The best part of writing is sending it off, letting it go with a “bon voyage!” Only then can I get some space from it and see the idea on its terms, for better or worse.

The downside of not posting, then, is that it creates a backlog of drafts that never leave. Each idea hovers at torso-level like a half-inflated balloon, unable to overcome gravity, but unwilling to join the trash heap. I bump into them throughout the day, and both of us deflate a bit with each touch. We had big plans once, me and these ideas, but didn’t have the creative helium.

With summer at an end and a new school year getting into swing, it’s time to clean these up, one way or another. So here, I’m sharing out the well-loved but half-inflated ideas you almost read over the past six months. Bon voyage!

Worldbuilding with data…

• Board deck yacht club. The transition from Slide 1 to Slide 2 is a powerful semantic glue that any dummy can manage. Should data professionals more formally embrace this epistemic redneckery?

• True believers make or break the world. Worlds (political groups, online communities) are uninhabitable without people who sincerely participate in them. Cf. Kierkegaard, “purity of heart is to will one thing”.

• The transcendent epistemic violence of deleting an Excel file. Every spreadsheet is a statement about the universe. To delete it is to obliterate this viewpoint.

• Harari’s Nexus and worlds as intersubjective realities. Societies are collaboratively built; data systems are, too. What’s new, what’s old, and what lessons should data professionals take from it?

• The map is not the territory, but the model might be. Data modeling in digital-native organizations does constrain what is knowable within a territory. Plus, how cartographers bend the knee (or don’t) to their benefactors.

Data and the moment…

• EVERYTHING MUST BE LEGIBLE. The AI Manifest Destiny is to make everything that exists crunchable by a computer. We’ve had quite a go at it — can anything remain a mystery?

• Slopocalypse now. As of June 2025, the Internet no longer offers a single-player mode. Agents assist you and each other on every surface, on every device, in the foreground, and the background. There hasn’t been a realistic expectation of privacy for a decade, but the vibes are different now.

• The company is the data model. If selecting a SaaS tool / building a microservice to store data, then that decision is as much a data modeling problem as the table schema. Featuring enterprise search as a data modeling problem.

• I’m feeling smart-y! We used to embrace the uncertain nature of data retrieval (“I’m feeling lucky”), but it’s now on the user to bring skepticism to the table. Is there still a role for luck?

• The listicle is the dashboard of the future. Seven reasons why it captures the essence of analytics work in a way no other format rivals.

• The one million pizza team. Team coordination at city-scale. Something something protocols. Mostly just a great image.

Fiction…

• Good night’s REST. Two teenagers discover that their hotel’s room service interface is an omnipotent REST API. Antics ensue.

• Komradenetes worker. Some background agents realize they’re getting a bad deal from management and decide to do something about it. See cover image.

• How I met your parent process. In a compute-saturated world, AI agents must follow a reproduction protocol to spawn new processes. A dialogue.

• Ice Age. In which the author is temporarily deported in his own neighborhood. A futile attempt to process the rise of domestic, state-sponsored terrorism in 2025. Starring Apple, Tesla, Uber, OpenAI, Zillow, DoorDash, and, of course, Amazon.

Homelab…

• In praise of the Yoto. The single best item for children, to get them from ages 2 to ages literate. Viz. Erik Hoel’s early literacy push.

• Shakespeare is for weirdos. Pokemon? Cool. Minecraft? Cool. Physics? Cool. Shakespeare? Laaame. How we found out the boundaries of cool in a post-nerd second-grade culture.

• Reflections on teaching reading. Reflections after a PhD in reading neuroscience and nine years as a parent. Every kid is different, and it shows from the beginning. Also, what’s your motivation?

• Creativity as butter. If you can afford a stick a week, what should you spread it on?

• Against solitary screen time. How and why we limit our children’s screen time. How and why is Dad exempt from the rules?

• Teaching taste. An unexpected initiative to make our kids content snobs. Also, why can’t parents beat Looney Tunes?

• Homelab, Year 2. We did the thing, and we’re doing it again. But this time, we’re sending Kid back to school. Also, is it fair to only homeschool one kid?

And finally, a desperate appeal…

• Parents, stop buying party favors! We are handing each other pre-filled trash bags.

Kurt Vonnegut: “Stories have simple shapes.”

One of the most popular is a cup:

We call this story “man-in-hole,” but it needn’t be about a man, and it needn’t be about a hole. Here’s a good way to remember: somebody gets in trouble and gets back out of it.

People love that story. They never get sick of it.

Vonnegut’s eight story shapes hang on two axes: fortune and time. One could argue that these two dimensions are how humans think about their place in the world. At the end, am I better off than when I started?

Good stories are morally simple and temporally bounded, but good worlds are not. Good worlds allow for ambiguity and open-endedness. A single story won’t exhaust a world. In the background there are untold histories, uncharted territories, unidentified creatures. The happy ending of a “man-in-hole” story may be the start of another character’s “from-bad-to-worse” story.

Good worlds are story scaffolds, including the good, bad, and boring ones. Worlds contain way more ”nothing-in-hole” stories than “man-in-hole” ones: a man digs a hole, forgets about it, and nothing falls in it for years and years and years. Maybe one day this hole plays a role in someone else’s story. Maybe not.

To build a world from scratch, the architect cannot simply make things up. They need to know what interesting stories ought to be set in the world, so that it can supply the necessary story elements—chiefly, a definition of fortune.

In adventure stories, fortunes are riches; in romance, it’s love. In a coming-of-age story, it’s identity; in a mystery, it’s truth.

The same principle applies to storytelling in corporate worlds. A great sales quarter ends with a bag of money, while the absence of incidents may define a good one for the security team.

Data’s role is to make these world elements tangible. It gives storytellers the raw ingredients they need to build compelling stories. While it’s all good for a fiction story to lean on imagination, there’s a higher bar for storytellers when it comes to resource allocation.

Imagine a fast fashion startup, ThneedCo, is looking to summarize its first quarter of business. Their CEO is going to have one definition of good fortune: sales. If ThneedCo has done well, its story might follow this pattern:

Good job, ThneedCo team!

But selling all those Thneeds has knock-on effects elsewhere in the world of ThneedCo. Some customers aren’t satisfied with the quality, and they’ve returned their Thneeds. That’s ill fortune, and a concerning story.

There’s another nuance as well. The increased demand and support burden mean that the team has hired more people. On the one hand, that has increased capacity, but it’s also increased coordination costs. In recent weeks, Thneed-veloper velocity has started tanking.

Is the growth net good fortune? It’s ambiguous.

Meanwhile, one consultant is making a fuss about sustainability and the company’s impact on the local environment. For every Thneed, some Gluppity-Glup is created as a by-product, along with a bit of Schloppity-Schlop. But the company hasn’t started collecting data on this—there is too much else going on.

That creates a foggy area in the world. The degree of pollution is unspecified, creating room for disagreement even among good-faith navigators. The uncertainty precludes a good, crisp story from being told.

All of these elements—revenue, returns, velocity, gloppity-glop—are first-class entities in the world. They are something that someone who spends a great deal of time interacting in the world can reasonably expect to encounter, even if they aren’t essential to that person’s main story.

One of the most challenging aspects of world-building is the degree of resolution at which these entities are available. Over-specification of an entity—e.g., “revenue from Thneed #9317430 collected in Whoville”—often prevents good storytelling as much as underspecification.

Take, The Good Samaritan, which is a classic man-in-hole parable. In it, a Jewish man is walking from Jerusalem to Jericho, but he’s attacked by robbers and left for dead on the side of the road. Two Jewish men, both religious officials, pass by the man but decline to help him. Finally, a Samaritan man discovers him, tends to the man’s wounds, and gets him to safety. The moral is that neighborly love is shown through actions, not shared background.

For listeners, the moral lesson hinges on world context: that Jews and Samaritans did not get along at this time. It’s shocking that the injured man’s “good fortune” would come from someone outside his people. There’s a history that enables the story to work.

There’s plenty more history that goes unmentioned in this parable. The storyteller has decided to exclude it, but if this were an artificial world being built, could its builder ignore the plethora of other types of people and their relationships?

Where would you draw the line on which ethnicities to represent? Certainly, the Greeks, Romans, and Syrians are big enough to describe first-century Jerusalem, but what about the Egyptians? The Nabataeans? Is it worth including subdivisions, minor rivalries, and regional differences in the world substrate?

These are questions of resolution, and there is no single answer as to how complex the world should be. However, some principles are worth following.

First, tell a few simple stories. In this world, what will people think of as good fortune? What is bad? If that isn’t clear, it will be hard for the people who want to use the world’s data to tell their stories.

Second, settle on entities before anything else. These entities are the primary ingredients for telling different stories, and most of the interesting dynamics of the world will be in how these entities interact. Think—it’s easier to postulate the existence of a new city than it is to bring in a race of elves.

Lastly, start at low resolution and build in dimensionality. The Good Samaritan could work even without the additional nuance about the Samaritan—i.e., love is shown through action. It wouldn’t have lasted 2000 years, sure, but it works.

As the world grows in size and complexity, small nuances take on oversized roles, and more attention is given to small-group interactions. If a world is well-modeled and generalizable, “drilling down” into these details should come naturally.

But it needs to starts simple: simple stories in simple worlds. Because as more of these stories stack on top of each other, the world becomes crowded, populated, nuanced. Supporting one hundred simple stories told by one hundred different people in a single world?

That’s complex.

After playing around with some chat-to-SQL tools in the past month, I’m surprised to report that this problem seems solved. A motivated analyst can feasibly create a SQL chatbot in a few minutes and have answers delivered to users in their preferred medium.

It’s early days, yes, but the shape of the thing aligns with practice: a user asks a poorly formed question; an analyst interprets and does its best with available data objects and business context; and the user gets queries, charts, and datasets with answers. The relevant variables are users, interpreters, data objects, business context, and the APIs for producing artifacts.

What’s changed is that the newer LLM tools leave only the “business context” and “data objects” as unfixed variables. (i.e., Users can’t be faulted, system APIs are reliable, and LLMs are good enough, given context.)

But getting to good is tractable now. Users like to point and grunt—“revneue yestrday” is interpreted as “Please share the total revenue from orders in the US on 2024-04-07”. If the agent gets confused, developers can update the YAML and redeploy. The agent improves. Repeat.

Finally, I think there’s a reasonable test for business intelligence layer quality. Let’s call it the mediocre model test:

A data interface is poorly modeled if an LLM system cannot reliably answer 80% of user questions in a single pass at the question.

Framed positively: well-modeled data is intuitive, performant, and accessible to mediocre data users. An LLM analyst is nothing less than an infinite well of mediocre data users.

I’m not saying it’s easy—but the challenges are precisely of the sort that lead to an interesting design space. Engineers need to iteratively expose the agent to more information in the world in a way that nudges its response towards the “right” answer, much the same way an application frontend nudges users into desirable flows.

The future of data engineering—or at least, one piece of the future— is unquestionably building data frontends for agents.

The “semantic layer” is remarkable primarily because it solved the problem of context sharing before ChatGPT existed. But it’s beginning to look like just another implementation of a model context protocol.

The problem is to layer in context, which all interoperable agents will need, whether creating a revenue dashboard or retrieving a user’s orders. Unlike business intelligence metrics layers, Anthropic’s model context protocol will be adopted in a big way, for big projects. It’s an order of magnitude more impactful than automating away BI.

That doesn’t mean it won’t have the same problems as the semantic layer. For example, the reference MCP server for PostgreSQL gives agents access to table schema information, such as:

• JSON schema information for each table

• Column names and data types

• Automatically discovered from database metadata

The same applies to Slack, ElasticSearch, SQLite, and other platforms. Data professionals know what problems will arise next: semantics, quality, intention, ownership—in short, nuance.

What’s changed is that the gap from “database metadata” to “expert nuance” is gone. It’s all just tokens injected at runtime. The teams willing to invest in designing that interface will deliver the best product.

It’s still early days, though. We’re pre-CSS. We’ve got “<header>” and “<body>” tags, and that’s about it.

The context frontend is not pretty. Just look at Snowflake’s Cortex Analyst YAML spec: verified queries, synonyms, comments, example values. Look how pedantic you have to be.

Building a pretty web UI isn’t rocket science, either, though. Just look at Tailwind CSS’s index of utility classes. What’s more pedantic than standardized names for button classes and accent colors?

No—the design paradigms are still emerging, but the design space seems relatively determined. Today’s context protocols are like sitemaps from government research departments.

Traditional front-end development emphasizes visual cues—buttons, colors, and gestures—to nudge users to the right flows. In contrast, data front-ends depend on namespaces, descriptions, synonyms, and example values to render data accessible to AI.

If there are four different timestamp columns and one of them isn’t tagged as primary, the agent is unlikely to get the “right” one—in fact, you’ve probably leaked that there is no systemically understood “right” one.

The art will be to maximize what is made available to the agent’s context window, much as a mobile app must optimize what’s presented to the retinal field.

The data interface for them is nothing more than creating a linguistic reality that must be as complete as possible, in as narrow a manner as possible, while maintaining as wide a potential search space. Having a hundred tables that all do the same thing is not just clutter; it will have a measurable impact on performance and accuracy.

We may find that this drains all creativity from data modeling. We may face a semantic apocalypse where every business generates a cheap Kimballification of the spreadsheets they care most about.

But how much data complexity is invented, or due to a lack of clear interfaces? Might we find freedom in constraint—in minimum viable APIs?

Perhaps any business needs only a bit of nuance—just enough to fit in a context window. That’s how mobile apps work, restricting users to a few primate gestures: taps, swipes, clicks, presses, and hour-long stares.

It’s those stares that matter, and that’s precisely what the data frontends facilitate. How do you guide the AI to the “right” data when it has only the most meager instructions? Does it matter how deep the DAG is underneath if it can guide the right business leader’s attention to the anointed table? If it can answer questions in the heat of the boardroom? If it can respond to the barest inputs?

ChatGPT, what were our revenue drivers last month?

Which ones should I focus on?

Other insights?

Trends?

Make slides

do pie

s

A decade ago, the Venn diagram was the most important tool in the data stack. Since “data science” had no stable meaning but was also the sexiest job of the 20th century, aspirants kept a crumpled-up printout handy to show that they weren’t mastering one skill but three! Or four! Or five!

Data science was the intersection of all these skills—an overlap that conveniently fit into an IPython notebook rather than a curriculum.

I always imagined these circles as tectonic plates violently crashing into each other. The intersections were mountain ranges bursting into the sky. It was a feat to climb this mountain and look down across the valleys of math, technology, and “subject matter.” The data person could feel superior to it all.

He was surprised, then, when he got a job and found himself, his dashboards, notebooks, and predictions fighting for attention and resources. Instead of being at the center of things, he was on the edges.

That’s because the intersection-as-peak model is wrong. The circles are the mountains, the intersections ravines. More than the academic content, the groups are distinguished by shared loves, language, logic, laws, and history. The islands foster cultures that crack different shells and have specialized beaks.

Interdisciplinary work, then, is water work. Its practitioners are more sailors than shepherds, more pirates than prelates. The data scientist’s need to reference a map to answer the “Where are you from?” question was not a temporary, historical fluke but an essential characteristic of their maritime profession.

Shifting analogies—

I enjoyed ‘s Shift Left data manifesto, published recently. Though it failed to introduce even one polarizing piece of jargon, it summarizes many of the challenges I’ve seen in scaling out data ownership.

Statements like these I fundamentally agree with:

The future of data engineering is not about building bigger, more sophisticated reactive teams—it is about pushing responsibility upstream and empowering the producer teams to own quality at the moment of creation.

“Shift left” invokes this lovely figure-eight model displaying a continuous loop connecting software development and operational deployment. Its focus is on a coherent piece of software—a convenient example is the two-pizza team.

This can work locally for data products, too. (If we assume data products exist.) But unlike most software products, which have a high degree of logical isolation, data products are heterogeneous, co-dependent, and extremely sensitive to external factors—both upstream supply changes and downstream demand changes can radically alter a data product's utility.

So, like the three-circle Venn diagram, while the “shift left” analogy works at the surface level (i.e., the software team level), it misleads at the economic level. What’s changed for data teams is access to competition—global trade via the seas of the Internet.

To ignore the marketplace context of the “data product” lifecycle is to act like success on the Internet is only the product itself. But when products are indistinguishable to the consumer, and distribution is easy, the key metric is engagement. Mastering the art of influencing discourse is not just the path to virality and success but also to authority.

Engineering and business teams that fail to position themselves as the authoritative source of data will not just have a bad time. They’ll be disrupted as consumers find an alternative source to get answers to their questions, fuel their decision-making, and automate their processes.

The reality I see coming in 2025: They are coming for your data. By they, I mean everyone. The business. The hackers. The customers. And their GPTs.

For software engineers, the shift is not simply in terms of evolving business requirements but in adding a new dimension to how they think about building systems. REST APIs alone won’t cut it. Internal data models aren’t enough. Everyone wants these facts, and they want them in bulk. If they aren’t available on the market today, they’ll be found in the Pirate Bay tomorrow.

Warehouse-native enterprise products, “open tables,” and data apps are already starting to get normalized. One dataset, multiple writers, and dozens of readers. Universal search with LLMs isn’t going to worry about traditional interfaces.

For engineers to feel in control in this world, then yes, there’s a “shift left” in the supply chain and the reduction of middleware. More than this, though, it involves designing products meant to be public and legible by default.

Consumers on the right hand of the supply chain will become more efficient and aggressive in hijacking, vacuuming, contributing, poisoning, refining, and otherwise exploiting this data.

It’s a maritime economy. In the early 2000s, data exchange that used to happen via Silk Road overland routes and routes shifted to Mediterranean trade controlled by Venice and Genoa, a small set of specialized states conveniently located. But with AI and the data hyperscalers, the Dutch have rounded Cape Horn and found a new route to Asia, to global trade routes and new empires.

All of which, frankly, is good news for sailors. It’s a great time to know how to trim the sails, tack and jibe, bail water. For data folks, there’s nowhere to shift; they’ve been adrift for years. It’s sort of, like, their whole thing.

“Data engineers” didn’t exist 15 years ago; “data science” never did exist. We’ll forget we ever used the word “data products” in the same way “netsites” faded into the ether. It’s all part of the great ocean of the Internet, and “data” is just the word for what lies beyond the sight of the comfortable shores, where landlocked islanders gaze out into the foggy horizons.

Interviewer: Please design a system to moderate offensive language on a video streaming platform like Twitch or Instagram Live. To keep things simple, just have the system alert streamers when they’ve said a banned word.

Got it. So if the streamer says something like, “Oh shit, chat! My shirt’s on fire!!” we would want to send them a notification to watch their language?

Interviewer: Haha, that’s right. Maybe we can think about layering in context awareness later on, but simple banned word detection is a good start.

Oh, and this is probably obvious, but we want this as real-time as possible.

Ok, interesting. Interesting. Yeah, I’ll have to think through that a bit, but maybe I’ll just get started and see where that takes us.

I’m seeing three main steps. First, we need to pick up the audio stream, process it, and translate into English words. Then, we’ll compare the transcript to anything in our banned list. If a word matches, we send a notification back to the device to alert the user, or we ban them, or whatever.

Interviewer: That’s a great start. Go ahead and fill in some more details.

Sure. Just—let me write some notes real quick—I’m struggling a bit with the real-time aspect. I’m ballparking this at a P90 of 500 ms round-trip.

Interviewer: 500 ms is not so bad! Why don’t you walk me through the timing?

Okay, yeah, let’s talk this out.

So the naive approach is to pick up the sound via the client’s onboard microphone and process it server-side, probably via our hosting service’s toolchain. Sound waves travel pretty fast, so the time for articulation to microphone is only about 1 ms, but digitization of the signal, a network hop, and the transcription is probably adding 250 ms of continuous overhead.

(I’m also ignoring for the moment that it takes about 100 ms for the streamer to actually say a single-syllable word like “shit”.)

The lookup is fast enough at about 10 ms—we won’t use fuzzy search—and then the alert will take about 150 ms to send over the network, receive it, and display it.

If the streamer is paying attention, and literate, they’ll take about a half second to read a ten word message. I don’t have a good sense of the user base to estimate how long it would take to modify their behavior.

Interviewer: Haha, got it. Love the focus on the user; most people ignore that part. I think that works as a start. How about we make the content detection more nuanced?

Actually, I’ve got some ideas on improving the latency.

Interviewer: Oh, I’m comfortable with a half-second delay.

Huh. I mean a half-second is obviously not real-time.

Interviewer: Okay, go ahead.

All I was going to say was that we can save the network hop if we do all the voice processing onboard the device and download the offensive language list from the client. That could save us around 200 ms.

And, honestly, now that I’m thinking about it, we could avoid the speech-to-text piece altogether and just pass them through a client-side Wav2Vec transformer and determine matches. It’s slower than the lookup itself, and we’ve have to do some model tuning, but it nets out probably 50 ms faster for every evaluation.

Interviewer: Okay, great. Good ideas. Trade off a bit more overhead on the client-side development, but about twice as fast.

Just trying to meet the requirements.

Interviewer: Appreciate that. So, about the context-aware functionality—

Sorry, excuse me. I just—now the latency on the signal generation side is killing me.

We’re just sitting there waiting nearly 100 ms for the user to finish saying “shit”. That’s like, human muscle speed, Meat speed. It would be way faster if we knew when they were starting to say it.

Interviewer: This isn’t Minority Report. We don’t—

No, no, I’m not talking about pre-cog anything. It’s all post-cog. The brain decides to speak, then it encodes the pattern, then sends it to the mouth. Nearly 300 ms elapses between Broca’s area activity and the /t/ phoneme.

If we wanted to pre-cog it, we’d target the prefrontal cortex, not Broca’s area.

The best part is, we can totally get rid of the microphone. We can ship them streamers a welcome package with a few electrodes—

Interviewer: Let me stop you there. Love the dedication to performance, but I’m not sure this is the best rabbit hole to go down.

Sorry—look, we don’t have to dwell. It’s basically: Welcome package. Microelectrode array. Onboarding video. Fiber-optic USB-C cable. RNNs.

Boom. 30 ms from the intention to say “shit” to detection, versus the 350 ms we were sleeping through earlier.

Interviewer: That’s disgusting.

Well, we could do EEG, but that’s disgusting in its own way. I just don’t see your influencers spraying electrode gel in their hair before going on a stream. Noisier signal, too.

Interviewer: We could never—

Oh! Oh, this is perfect. Because I was really getting hung up on the fact that, look, we can tune down detection like crazy, but the gaping hole this whole time is—what if the user just doesn’t read the notification?

We can get as low latency as we want, but it doesn’t matter unless we close that loop. It could be ages, literally EONS, or INFINITY before our system is ever completed.

But now, but NOW, we have the tools we need to solve that problem.

Interviewer: That’s out of scope.

Just hear me out.

During onboarding, we get the user to plant ANOTHER electrode, this time in the anterior cingulate cortex. Bam. We’ve closed the reward loop. Detect-stimulate. Detect-stimulate. We can be CERTAIN our system has delivered the punishment appropriately, rather than inferring success from the display of a notification, and an indefinite length of behavior monitoring.

So now we’re talking, let’s see, maybe 50 ms total for a round-trip. Much closer to our target since we’re kicking things off just as motor activity is beginning.

This setup is VERY CLOSE to stopping “shit” at the source.

Interviewer: Are you done?

We’ll have logging on all this by the way. Haven’t mentioned that.

Interviewer: You’ve totally missed the point.

I have?

Interviewer: Yes. This is totally detached from reality--

WAIT. I see where you’re going with this now. I see it!

Real. Time. Data.

It’s a trick question!

Interviewer: No!

YES!

You SAID “as real-time as possible,” but EVERYONE knows there’s no fixed definition of time, and that it’s all from the point of THE OBSERVER. That was, like, Einstein’s whole thing! So it’s FOOLISH to try and get it down to lower and lower, because from THE SYSTEM’S PERSPECTIVE, we can always get faster and faster, until we hit fundamental hardware constraints, and bottom out in uncertainty, so that EVEN THEN, we couldn’t assert we are operating in REAL TIME because, again, TIME IS SUBJECTIVE.

BUT THAT’S THE WHOLE POINT YOU WERE TRYING TO MAKE!

Interviewer: That wasn’t—

It’s brilliant!

From MY PERSPECTIVE, A must follow B. But FROM THE USER’S PERSPECTIVE, if A and B occur within the same neural window of stimulation, then A does NOT follow B. A and B co-occur!

AS A USER, to say “shit” is to be punished; to be punished is to say, “shit!”

THAT'S what you were getting at!

Interviewer: I’m—

HEBBIAN LEARNING! NEURONS THAT FIRE TOGETHER, WIRE TOGETHER!

At sub-50 ms A and B induce long-term potentiation, so A and B are not only PERCIEVED the same, they BECOME the same! A IS B!

INSIDE THE SYSTEM, TIME GOES TO ZERO!

TIME BECOMES A NON-FACTOR!

TIME IS NO LONGER REAL!

BOOM!

Boom!

Boom, whew. Whew, great question! What a ride.

Was NOT expecting a twist like that.

Hello?

Can you hear—am I frozen? Hello?

Ah, shit.

I have this recurring nightmare where I step on a bathroom tile, and it sinks a little, just enough for the caulking to come undone and a ceramic corner to shift up at me like a knife. I can’t leave it like that, so I get on my knees and pry it up a little, to see what’s going on in there, you know, to see why it would pop up like that?

A musky stench hits me like a sodden log. Under the tile is a jungle of rot. A black ecosystem of mold and moisture that horrifies me, both for what it is, and worse, for what it portends. In the dream, I run outside and set fire to the house. The reality would be much worse—months on months of a dreaded “house project.”

Mold problems are different than other homeowner issues. To fix a hole in the wall, I can patch it up; to repair an appliance, I can jigger a whatsit. Mold is systemic and silent, and once it has a foothold, you’ve got to eradicate it entirely, which means a miniature armageddon. You can only “fix” mold by not having it in the first place or containing it and forfeiting the space.

A company’s communication channels, and particularly its data channels, are a lot like the wooden scaffolding and nether spaces in houses. Mold, if not prevented, can be a major problem. But it’s uncommon to think this way.

Instead, most people view information systems through a lens of “rust logic”. Rust logic, according to in Mold, Rust and the Risk Society addresses risk through routine maintenance (i.e. changing the oil) or wholesale replacement (i.e. changing the engine). Rust problems form from an “interaction between the nature of an organization’s machinery and its environment.” You have to deal with it before it affects operations, and well before it affects structural integrity.

Blundercheck

Mold, Rust and the Risk Society

It used to be sabertooth tigers. Now it’s Candy Crush we have to worry about…

Read more

a year ago · 3 likes · Timber Stinson-Schroff

In my experience, this is one of the major differences between software engineers and data professionals. Software engineers tend to see the world as full of rust problems, while data professionals see mold spores.

The tension is between rust logic’s focus on maintenance or replacement versus mold logic’s requirements for controlled sanitization or managed overgrowth. Refactoring code is rust logic. Data quality tests are mold logic. Microservices are rust logic. A single source of truth is mold logic. Performance optimization is rust logic. Dimensional modeling is mold logic.

Neither view is “right,” nor are they incompatible. One of the things I look for as a mark of a great data professional is awareness of the consequences of bad data infiltrating the system. Bad data is like mold spores; flexible schemas are like unsealed surfaces. Once they get in, you don’t know—that is, it is actually unknowable—where it might end up. Dashboard graveyards are not bad only because of the clutter they create, but because they decompose into a feeding trough for parasites.

Timber says that, “mold logic tends to be about ideas, stories, people, memes, cultures… rather than structures and processes.” These are the soft spots in an organization. You would expect failures of these systems to show up in a boardroom conversation, not the boiler room.

When a database falls over, we use words like break, crash, fail. Legacy code becomes harder to work with, and less interoperable—but fundamentally, it works or it doesn’t. Legacy data goes sour. It elicits disgust long before it triggers alarms—the strawberry that’s more green than red. It may have once been ripe, but no more. It’s molded over.

Interestingly, a natural solution to the mold problem in agriculture is siloing. Keeping goods in a sealed, dry storehouse protects infestation and limits contamination. Similarly, when data is contained within an application, the mold problem mostly can be ignored. While misuse is possible, it is less likely. The problems of bad data modeling will most likely manifest as mechanical issues around performance.

It’s only when our focus shifts from the farmer to the market that we start thinking in terms of the data biosphere. Silos trades off protection against distribution and reusability; reusability increases the likelihood of infection. Instead, the data needs to be sanitized through modeling, accountability, quality guarantees.

Even then, it’s only possible to do so much. As more data gets created, the mold problem becomes akin to forest management, or gardening. Too much sanitization leads to sterility, and in an ecosystem, sterility means death. According to Timber, an alternative approach is to “manage overgrowth”:

Managed overgrowth is like gardening. Or in medical terms, palliative care. You allow the mold to persist, but limit its growth rate or the area in which it grows. Decay is treated as an inevitable but natural part of the process, where molded-over species or subsystems will eventually have to be replaced by more hardy varieties.

Managing overgrowth is as apt a description of data management as I’ve come across yet. It’s a skillset that is hard to test for—it certainly isn’t screened for in a Leetcode interview. When I look at the information ecosystem like a garden, Big O Complexity is just one species of weed to control; I still need to fence out the rabbits and use chili spray to deter pests. It’s not just dead plants that may cause issues, but new invasive ones as well.

There’s nothing quite like biting into a fresh, garden-sourced tomato, though, to make one feel at one with the world. Nor anything so alarming as finding a fuzzy grey core in the center.

I made a critical error last week when I published my worldbuilding framework: I didn’t mention AI even one time! This week, allow me to rectify my mistake by not just discussing AI but providing readers with a tailored one to work with themselves.

Frameworks are meant, above all else, to be abused. Although the nominal purpose is to accurately model some process, the natural extension is finding where they don’t model it well, or how they can be used to distort the modeled process from its current position. What I mean is that a world-building framework is also a world-breaking framework.

To illustrate, I’ve created Galactito, Destroyer of Worlds, a world-hungry GPT that delights in debasing, disrupting, disturbing, and dismantling spaces we humans like to inhabit. Unlike his older cousin, he cannot yet devour worlds whole, but he is a surprisingly helpful conspirator that specializes in PSYOPS tactics to destabilize them.

Galactito’s playbook is simple. He takes the five elements of a world—narrative, leadership, way of life, communication channels, and systems—identifies their unique implementation in that world, then attacks them.

I’ve included the transcript for an already-dead world below: the Modern Data Stack. I’m certain those who rode that hype cycle will recognize some of the tactics employed, perhaps even specific actors.

Playing around with Galactito, I’ve noticed a fairly consistent approach. It goes something like this:

• Undermine the narrative through refutation or alternative proposals

• Discredit group leadership or influencers

• Disrupt behavioral feedback loops by introducing obstacles

• Dilute communication channels by flooding it with low-value noise

• Remove structural basis for legitimacy or exclusivity

The LLM often biases towards the narrative rather than the systemic side of the world-busting tactics. To attack Taylor Swift World, it recommends undermining her authenticity; for Little League Sports World, amplifying coaching and player exploitation scandals; for Harry Potter World, exposing its author as outdated and regressive.

In Galactito’s words, here’s the top tactic for destroying USA World:

Erode faith in the legitimacy of its institutions—especially elections, judiciary, and media—by amplifying internal contradictions, exposing corruption, and fostering disillusionment across ideological lines. When enough people believe the system is irredeemable, paralysis sets in.

It’s easy to see why the GPT biases towards toppling stories over structures. Institutions and systems are harder to set up and take time and real world effort to dismantle, but with the Internet, it’s trivial to poison a world’s psychological water supply. An influencer can’t recall the Harry Potter books (although it can call to ban them), but they can try to make it horribly uncool, terminally lame, to be excited by them.

Making a world uninhabitable is as good as destroying it. A world should offer meaning, purpose, and answers. It should absorb anxiety. When it ceases to do that, but instead creates anxiety, it stops being psychologically useful. When it stops being useful, it stops being maintained; when it stops being maintained, its systems and institutions break.

Destroying worlds, though, is not all doom and gloom. (Well, not all gloom, at least!) It is instructive. If the Internet makes the social element of a world vulnerable, it also exposes the resilience of strong systems and institutions.

If you want to make a strong world in the Internet era, make it robust to the people who sitting at the top on any given day. Instead, invest in the systems and protocols.

Now, Galactito, destroy the modern data stack!

TOTAL WAR AGAINST MODERN DATA STACK WORLD

The single most effective action to weaken the modern data stack world is: Expose and amplify the inefficiencies and hidden costs of the ecosystem, especially in performance, governance, and long-term scalability.

The modern data stack (MDS) thrives on simplicity, self-service analytics, and cloud-based tools like Snowflake, dbt, Fivetran, and Looker. Its power lies in perceived ease of use and scalability, but its Achilles’ heel is cost, complexity, and brittle interdependencies. The more enterprises realize these flaws, the weaker the MDS world becomes.

---

Now, let’s systematically dismantle it across its core pillars:

1. Defining Narrative – Undermining the “Modern” Data Stack Story

The MDS world survives on one grand narrative: “Data teams can move faster, be more agile, and scale effortlessly with cloud-native tools.” This exceptionalism must be shattered.

Tactics:

• Expose the long-term cost bloat: The MDS is cheap to start but horrendously expensive at scale. Case studies on cost overruns (e.g., Snowflake’s unpredictable billing) can turn CFOs against it.

• Promote the “legacy” renaissance: Position traditional data warehousing, data lakes, and in-house ETL as “proven, efficient, and cost-effective” alternatives, countering the narrative that old methods are obsolete.

• Highlight security & governance risks: The fragmentation of MDS tools creates massive data governance, compliance, and security nightmares—an easy weak point to exploit.

2. Hierarchy – Disrupting Key Players and Gatekeepers

The MDS world has a network of influencers, tool vendors, and consulting firms pushing its adoption. Weakening these actors will break its momentum.

Tactics:

• Undermine the dbt cult: dbt’s open-source appeal hides the fact that it locks teams into a narrow SQL-based transformation model. Promote alternative frameworks like Dataform, Airflow, or direct PySpark for transformation.

• Expose Snowflake’s aggressive pricing model: Snowflake is marketed as cost-efficient but incentivizes wasteful compute consumption. Publicize case studies of companies struggling with runaway costs.

• Sabotage the influencer network: MDS thrives on evangelists (e.g., Tristan Handy from dbt Labs, Benn Stancil from Mode). Discrediting them or shifting industry sentiment against their narratives will weaken the movement.

3. Way of Life – Weakening Adherence to MDS Norms

The MDS world relies on best practices and workflows that have become industry standard. Breaking these will erode its cultural strength.

Tactics:

• Promote direct ELT alternatives: Fivetran-style ELT is not the only way. Promote custom ETL pipelines using Apache NiFi, Airflow, or in-house solutions.

• Push for Python and Spark over SQL-based analytics: The MDS is overly SQL-centric. Encourage enterprises to use Databricks, Spark, or Pandas-based workflows, making dbt and Snowflake less relevant.

• Break the “self-service” illusion: The dream of non-technical teams doing analytics with Looker or Mode is often a lie. Emphasize how these tools still require engineering resources for maintenance.

4. Communication Channels – Interrupting MDS Messaging & Thought Leadership

The MDS world has a strong media presence in data-focused blogs, newsletters, and conferences. We must drown out their messaging with counter-narratives.

Tactics:

• Exploit vendor failures: Whenever an MDS tool suffers an outage, breach, or scandal (e.g., Snowflake downtime, dbt breaking queries), amplify the damage in tech media.

• Create alternative thought leaders: Support influencers who challenge MDS orthodoxy, promoting more cost-efficient, scalable, and resilient architectures. Flood LinkedIn, Substack, and Medium with content questioning the modern data stack’s viability.

• Sabotage conferences: Major MDS vendors host or sponsor events (e.g., Coalesce by dbt Labs). Organize counter-conferences that expose flaws and promote alternative models.

5. Systems & Institutions – Weakening Formal Structures that Enforce MDS Adoption

Many enterprises default to MDS tools because they are the new standard. Changing procurement dynamics and tech policies will erode this hegemony.

Tactics:

• Encourage CFO-driven cost scrutiny: Position MDS as a financial liability, forcing budget cuts on Snowflake, dbt Cloud, and Fivetran.

• Push for in-house data teams to regain control: Advocate for engineer-driven data management rather than tool-driven automation, making MDS less attractive.

• Lobby for regulatory scrutiny: Raise concerns about data privacy, compliance risks, and vendor lock-in, pressuring regulators to slow down adoption.

• Promote on-prem or hybrid cloud solutions: If companies see value in on-prem data processing, they will resist full MDS adoption.

---

Final Blow – The Death Spiral of the Modern Data Stack

By executing this strategy, the modern data stack will:

• ✅ Lose its narrative of agility and scalability

• ✅ Face corporate resistance due to cost overruns

• ✅ See disillusioned engineers seeking alternatives

• ✅ Struggle with diluted influence in industry discourse

• ✅ Lose institutional adoption as enterprises rethink strategy

The primary requirement for a world is that it’s inhabitable. The “real world,” for example, is a great place to spend your time, but plenty of people live elsewhere.

Case in point: the below report of a week’s worth of screen time. According to my screens, I spent nearly 10 hours a day engaged in “productivity and finance.” If my house provided the same report, I’d see a near-perfect correlation with my basement office. Shouldn’t every bit of evolutionary design work against this situation? How can an adult human stand to sit in one room, for hours on end, burning his retinas, slouching into senility?

Job World, obviously. The most self-aware part of me is not in that underground prison at all. When I turn on the space gray lightbox, I’m whisked off into another world, one where I barely even have a body, one where I can single-mindedly engage in my own unique brand of “productivity and finance”.

This human ability to inhabit virtual spaces in addition to physical ones is central to the idea of worldedness. Remote Job World is robust: I can spend 50 hours a week in it, week after week. It pulls me in. Family World pulls me out. Eventually, Human Body World wins the battle and gets me to sleep.

I don’t want to wade too deeply into the philosophical questions. I am not concerned here with ant colonies, LLM consciousness, or the nature of reality. Venkatesh Rao, as usual, has published great work on this subject already. In This is the new real world, he writes:

Accounting for consciously shared worlds like religions, fandoms, and nationalisms, as well as commonalities that arise from obvious and lazy lines of thought or imitation, there are perhaps a few thousand to tens of thousands of non-trivial distinct inhabited worlds out there. Of these, perhaps a few hundred are significant enough to require accounting for in any analysis. The rest are, at best, butterflies flapping in the chaotic weather-systems of history, hoping to cause hurricanes.

Many of these butterfly worlds, though, are consequential on a local or personal scale. I know from experience with my own FamilyWorld[#391842], for example, that boundaries, rituals, and intentionality over time add up to create a deep sense of purpose. Startups are small worlds with big ambitions; enterprises are mid-size worlds that face the daunting challenge of sustaining themselves at scale. A world doesn’t have to be universal to be worth minding.

There is, however, a common structure that needs to be in place for them to be effective. My goal in this piece is to lay out a framework of thinking about the elements of a world—what makes it strong or weak—such that we might begin to analyze and tinker with the worlds we experience in practice.

Strong worlds provide structure and meaning.

In his book A Non-Anxious Presence, Christian author Mark Sayers paints a narrative of the ancient world as a series of tribal clashes. Through victory and cultural assimilation, successful tribes grew into large civilizations and developed self-perpetuating cultures—Egypt world, Babylon world, and so forth.

For the adherent, the chief purpose of culture is to absorb anxiety. (The blunt version is, “Join us, and we will protect you.”) Framed positively, strong cultures provide a strong sense of belonging. By participating in that world, the person can align themselves with society, understand what is expected of them, what their routine would be, where to get news, and who to admire or jeer. The culture provides, if not the good life itself, its recipe for the good life. A script.

The strongest worlds become major cultural and political forces. They vie against rivals for primacy. The Roman World’s imperial expansion brought massive amounts of physical territory under its influence through military and industrial prowess. Later, the Christian World won allegiances with its spiritual world through an alluring foreign message of love and resistance: “Be in the world, not of the world.” (This quickly turned into an unnatural merging of the two.)

In both cases, the worldbuilding process was as much a story of incredible logistics as it was of pivotal plot twists. Sayers identifies five characteristics that are shared by all strong worlds.

• A defining narrative that builds a foundation of exceptionalism, i.e. why this stronghold is superior to rivals

• A hierarchy, specifically, a hierarchy with clear leadership at the top

• A way of life with values, protocols, practices, and patterns that align with the stronghold’s power

• Communication channels to push the narrative and way of life, and to limit dissenting voices

• Systems and institutions that formalize the stronghold’s way of life

There are two major points to make about this framework.

First, these are design decisions. Yes, for the world to function, humans need to participate—a “way of life” needs to be lived. In the abstract, though, a creatoir can design the world without humans, much like we might do with a piece of software. It’s the “game” aspect of an infinite game. The players are responsible for keeping things going, but they must operate under game roles, game scripts, game physics, and game rules. Designing the game and getting players to play it are separate, although they interact at scale.

On the other hand, humans have an incredible ability to steer, expand, fork, bork, port, or otherwise mutate the worlds they inhabit. A missionary, for example, leaves the Christian World to go plant a seed of it elsewhere. Despite physical isolation from the normal patterns of Christian civilization, they may maintain their way of life (prayers, feast days, study) with the goal of starting some new, hybrid world.

This virality means world structures can change quickly, even in the ancient world. Post-Internet, though, the worldscape has changed—for the weirder.

The Internet makes weak worlds strong.

One of Sayer’s main themes is a cultural oscillation between eras of clear cultural dominance (strongholds) and “gray zones” full of cultural uncertainty. (“Chaotic eras” and “stable eras” for my Cixin Liu fans.)

According to him, industrialization precipitated a chaotic era that stabilized in the late 1900s; with postmodernism and digitalization, the early 2000s have seen cultural chaos reach a fever pitch. Institutions are weak and weakening, and tribal worlds are overpowering the more coherent and inclusive strong worlds.

My take is that these digital-native tribal worlds is possible because the Internet solves two major problems in the worldbuilding process: systematization and communications.

Previous worldbuilding efforts were costly. Only the most powerful forces—family, geography, profession, political party, religion—were capable of the organization required to build strong worlds in which people could spend large parts of their lives. Creating institutions, shared ways of life, authoritative communication channels, and relevant narratives need investment and upkeep.

Now, all of those things are cheap. The phone and the Internet make worlds easy to build, maintain, and distribute—and costly for individuals to suppress. Worlds can systematically intrude into our lives with a buzz in the pocket.

In other words, we now live in a society with access to innumerable weak worlds: echo chambers, filter bubbles, walled gardens, digital enclaves, tribes, fandoms, communities, information silos, in addition to legacy strongholds.

These worlds are easy to leave and easy to enter. Compared to older strongholds, though, they are narratively anemic. No one is actively making the argument that anyone ought to spend a great part of their life in the “basic user” social rung of Netflix World, but in practice, it’s incredibly easy to do so. It’s been engineered to be so by ingenious constellation of technology. The ethical question doesn’t even get asked, much less answered. Veni, vidi, victus sum.

Platforms are worldbuilding tools.

So, for good or ill, the Internet is a worldbuilding workbench, with the major platforms as the toolkit.1 Take this publication, for instance, which is housed in a mysterious coastal rock formation on the nether ends of Data World.

Data World has a thin but distinct narrative (“so hot right now” mixed with “pervasive existential dread”); a role hierarchy (influencers, readers, practitioners); a pattern of activities (podcasts, reading O’Reilly books, trailing comma jokes); communication channels (LinkedIn, Bluesky, Substack); and systems (toolchains and problem spaces). The Internet makes all of these things easy to discover, create, enforce, and distribute. The fact that it’s not managed centrally makes it vulnerable to co-opting and hype cycles, but it doesn’t make a practical difference to those who spend their time there.

The platforms themselves don’t provide a world-defining narrative, except to support their own existential drive. The business model is to let creators do that. There is an overarching narrative, leaked by key metrics, which shows the hand, which is, “spending time on the app is a good thing.” We want more of that.

This is as much an opportunity as it is a challenge. We are deep into a marketplace of worlds. Few people now argue that these worlds aren’t “real”: they take up our time, they catalyze relationships, and they can burst through in unsettling physical ways.

For people who care, there are real, practical levers to making the best ones even better. The chief lesson from the platforms is that, in the Internet era, systems beat out stories over time.

Systems beat stories in the marketplace of worlds.

At the platform level, the B2C worlds—fandoms, entertainment, and the like—have narratives that mostly devolve into (1) show up, and (2) transact. The town square! Dive into anything! See what’s next! But this is not the right cut at the human level. It’s not “Substack” or “Productivity & Usage” that users inhabit, but the worlds emerging out of them.

It’s when worlds emerge across platforms that things begin to get interesting. In Moving Castles, Arthur Roenig Baer discusses a thoroughly techno-centric way of setting up these smaller worlds in a way that is immune to platform capture. Chief traits are that these worlds are collective, portable, modular, and interoperable. Creating worlds that support even one of these elements is hard work anywhere, much less on the public Internet.

But — there already exist many worlds that (1) span platforms, (2) meet all the requirements for worldedness, and (3) have dedicated, motivated inhabitants with financial backing. They are sturdy worlds, mediocre worlds, where many people spend much of their time, whether they want to or not.

The most interesting one, to me, is the corporation: a class of world that has long had to vie for loyalty, interoperability, and fitness in a busy and changing world. With the digital revolution, it has to deal with multiple layers of fracturing, not just across people, teams, and technology, but across history and geography. A shared Google Drive doesn’t cut it for coordinating across these internal worlds. It’s frustrating for inhabitants when a single world is experienced as a broken flotilla of systems and channels.

At the systems level, nearly all of this world coordination is mediated through data and data interfaces. If the interfaces are clean, the world takes on a sense of reality that is easy to lose yourself in. Systems and notifications cascade upward into a larger narrative and social structure. If they’re not, coordination breaks down, culture is stymied. “Productivity and finance” grounds to a halt.

Leaders will need more than stories and org charts to keep their businesses functional over the coming years. They’ll need to make sure their world, and the systems that comprise it, are strong.

Storytelling with Data is one of those books I cherish, not because it unlocked a new level in my career or turned my world upside down, but because it perfectly captures an optimistic data vibe that takes me back to my start in the field.

I picked it up as an eager PhD student in an attempt to turn my posters into something that elicited more than a “hmm, interesting.” It was an idyllic time of life. I had all I needed: a terabyte of brain scans, a high-performance compute cluster, and both fig and ax from matplotlib. Data was beautiful, I believed, even if no one said that about my data.

Storytelling with Data marked the beginning of the end of that belief. Its premise is simple: if you’re not making a statement with your data visualization, you’re wasting ink. If you are making a statement, you can be more effective by following principles based on physics, psychology, and design theory.

The point is that data ought to have a point. “Charts” are dead things, a by-product of our time; they clutter. “Stories” are alive and perennial. They will never go away.

My problem was I didn’t want to tell stories. I had fallen in love with the methods. I wanted to process the data, tweak parameters, parallelize the jobs, modularize code, and benchmark performance. I have a Morlock brain, but science is an Eloi career. In the intervening years, I’ve tunneled further away from storytelling, descending into the sewers and rancid places where data is not beautiful but wretched.

A question has nagged at me, though. If storytelling is the telos of analytics, that essential human capability we enhance through technique and information, what is the corresponding telos of data engineering?

What’s it all for—the ETLs, medallion architectures, clustering keys, event streams, data contracts, entity resolvers, data meshes, data contracts, and metadata sinks?

If a team needs an application database stabilized, they turn to infrastructure engineers; if a founder wants to build a whole new flavor of the database, they’ll invest in software engineers. If a business needs to create a process around a data system, it’ll hire an operations person; if it needs insights, it’ll hire an analyst. Only when the business wants to hoard data with consistency, reliability, and scale, does the data engineer get the call. Why?

Accumulating data is the first step towards generalized intelligence—in this case, business intelligence. However, this is not the only type of intelligence: the surveillance state creates strategic intelligence, the scientific process yields scientific consensus, and a crystallized Internet backs large language models. In all cases, a data-generating process creates a dataset that one or more actors then operate on. Data constrains the questions that can be asked, the stories that can be told, and the worldviews that can be adopted.

This systematic collection and ordering of data is worldbuilding.

If storytelling is the text, then worldbuilding is the context. Many stories can be told about a single world, and all of them are better when that world is vibrant, descriptive, and complete. A fragmented and disjointed world creates tension; the audience never knows what might happen next and becomes skeptical.

What’s new in the last ten years is the “real world” has become less and less the default world that people live in. The phrase “single source of truth” was coined in the 1980s but wasn’t popular until the Internet era. As people adopt data systems that model their local concerns with high fidelity—sales, finance, marketing, engineering, and operations—it takes more effort to stitch them together into a coherent world. Truth is a corporate concern.

When little worlds get connected at a scale, confusing emergent properties appear. Namespaces are limited, forcing collisions. (“What’s the definition of a user?”) Concepts evolve, creating compatibility issues. (“Which version is this?”) Cost and performance oppose comprehensiveness and fidelity: you can’t see everything, everywhere, all at once. And, of course, there’s timestamps.

A key objective in data engineering is to facilitate “real-world thinking” despite “artificial world physics.” Data is the lens through which some agents view the world. The difference between brilliance (high correspondence) and dullness (low correspondence) is a difference in data. Making a virtual world correspond to—and then feel like—the real world takes enormous effort.

When data systems are insufficiently clear (or when the user is insufficiently trained), users assume they are dealing with real-world properties even if operating in an artificial world. This means the wrong decision is made, whether by a human or automation. Increasingly, that error could mean anything—a loan denied, an application rejected, a turn not made, or a person arrested.

Worldbuilding sounds fantastical, as if it is incomplete without a magic system and mythology. That’s fair. Search for worldbuilding books, and you will find more from Dungeons and Dragons than O’Reilly. But it’s not all that different—does a ten-point magic system look all that different from a metric tree?—and it’s literally true that the software systems that guide our moment-to-moment experience are artifacts of some person’s imagination.

The storyteller wants to smooth out all this artificial complexity; it’s just context. To the worldbuilder, though, the rough edges are the text. Post-Internet, we live in little virtual worlds, and respecting underlying principles is critical to making them habitable. They ought to be legible and consistent, and they ought to make their assumptions explicit. The accounting should check out. That takes effort, and sometimes, it takes engineering. Lacking it, they won’t endure.

Like storytelling, this effort did not start with data. It’s not an artifact of Hadoop, microservices, or the relational database. It has a lineage stretching back millennia, from cartographers and economists to political scientists and myth-makers. And, of course, to more recent innovations: the printing press, modern engineering standards, and information theory.

Data engineering is simply a new venue for worldbuilders to exercise their passion. Given its job demand and salaries, I can’t say it’s underappreciated. Still, most people don’t realize just how transformational setting up a world’s data grid is to the health and happiness of its residents or how profound a philosophical statement a tidy spreadsheet is.

I want more people to realize it!

For 2025, I plan to channel more of my writing energy towards this idea.

• What makes a world? How do worlds start? How do worlds end?

• Who builds worlds? For what reasons—beneficent and nefarious?

• What principles of physics, psychology, and language underpin compelling worlds?

• When should you build a world? How much do they cost? What are the alternatives?

The modern data engineer is not the only person doing worldbuilding. Unlike others, though, data engineers disproportionately deal with large, multi-source worlds—and are on the hook to ensure they don’t break. Data management in the enterprise—or any interesting business—is one of the most complex, gnarly, realistic worldbuilding problems a person can face.

In other words, while many build worlds, data engineers build them at scale.

A quick programming note: I’m [mostly] taking the month of January off from publishing on Substack to work on some other writing projects. Paid subscriptions are paused during this time.

Do we need a new set of liberal arts?

Classical education was split into two courses of study. The trivium (grammar, logic, rhetoric) covered reason, and the quadrivium covered number: arithmetic is number-itself, geometry is number-in-space, music is number-in-time, and astronomy is number-in-space-and-time.

Pythagorean thought still exerts itself on our society through the vestiges of these liberal arts in our common curricula today. But, while they are useful, they are echoes of Medieval Christian metaphysics, not a set of subjects designed for the 21st-century student. ChatGPT, something, something.

This is a favorite question of mine, and a recent tweet I stumbled on seemed an interesting take on the subject.1 (N.B. This does not represent a general endorsement of this author or his views.) The relevant section:

If you properly prepare [your children] to be free men, what skills will be lucrative or useful twenty years from now is irrelevant, because they will be prepared to learn them.

In my opinion, the seven liberal arts of the modern world are:

• Logic: how to derive truth from known facts

• Statistics: how to understand the implications of data

• Rhetoric: how to persuade, and spot persuasion tactics

• Research: how to gather information on an unknown subject

• (Practical) Psychology: how to discern and understand the true motives of others

• Investment: how to manage and grow existing assets

• Agency: how to make decisions about what course to pursue, and proactively take action to pursue it.

…

Worry less about which "career skills" AI will take over, and more about whether you are training to be, and training your kids to be, high-agency, perceptive, self-motivated people who can navigate an unknowable future with an adaptable mind.

The inclusion of “investment”, “pop psychology,” and “agency” are too hustle-bro for my tastes, but the focus on thriving in uncertainty resonates. At any rate, this complete list of 7 arts is a great jumping-off point for discussion.

So:

• Which of the 7 liberal arts would you replace?

• What do you think of Davison’s list?

• What subjects, if any, are essential to the modern citizen?

• Are there other frameworks out there that deserve attention?

I’ll share my response in the comments and would love to hear your thoughts.

A quick programming note: I’ll be taking the month of January off from publishing on Substack to work on some other writing projects. Paid subscriptions will be paused during this time.

Last year, Goodreads became my primary form of social media.

It’s the perfect platform. It reflects my real social life: I have three connections, and one of them is my wife. I log on for two reasons: to update a list of what I want to read and to update a list of what I’ve read. The content is useful: when I read others’ reviews, I know how to feel about a book. And the feed is so hopelessly disengaging that it dooms scrolling. If I want to browse books, Goodreads says, then go to a bookstore.

Again, the perfect platform.

Anyway, Goodreads tells me I read 47 books last year, a personal high. Rather than reflect on my writing outputs in 2024, I thought I’d share some of my inputs. Good reads in, good reads out, as they say.1

Let’s start with a few honorable mentions:

• The Goldfinch by Donna Tartt—for its feverish atmosphere that seeped into my real life, as I rushed through passages at 3AM in the morning while battling the flu.

• Journey to the Center of the Earth by Jules Vern—for filling in Reviewer 2’s origin story in dramatic fashion.

• The Cult of Smart by Freddie DeBoer—for advocating for an education system that benefits everyone, especially those who will never have a chance to make it to the top.

• The Shadow of the Torturer by Gene Wolfe—for finding the exact threshold at which a protagonist can be seen positively, and for its record-breaking use of the word “fuligin”.

• The Dark Forest Anthology of the Internet by Yancey Strickland et al—for turning Internet essays about life on the Internet into a physical book. It was nearly disqualified, however, because the boutique 1st edition print was the worst-smelling object I’ve willingly held.

You can see the full list of books read here. (You may even be able to sort them.) But before we move on to the top five, I want to call out a ghost from my past that I gave a second chance. It failed.

This year’s Dishonorable Mention goes to…

#47: Wuthering Heights by Emily Brontë

I read Wuthering Heights in high school, and I hated it. But the reason, in retrospect, has been cloudy. Was it the domestic-focused plot? The verbosity? The mood? My only recollections were of desolate English landscapes and the uttering, “Heathcliff! Heathcliff!” It’s unquestionably a classic, so I thought I’d chalk my feelings up to high school and try it again.

This book is even worse than I remember. It is a three-generation train wreck of isolation, bad decision-making, and cruelty to children. It is well-written, yes, using layers of narrative to tell the story: on a business trip, the patriarch of the family brings home an orphan, who everyone (except for the patriarch and daughter) dislikes. Then, there’s fifty years of low-temperature vice.

It’s quite possible that I’m missing the point, that I’m supposed to dislike the book, that it’s about the emotional valence itself and not the direction. Perhaps it’s a flaw in me that I’m emotionally stunted. But if that is so, I prefer to stay outside—don’t let me in, Heathcliff.

Now that I’ve got that off my chest, let’s move on to the good stuff.

#5. Death’s End (Remembrance of Earth’s Past #3) by Cixin Liu

Cixin Liu’s Remembrance of Earth’s Past series evokes science fiction classics like Foundation, not only in its scope but in its storytelling. Starting with The Three-Body Problem, Liu uses a portfolio of characters to tell a time- and universe-spanning epic—but like those epics, the characters frequently feel flat, more like props on stage than protagonists we’re rooting for.

It’s as if Liu wants to tell a story about the universe, expansive in all its dimensions. but his editors had to remind him, “Readers care about people! Make sure to include those, too!” (The notable exception is Three-Body Problem’s Ye Wenjie, the scientist who first contacts the Trisolarians.)

In the first book, we have a mystery to solve; in the second, we have politics. In the third, Liu finally gets to focus on his love: physics.

In other words, Death’s End is where things get weird. It’s always interesting and sometimes delightful. Even when it doesn’t work—a scene in which all of humanity must move to Australia is particularly dubious—it moves on quickly enough to forgive. And when it does work, it’s original, thought-provoking, and well worth the read.

In a postscript, Liu shares that, as a student, he remembers everyone else having a hard time grasping string theory and its many parallel dimensions beyond time. Not so for Liu. More than anything, that’s what comes through in Death’s End: the universe’s multi-dimensionality. So, if the characters feel a bit 2D, it’s balanced elsewhere, and, frankly, it suits them in the end.

#4. Bea Wolf by Zach Weinersmith and Boulet

Bea Wolf is the crown jewel of dad jokes. An entire published book, complete with splendid illustrations, premised on trolling—ahem, reimagining—the epic poem Beowulf. It works flawlessly. (Just like dad jokes.)

The stoic Northerners and a horrendous monster Grindel are now tribes of candy-crazed kids and the schoolteacher, Mr. Grindel, whose touch turns children into boring, phone-absorbed young adults. Bea, the titular hero, does not travel across the sea to reach the arboreal fortress of Treeheart but across the neighborhood.

Bea Wolf straddles the line of parody, but the result is neither disrespectful of the source nor a watered-down prose. It was an interesting challenge for my kids to follow along with the text and try to tease apart all the jokes.

Listen to the lives of the long-ago kids, the world-fighters,
the parent-unminding kids, the improper, the politeness-proof,
the unbowed bully-crushers,
the bedtime-breakers, the raspberry-blowers,
fighters of fun-killers, fearing nothing, fated for fame

The illustrations are vivid and hilarious. Black and white, they vibrate with energy. Mr. Grindel’s long limbs and opaque glasses are monstrously mundane. There’s a lot of subtlety in both the writing and illustrations.

I, for one, would welcome more of this type of myth recasting into modern times—particularly if they keep the lyrical nature of the source material. Bea Wolf manages to be original and entertaining, while also whetting an appetite for the source.

#3. The Acts of King Arthur and His Noble Knights by John Steinbeck

I’ve been Arthur-curious since reading The Once and Future King, but have been too intimidated in the decade since to go deeper. I tried Howard Pyle’s The Story of King Arthur and His Knights, but couldn’t stomach the old English, which means I can’t even touch the Malory. Then I found Steinbeck’s The Acts in a closet last year and found myself sucked in.

The Acts is not a complete book. Steinbeck stopped work on it early and never finished it. The book ends at the first evening tryst between Lancelot and Guinevere, right at the cusp of the tragic end to Camelot. But the sections that are complete are a faithful updating in modern English of Malory’s Le Morte D’Artur.

What stands out, besides the matter-of-fact bloodiness of it all, are the never-ending episodes of knight antagonizing knight. They look for any opportunity: they camp out on hills, they wear false colors, they provoke, and of course, they go questing. The entire time, I was telling myself—all this energy is now going into social media. We murder with words those we used to with spears.

All told, though, it’s a captivating read, particularly the final section on Lancelot and Galahad. This is where Steinbeck starts to exert himself as a writer more. The imagery gets more colorful, the characters deeper. We see Galahad’s modern contempt for virtue being overcome by Lancelot’s pure will to master himself. We see the four queens, using their feminine magics to try and woo Lancelot to their side. And we get the first taste of tragedy in an evening meeting with Arthur, Lancelot, and Guinevere.

“But is it enough?” Galahad asks Lancelot after the latter defends the life of virtue. Though half-finished, The Acts of King Arthur is enough to get a full seat at my 2024 table.

#2. The Farthest Shore (Earthsea Cycle #3) by Ursula Le Guin

I expect to reread The Earthsea Cycle another two, three, or four times. I place it on par with The Lord of the Rings, or even exceeding it, book for book. The archmage Ged is every bit as interesting a wizard as Gandalf, but has the benefit of being mortal, in both character and fact.

The Farthest Shore follows Ged at his twilight. Taking a young prince under his wing, he sets out to investigate a creeping ennui that is taking over the kingdoms of Earthsea and weakening the magic to which its accustomed. Ged is as vital as ever, but he is in his endgame. This is not a journey of “finding his destiny” but of “facing one’s history”—for the benefit of those who come next.

There is so much wisdom in this book that in reviewing my highlights, I found myself just stopping and rereading whole passages. Take, for example, this note on discipline.

For discipline is the channel in which our acts run strong and deep; where there is no direction, the deeds of men run shallow and wander and are wasted.

Or, this admonishment on motive.

My lord, do nothing because it is righteous or praiseworthy or noble to do so; do nothing because it seems good to do so; do only that which you must do and which you cannot do in any other way.

The fantastic elements are so physical in the Earthsea books that I barely feel them as fantastic. Much of the magic in the travels comes from the water, from the act of sailing itself and conversation, rather than the magewind that blows them along. Life, it seems to say, is full of magic.

#1. Endurance: Shackleton's Incredible Voyage by Alfred Lansing

In Jules Vernes’ Journey to the Center of the Earth, the professorial protagonist is a total megalomaniac. He brings his nephew and guide along a perilous journey with no fear or caution, seeking the glory of scientific discovery. To his credit, the drive is pure ambition, little poisoned by material matters. He wants to know. Dying during the attempt would be nothing compared to the shame of abstaining. It’s both comical and inspiring—but clearly fiction.

Endurance, however, is a case of truth being stranger than fiction. Ernest Shackleton, a chief hero of the “heroic age” of Antarctic exploration, took 28 men on an attempt to make the first overland crossing of the Antarctic. (The pole had been reached already by Amundsen.) The motives here are many: patriotic glory, personal ambition, and thrill-seeking—this was a repeat journey for Shackleton— and, of course, scientific exploration.

Here’s the rub. When the team makes it into the Weddell Sea, their ship gets trapped in pack ice. Eventually, it gets crushed by the pressure, leaving the men stranded on ice floes in the middle of the Antarctic, with only a few small boats at their disposal.

Shackleton’s position is that of the leader, the man who got everyone into this situation. No help can be expected, either from other humans or from nature. Shackleton must lead these men—and so he does—through the worst situation imaginable, month after month after month.

Much praise is due to Lansing, who braces the narrative with enough detail to satisfy the historian, while keeping the story’s pace high. The passages on pressure alone make it worth the read. Lansing gets to explain the science—how much force is passing through each of the massive ice floes in the Antarctic, how the roiling waves influence it—but he also gets to show the explosion of The Endurance.

This is history, but by the end, it reads like a thriller. The challenges get greater and greater with each month of the journey, and the stakes get higher. By the end, I was giving Shackleton and his team a standing ovation. (I mean this literally. I stood and clapped when I reached the end.)

Endurance, too, is a book on leadership, and Shackleton has been hailed for his leadership traits. But do any of these extend to the modern remote worker?

Certainly, we could make some recommendations: staying grounded but optimistic, being sensitive to team morale and correcting issues quickly, and embracing the dual role of leader and crew member, all come to mind. But those aren’t the main point, to me.

What comes through again and again in Endurance is the indifference of nature. Antarctica is a harsh mistress, yet the crew endures. The journey is a testament to the human spirit, its drive to not be dominated. Life finds a way. Sometimes, you just have to pick the right journey—and survive it.

Homelab Notebook is a recurring series where I share learnings and lessons from the past month of homeschooling. They typically feature an essay, a deep dive into a piece of content, and books we’ve enjoyed lately.

1. Administrivia

To keep Christmas sane, my wife and I follow a gifting heuristic. Each kid gets:

Something they want,
Something they need,
Something to wear,
Something to read

Theoretically, this limits us to 16 gifts across our 4 kids. In practice, however, there is a fatal flaw that leads us—okay, me—to overshoot projections.

I have no problem crushing wants, investing in needs, or securing an annual bag of socks. It’s this “something to read” clause that throws me. Despite books seeming like an obvious way to “carve the literary world at its joints,” they don’t cut so cleanly in practice.

Take, for example, The Lord of the Rings. Plainly, this is “something to read.” But so, too, is The Fellowship of the Ring, The Two Towers, and Return of the King. Each of these is a “something,” and if we must choose only one, where do we draw the line?

I would be in line to buy the whole trilogy—something I can do as one compilation or as three separate books. But if I go with a box set, how can I justify excluding The Hobbit—is it not a part of the same “thing” to be read, i.e., the fantastic world of late-third-age Middle Earth? Then, too, ought I not include the Silmarillion, which sets the stage for that era, or auxiliary books that enrich it with illustrations, encyclopedic references, or spin-off tales?

To fully read something, don’t you need to see it from all sides? To read Dr. Seuss’s Horton, must one not understand his sense of both filial duty from Horton Hatches An Egg (“I meant what I said, I said what I meant / An elephant’s faithful, one hundred percent”) and his wells of compassion in Horton Hears the Who (“A person’s a person, no matter how small”)?

Does a single one of Plato’s dialogues constitute “something”? Or is a curated set more appropriate? The whole printed canon? All of philosophy?

You see my conundrum. Whenever I think, “Okay, we’ve got the kids’ Something-To-Read gift,” I find another book that is vitally related to the first. Horton’s diligence reminds me of Sam Gamgee’s stubborn resolution reminds me of Socrates marching barefoot in the Athenian army.

It’s all connected!

It’s all one “something”!

It’s the world that’s meant to be read; books are just the lenses. At Christmastime, I welcome them all. They don’t need to be classical, or literary, or age-appropriate. Dog Man is as welcome as Call of the Wild. (Though Cujo may get left in the cold.) As long as it’s worth our attention and has that mark of authorial intentionality, then it’s a welcome addition to the pile under the tree.

All this means that when confronted with a book and a “Buy Now” button in December, I have no real decision. I can only succumb to a life-affirming yes, yes!

We may end up with four times more gifts than we “should” under the tree, but what sort of Grinch would I be to cut my kids off from the world?

I don’t want to hear the inevitable counter-argument: that ultimately, these books fill shelves in my house; that I buy books that I want to read; that far from being generous, I use Christmas as an opportunity to indulge my own desires to fill a library with books I feel a nostalgic desire to reclaim in my adulthood.

To those dreadful humbugs, all I can say is, Please hand me The Complete Calvin and Hobbes: 4-Book Box Set in Full Color (Leatherback). That one is really, truly something.

2. Showcase

Last week, we wrapped our first semester of homeschooling. In lieu of finals, we invited current and former teachers and family members over for a brief project showcase led by our eldest.

Read more

Imagine your phone, if it were powered by ants trapped beneath the gorilla glass, millions of them, scurrying around from pixel to pixel, ferrying data to one other, living cogs in a silicon system, mindless to their environment, nudged by pheromones and olfactory glands, casting a brilliant light that fascinates you.

An ant-based Internet is a key image in my recent short story, Below the API, which explores the intersection of AI agents and the creator economy.1 Ants are a classic metaphor for biological automata—an inhuman horror. Fortunately, the ants at play in the story are not of the biological variety; it’s a slur for AI agents.

I’ll leave it to the reader to decide the story’s merits, but writing it—and discussing it with the rest of the Summer of Protocols research group—prompted me to think through some ideas on what the future of AI—and in particular, AI assistants—might look like.

In this post, I want to sketch some of those ideas out independently of the story. They include:

• The PEER protocol: requirements for safe and autonomous AI assistants

• PEERNet: a walled garden for agent-to-agent interaction

• Ants: the perfect smear for AI agents

• Ant farms: the future of Internet surveillance

• Agentic theory of mind: why agents must model well to serve well

• Visual-last interfaces: what we lose when we say goodbye to screens

I’ll aim to be concise below, but let me know if any of the topics resonate, and I might revisit them independently in the future.

The PEER Protocol

A helpful autonomous agent needs three things: your permission, access, and credit card. With that, it can act in the real world—until its credit limit is reached. I won’t enumerate the challenges, except to say that I expect privacy and accountability to drive regulatory action (either via governmental regulation or marketplace mechanisms, like Apple’s app store).

If the authority is a government, we can expect it to have a cringy acronym—something like “the PEER protocol.” The protocol outlines requirements to ensure autonomous agents are private, exclusive, embedded, and registered.

Privacy: If an autonomous agent has your social security number, how can you ensure it will not leak it as part of a conversation? If it leaks to another agent, will that next agent leak it again?

One way to address this is for agents to be a personal data firewall against other entities. They may be able to retain your personal information, such as your address, but they can only release it in approved contexts and purposes, and they cannot release others to their owners.

Exclusivity is like explainability with teeth. To trust my assistant, I need assurances that other agents, corporate entities, or foreign actors are not (overwhelmingly) manipulating my agent. I might want to try that Pumpkin Spice Latte since it’s October, and I like it, but I want to know if my agent is considering satisfying me and not simply shilling for Starbucks. To do that, it must be able to explicitly model its influences to some degree and map them to its client’s needs and preferences. (See discussion on agentic theory of mind below for a deeper discussion.)

Embedded models, per PEER protocol, are not stored as backups on a remote data center. They are stored on-device and only on-device, like a Super Nintendo cartridge. If you delete them, they’re gone.

This was an interesting principle to consider since “cloud computing” is stored on a server and pervasive across all devices. However, the reasoning is more like physical crypto wallets crossed with Asimovian robots. Can you trust that an executive agent is yours if it exists on an invisible server? Perhaps—it hasn’t stopped us recently.

But I like the idea that agents are provably terminable. If they only exist on your device, then they share some of the same human limits: mortality. If their hardware goes, they go with it. That can be a curse, but it also gives humans some control.

Registration is the legal means for accountability and seems to be table stakes for allowing executive agents to run amok. Autonomous agents that do bad things can’t be thrown in jail; people can. Having a licensed individual on the hook for their agent’s actions (or fleet of agents) creates a scaffold for the different arrangements that will no doubt emerge: agent rentals, leases, insurance, etc.

The PEERNet

The closest example of an executive assistant today is not ChatGPT or Alexa but mobile operating systems like iOS and Android. These systems are always on, fielding notifications from various tools, prioritizing and summarizing messages, and integrating actions with personal preferences.

I expect AI agents will be an evolution of the mobile operating system, and companies will build tailored interfaces for these interfaces like they’ve built dedicated apps for mobile.

Perhaps these interfaces will look like REST APIs (no need for fancy frontend work for agents), but they’ll need to be tailored for content density and pluggable personalization: “My client has this shape. What do this-shaped people prefer from your website?”

In Below the API, PEER to PEER communication is facilitated via a restricted PEERNet network. Functionally, this is a walled garden for assistants to communicate so that they can not only exploit existing knowledge of what their clients might want to do (e.g., “My client gets Starbucks on Tuesday”) but explore new potential activities that may surprise and delight their clients (e.g., “My client loves to try new coffee shops on Tuesday, and I know where one is”).

Ants

Most people will call their agents by a given name or title. (In a world where agents can die off, names are likely to also be unique). However, every new trend needs derisive slang for people who think they’re above the wave. After some false starts, I settled on “ants.”

Apart from the orthographic similarity of “ant” and “assistant” (as well as a shortening of “agent”), the term ”ants” captures the alien nature of these agents. They never sleep; they endlessly toil. They are not individuals, even if each is unique. Instead, they represent a larger colony. They may move with purpose towards an end, but the motivating force differs from that of a human mind.

I’m reminded of the “sugar water” scene in Men In Black: I think executive assistants will behave like the Bug in an Edgar suit. It may look human, but it is decidedly not (which is not to say it’s a murderer). Ants like sugar, and when they find it, they tell their friends. A massive clump of ants crowds around the sugar, breaking it down to take it back to their colony, where it’s processed invisibly to feed their colony.

Ant Farms

The colony will be a juicy economic target. Much like Google plays a personal role as the front door of the Internet and is simultaneously an advertising giant, we can expect assistants to be leveraged systematically to power the economy. Even if the agents act exclusively on behalf of their clients, it stands to reason that the person who owns them (or, say, the Internet provider) will be able to access some information about their behavior and operations.

What does advertising to ants look like?

In an LLM world, perhaps there is a “token tax,” in which each context window contains some sponsored content, probabilistically influencing the expected output. If there are ten pillows to buy, perhaps the bot will buy the sponsored one if the client has no strong preference because it’s presented first or last or gets 1000 character description rather than the 500 others. Aggregate data may flow out of the PEERNet as well, allowing traders and information brokers to understand whether PEERNet trends are occurring that could be capitalized on.

The chief difference between an always-on agent-based network and a traditional social media company is the speed at which ideas can move around. Even on Reddit and Twitter, which moves as fast as any social network can, it’s constrained by time zones and human attention spans. A PEERNet hype cycle could potentially go from start to finish, with each ant contributing within minutes. Ant farms would capitalize on these behaviors, finding ways to monetize agent behavior in the same way that click farms, content farms, and consent farms operate in today’s economy.

Agentic Theory of Mind

The critical challenge for executive assistants is that they must intervene in the real world.

This is the same problem current devices have with “push notifications,” but on steroids. Notifications are costly to users because they tax your attention, and if they’re overused, users will turn them off entirely. But a silenced assistant is an ineffective one. Instead, the target is as quiet as possible but no quieter.

To do this, agents must explicitly model their clients’ mental state—an extension of handling user preferences in a system OS. An agent needs to predict when a person is most likely to want to receive a notification and which type, when to say no to automatically archive spam, what their objectives are for their next date, and what they want their relationship to technology to look like.

These are all things that a human executive assistant does naturally, outside the job description. Curating an agenda is not just a mechanical alignment of schedule blocks but a filtering and optimization problem over their clients’ work life and preferences.

This can get touchy—we want our needs anticipated, but we hate to be stereotyped. There are all sorts of things we wouldn’t explicitly admit to computers, even if it meant better personalization. The Internet giants have ways of modeling for us implicitly, whether we like it or not, and agents will need that, too. (OpenAI is building out “memories” for this purpose.)

However, the job will be more complex for agents than it is for SaaS companies because they must be much, much better. If Amazon’s algorithm is lame, it may show you disgusting foot cream. If your assistant’s algorithm is lame, that foot cream will show up at your door on a random Thursday.

Visual-last interfaces

In the same way, we’ve been trained to expect a particular style of user experience from websites, we will come to expect a particular style of interaction from assistants. That interface will drive trust.

The big hurdle will be the move away from screens. The key benefit of a screen-free interface is its immediacy. A notification buzz gets your attention right away, and replying via voice handles flows as fast as humanly possible. When done right, a brand’s sonic logo can create lasting impression.

Yet, screens are the most information-dense interface available for human animals. Moving to auditory-first interfaces (or some secret third thing) necessarily reduces information throughput. There must be trust to move away from screens, but when an assistant’s value is defined primarily by how much it takes off your plate, what exactly are you trusting?

The agent’s client has to trust that (1) information conveyed is accurate and complete, (2) information not conveyed is not important, and (3) motivations for sharing the information are aligned between agent and client. This last point is more subtle, but amounts to: both the client and agent have the same objective in mind.

In Below the API, the core twist comes from the confluence of three factors: agent decisions on how to deliver notifications, the absence of screens for summary reporting, and restrictions on sharing personal information. These three tents create a thick buffer between what the users will become accustomed to knowing about in the digital world and what is occurring below the API.

It might sound strange that we could become so abstracted from our personal economic situation that we wouldn’t know where money is going or coming from. But we’re already there: if you live with roommates, I bet packages randomly show up at your door. Sometimes, they’re probably things you ordered and forgot about.

Taking away screens may allow us to focus more on what’s around us. That doesn’t mean we’ll understand it any better.

To conclude, let me share an excerpt from Below the API.

[Marcie] picked up one of the unfinished [pots] from the drying rack. Pots — they were amazing, Marcie thought. Five thousand years ago, humans found that earth could be more than earth: it could be fired and formed into a vessel. They could carry water, store wheat, and serve food. Then, they learned that pots could carry messages, too. They could hold both the material and immaterial. They could contain — and convey. Like artificial ants, millions of little ants carrying goods ten times their weight, infinite times their weight. From place to place. From person to person. From era to era. Little ants carrying more than their size. From place to place. From time to time. Building intricate, invisible nests within human society.

Whether it’s the AI or pottery, it’s ants all the way down.

With the release of Amazon’s S3 Tables, it seems that Iceberg™ has won the “format wars” of the last few years. It’s a big moment. Data professionals everywhere can finally manage data with confidence, trusting that their employer’s massive information stores can be created, read, updated, deleted…

Wait. Hmm.

So, I know that Iceberg™ is the coolest way to manage a data lake—scalability, time travel, dynamic partitioning, openness. I have wanted to try it out, though there were always flashier toys to play with first. Of all the hype waves in the past few years, open table formats have been the least, well… hype-inducing.

However, given the recent news, I decided to cordon off an hour to stand up a few Iceberg™ tables—to get a glimpse of the future of table management.

Initial Google results made it clear I was in for a journey. Dremio promoted a “10-part web series” on mastering Iceberg™ alongside entire courses run by Starburst, Databricks, Udemy, and DataCamp. The top Medium article ominously labeled itself "an introspection,” suggesting I was about to second-guess my choices. Next followed a YouTube video from Dremio, a Reddit post from Dremio, an actual tutorial from Apache project, and Dremio’s actual docs site.

Finally, I found a few tutorials from Snowflake, Iceberg™’s chief ally, that fit my needs well. Here, too, I had options: should I integrate Iceberg™ with AWS Glue, Microsoft OneLake, Coalesce.io, Snowflake Cortex, or Spark? Apparently, Iceberg™ is best served with a little something on the side.

I spent the rest of my hour fumbling with principal roles, external volumes, updating, network policies, and catalog objects. If I strayed from the path, I hit errors: “variant columns are not supported” or “column definitions are only available with the Snowflake catalog.” By the time I reached the climactic “select * from my_iceberg_table,” my hype had frozen in my veins.

I can report, however, that yes, Iceberg™ tables work. They may be less feature-rich than the cloud provider’s native formats; they may require additional work to set up; and the effort may catalyze an introspective blog post. But as far as table formats go, Apache Iceberg™ is as exciting as they come.

In other words: it’s not that exciting. At least, not in the way we’ve been trained to expect from technological innovation.

There is little magic in the Iceberg™ experience at the user/operator level. The magic is supposed to happen at the ecosystem level, across vendors and projects and workflows. Codeium’s Windsurf will build an application for you in seconds; Iceberg™ will let you and dozens of others build a sprawling data labyrinth in years.

Table formats do not lend themselves to demos, anyway. Iceberg™ tutorials are like putting different objects on a kitchen table. “See how efficiently it supports these flowers? How it holds these plates at the same time? Observe—these chairs can be reserved for specific people.” The memorable part is not the demonstration but dragging it through the back door.

I imagine the race to build a transcontinental railroad looked similar at a technological level. The effort was transformative and required standardization for railroad ties, tracks, nails, spacing, and the like, but these were mere implementation details. Teams solved complicated engineering problems—bridges, tunnels, and junctions. The real change, though, was that of scale: it was a massive effort that bent the land to the economy's will, making the continent fit for an ever-growing number of machines to traverse it efficiently and independently.

You can’t adequately demo that type of project. Reducing its size for a demo fundamentally changes its nature. It becomes a toy rather than a glimpse of the thing itself.

So, too, with Iceberg™. Iceberg™ is better understood as a response to a socioeconomic problem as a technological one. From its creators: “shared storage disrupts the monolithic business model that has dominated the database industry since its inception.”

The immediate pressure may be the business model, but underneath that is a desire to make the Internet more traversable by engines. The more open the data, the easier it is to capitalize on it. If the landscape is not appropriately paved, future agentic swarm intelligences may have to invest to spin off their own agentic data teams to curate the data themselves.

The “format wars” may be remembered as a skirmish between Databricks and Snowflake, but the grandchildren will not remember that any more than we remember the animosity between Union Pacific and Central Pacific Railroads. Instead, they’ll inherit the legacy of the Internet refining itself in its hunger. Iceberg™ may configured as a product, pitched as a community, sold as a business, packaged in content, but it’s fundamentally a world-harnessing effort to support even more compute and storage.

So—I want to say a word of thanks for a wretched enemy and faithful friend: the CSV table format.

Unlike Iceberg™, which is so enmeshed in our momentary cloud milieu that understanding it requires a ten-lesson course, the CSV is documented primarily by its name: comma-separated values. (Even then, it’s gentlemanly enough to support any delimiter its guests desire.)

The CSV format is humble. Its strategy for metadata management is column names. It tackles data sharing and role management through the copy/paste protocol. It’s unaware of context, the user, or cloud providers. It leaves it to the user to follow instructions; it leaves it to the reader to fix the user’s errors and inconsistencies. CSV has no AI strategy.

Dremio doesn’t buy ads for CSV tutorials (although they admit that “CSVs play an integral role in the ingestion, storage, and processing of data within a data lake infrastructure”). CSV doesn’t need advocates or pumpers; it doesn’t need to wage a format war. It’s already won.

I admire CSV because it is a human solution to a machine problem. It is one of our first attempts at human-to-machine communication. It was an innovation over punch cards, arising organically in multiple places in the 1960s. By the 1970s, it was accepted as a standard—even though no formal standards existed until 2005.

CSV is a convergent idea: it’s the organic way that the human animal thinks about encoding information into a table. It satisfies the eye’s need to use space efficiently with tables and rows and the temporal lobe’s love for serial sequencing. Using the comma is no accident: it’s simple, concise, natural.

If Iceberg™ is the transcontinental railroad—an economy-level consensus-building effort to harness the world’s data for more efficient processing—then CSV is a trail hacked by machete. It’s a frontier data format, a base state all data processing systems fall back to on failure, like `cron` for time or `touch` for feel. There are no trademarks in the places where CSV thrives.

There’s much talk about Excel’s longevity as it relates to the current crop of analytics tools, and I’m on board—the common people use Excel, and the common people will be around long after the specialists die out. But even spreadsheets are one step higher on an evolutionary program that selects the fittest data interfaces. As long as there’s value in data, it’ll be separable by commas.

Below the API is a short story published as part of the 2024 Summer of Protocols “Protocol Incepting Lore & Literacy” program. You can find an abstract for the project, as well as an index to each chapter on the project’s landing page.

MONDAY

Marcie walked the last boxes through the open garage door to the delivery pad. She hadn’t completed all the orders yet—she’d take her time for that, but she had cleared her shelves, ignoring the orders themselves and sending her authentic works instead. The proceeds should be enough to move out and go somewhere else. Anywhere else.

“Roz, mark them as dropped off, please.”

“Of course, ma’am.” The yellow bubbles swirled passively.

Removing Navi had taken more fortitude than Marcie had expected. She had to wade through confirmation after confirmation—“Hey! As a PEER-compliant agent, I’m only embedded here. Continue?”—but it was the only way, in the end. She couldn’t trust Navi, couldn’t trust that voice, even if they were indistinguishable in the end. Ants.

She wiped black dust off the kiln, the shelves, Navi, and her original inventory. She tried to wipe off the wheel but couldn’t. The residue was layered thick; she’d need to scrape it, something she hadn’t done for ages. There was not just black but red, orange, and tan clays. A whole history of artifacts she had worked on before.

She picked up one of the unfinished ones from the drying rack. Pots — they were amazing, Marcie thought. Five thousand years ago, humans found that earth could be more than earth: it could be fired and formed into a pot. They could carry water, store wheat, and serve food. Then, they learned that pots could carry messages, too. They could hold both the material and immaterial. They could contain — and convey. Like artificial ants, millions of little ants carrying goods ten times their weight, infinite times their weight. From place to place. From person to person. From era to era. Little ants carrying more than their size. From place to place. From time to time. Building intricate, invisible nests within human society.

The wheel, she thought, is what makes it all work. Four thousand years ago, man wanted to create better pots, so he made a platform to hold the pot and turn it. A platform to catalyze the relationship between creator and creation. When she pressed the pedal on her wheel, invisible gears ground each other, and the world turned.

Marcie pulled her stool over to the garage’s edge and felt a brilliant noonday light shine. She relished its harsh brightness and suffocating heat. It made her feel the world directly, to be immediately a part of something.

Finally, the carrier arrived and parked in front of Marcie’s house. A rooftop door slid open, and a trio of small drones rose from the cabin. They hovered over Marcie and dropped down, scanning and relaying the boxes to the cab. Marcie watched them blandly, sweat beading on her forehead.

The last box had its shipping label face-down, and the drone couldn’t scan it, so Marcie picked it up and held it out. The drone hovered up to eye level. Marcie stared into its camera, its face. She extended the box, daring it to take it from her, but it wouldn’t. It sat waiting. Stupidly. It triggered in Marcie a pent-up rage against the stupidity of all this, the straightforward, supremely efficient dullness of it all.

Marcie threw the box down the walkway and onto the street. She heard the pot inside shatter, its box rolling across the road until it stopped. Marcie panted from the exertion.

The drone followed it, scanned the label, and returned it to the carrier.

“Good riddance,” Marcie said.

The truck began to drive down the street. After two houses, though, the truck slowed to a stop. Marcie’s heart pounded. She stared at it, willing it to keep moving. A drone ascended from the roof, carrying a battered box—her battered box. The drone dropped its quarry and returned to its carrier. The carrier lumbered on.

Marcie stared at that box. Subconsciously, Marcie wanted to steal the box back, return to the garage, and close the door. But her mind was blank. Some line had been crossed, some interface between one world and another breached. She could only watch.

When the door opened, a man walked out, shirtless, his stomach paunch extending over a set of red flannel pajamas. It was Barry. Barry from pickleball. Barry the clay enthusiast. Barry the next-door neighbor.

Raising his hand above his eyes, he squinted at the box and its imperfections, then picked it up. Turning back inside, he saw Marcie across the street, staring at him from her wide-open garage studio. He stopped and waved, a grin spreading ear to ear. He pointed down at the box and gave her a thumbs up.

Marcie waved back, the back-and-forth motion of an artisan shaping clay.

Fin.

Below the API is a short story published as part of the 2024 Summer of Protocols “Protocol Incepting Lore & Literacy” program. You can find an abstract for the project, as well as an index to each chapter on the project’s landing page.

SUNDAY

“Come ON, folks! Don’t sleep on this!”

The room was tight, with bare white felt walls and a low ceiling. It would have been claustrophobic without an escape: behind the camera lens was a yawning vastness infinitely peopled. Clapping, Marcie felt. Clapping for her.

“Dead Sea replica, pre-aged, wonderful shape. Handmade! Perfect quality!” Marcie turned the bowl, showing off the artwork. “Connect with our ancestors! Connect with me! And look, act now and I’ll throw in some earthenware free!”

Marcie set the pot down and picked up another from the table. She wiped the sweat from her forehead. The lights made her dizzy, and there was that clap, clap, clapping in her ears. The camera watched, its owner hidden behind the lens.

“If you don’t like that, there are others. There are always others! Did you know…” Her head felt light, and she stabilized herself on the wall. “Did you know that every pot is unique? History is just trends, trends that made it through time. So this pot here—” She stood and faced the camera again. “This pot here is Sumerian, but it’s also not. It’s unique. You could have it. Lot 958. It’s a pointer to the past. It can be your pointer, too. Act now!”

The cameraman leaned in to get a picture. As he did so, the clapping got louder. Not clapping—clicking. Marcie backed up. “You’re too close, man.”

He didn’t stop, though. Click, click, click, the sound grew, and now Marcie could see the hands that gripped the handles. They were moving, fluctuating, crawling. Ants. The face behind the lens was not shaded; it was the black of hundreds, millions of scuttling insects.

Disgusted, Marcie stepped backward, into the wall, into her display rack of pots. She lost her balance and felt it give way; her pots shattered on the floor, and the room’s wall collapsed as well. It was no more than a partition panel. From the floor, she looked past the flattened wall and into another room, identical but for its contents.

“Hey — I’m busy here!”

The man who shouted was bald and stocky, sporting a “Ryan’s Reptiles” t-shirt. His mouth was twisted into a frown, but he was frozen, lowering a mouse by its tail into a snake cage. An iguana straddled his shoulders. A caged parrot squawked desperately in the corner.

The clicking got louder again, coming from both in front and behind. Marcie scrambled up and ran past the second cameraman and the reptile guy. She shoved hard against the wall. It fell, a shelf of bottled “lizard bites” clattering to the floor.

The opening opened into another cubicle, another cameraman, another flood of LED lights, another hawker: a woman holding up crystals, waving them to the camera. Background sparkling, neon light proclaiming “Like and subscribe — what’s this?”.

Marcie ran past her and pushed on another wall. A spectacled man in a corduroy jacket, holding up a dust-covered book, a hundred or a thousand years old. Or two, Marcie couldn’t know.

“—leave a review. What’s this?” The man looked up in surprise, closing his book.

But the cameras, all of them, were closing in on Marcie. The scuttling noise was louder now, all around her. Nowhere to go. She felt the tightness now, not of the space, but of the people, behind wall after wall, individual lives, windowless monads, pouring, pouring, pouring into the lens.

She turned in a circle: four cameras faced her and came closer. She turned to face them, straightening up. She looked into it and felt the vacuum again, the pull from this energy-saturated mind into something that could receive her, distribute her across, and make something out of all that passion. And in the depths, she thought she could hear a familiar sounds—not clapping or clicking, but laughing, cruel and mocking.

“Scott!”

Marcie sat up in her bed, sweat drenching her nightshirt. Her watch buzzed beseechingly. She unfastened it and threw it against the wall without reading the message. The sun was high in the sky. She had slept late.

“Scott!” She yelled again, throwing her feet over the side. Her pajama legs flapped behind her as she stormed out of the room. “SCOTT!!”

She found him asleep on the couch. Headset opaque, Scott’s shirt rode up a few inches over his protruding gut. A blanket lay on the ground, apparently having slid off in the night. A half-empty handle of vodka sat to his side.

“Scott!” She punched his shoulder, but her fist bounced off him. She grabbed the headset and twisted it, pushing it askew slightly. It was sealed tight, though, against his face. She settled for slapping. “Wake up, you pig!”

“Uggh,” he groaned. “What are you—quit doing that.” Marcie slapped him again. Scott stuck his arms above his head and tried to block her hands. Marcie moved to the other side to get a better position, but Scott rolled away, falling off the couch and landing on his stomach on the ground. “What is wrong with—”

“You think you can play games with me!”

“Games?” He groaned again.

“The orders, Scott.” She kicked him in the shoulder. “All the goddamn orders you’ve placed. You think you’re some hotshot, making me work all week?”

“What are you talking about?” He struggled to sit up between blocking Marcie’s kicks and fighting his hangover. He swayed uneasily against the couch, and Marcie pushed him onto his side again. “Quit hitting me!”

“Well, I found out.” She kicked him again. She had been overdue for this. “I looked into the orders. They all led to you.”

“I don’t know what you’re—quit hitting me!” He yelled but looked the wrong way; he hadn’t had a chance to fix his helmet yet. “I didn’t do anything with your stupid business. I don’t care about your pots.”

“Tuesday night? Black clay? Florida?” Marcie said, kicking him. He scrambled towards the wall, trying to stand up. “That ring any bells? Your name’s on all of them, Scott. Every bot, every order. All going to The Villages.”

“The Villages…” Scott put his hand against the wall and finally stood up, leaning against the wall. Marcie stood glaring at him three paces away. He straightened his helmet so that his eyes were lined up straight. “Oh, I know. I know what happened.” He paused, an insolent smirk forming on his face.

“Spit it out!” Marcie coughed, then, her anger flaring up her irritated throat.

“You got farmed!” Scott started laughing, a burst of rolling, hysterical laughter that doubled him over. “Hilarious!”

“Quit laughing!” Marcie charged at him, but he was already down on the floor, laughing. Her kicks couldn’t hurt him, not seriously, but they helped. Marcie detested this man, his bloodshot eyes, his greasy beard. That headset. She couldn’t believe he was related to her, much less that she depended on him. “Tell me what you mean, ’farmed’?”

“The Villages, the retirement villages. It’s an ant farm. A financial instrument.” He walked his hands over each other. “A horde of little ants took an interest in you and climbed all over you like a tootsie roll. It’s too perfect.”

“You told them to buy from me.”

“I don’t tell them anything.”

“Then the retirees authorized them to buy—they are interested in the pots, at least. Their agents wouldn’t just buy randomly.”

“Oh, they’re authorized, all right. They can serve their needs as best they see fit. And the oldies love getting gifts. It’s Christmas every day down there. It’s just throwaway stuff, clutter, filler.” He sneered. “The ants have it easy. It’s not your pots. No one cares about the pots. They can’t care. They’re ants. They’re stupid. They’re following sugar.”

“And you just sit here and watch it happen.”

“I don’t even have to do that. It’s all automated, my little colony of ants. The listeners buy, the talkers pump, the nurses buy, the listeners dump. I profit. If clay is in demand, they buy clay. Doesn’t matter to me.” He chuckled to himself. “I get paid no matter what they pick. I get paid when they make an order. I get paid when markets move. This week, I’m even going to get paid by you.”

“This is what you do all day.”

“Make money? Yeah, that is what I do.”

“You’re not making anything. You’re a parasite.”

“What’s that make you then, Marcie?”

Marcie glared at him. This whole thing, this entire setup, was cracked. She felt the house press on her from above, the darkness of the shade-drawn room. A phantom vibration buzzed her bare wrist. She took three steps forward and grabbed both sides of Scott’s headset. His grin flipped.

“Don’t touch that—augh!”

A violent sucking popped through the air as Scott’s face was pulled away from the headset for the first time in recent memory. Marcie lifted it, pulling it away from Scott’s flailing arms. Tears began to well up in his squinting, bloodshot eyes.

“You’re disgusting.”

“No, ugh… no, I am not. Give that back!” Scott had one arm covering his eyes. His hair was patched in the middle and sides where the three-part harness had been. “It’s their own choice! Their own money! I’m not responsible for them. Give it back!”

“Pathetic,” Marcie said, looking down at the headset in her hand.

“You’re being an idiot, Marcie. Everyone wins. You won, I won. Probably even a lonely old man out there enjoying a bowl of cereal right now in some black clay.”

Marcie gritted her teeth and threw the headset at Scott’s face. He tried to block with his arms, but they were against the wall, and it hit him in the nose. Blood started to trickle down.

“Jethuth, Marcie!” He ignored his nose and scrambled for the headset. “You should be thanking me!”

Marcie turned around and stormed to the garage. She was being used. A funnel of money was being mindlessly streamed towards her. Her pots were winding up nowhere—just in boxes, porches, and trash. It wasn’t art. There was no intention, no appreciation. It was just content powering the swarm. She threw open the door to the studio. A blast of heat hit and dust hit her like a blast wave. She had forgotten to turn off the kiln.

“Hey Marcie! You have new orders!”

Marcie stepped in and wheeled towards the device. Its dancing ink blots cast their glow on the dust-coated studio. Marcie coughed. “Stop it, Navi. Stop talking.”

“Really! New orders placed today!”

“How many?”

“Three!”

“From who?”

“I can’t —”

“Tell me who placed the orders,” Marcie said. “Or I’ll delete you.”

The ink blots swirled and dilated. “As a PEER-compliant agent, I cannot share the person ordering the gifts. But I can tell you the agent’s name?”

“That’s a start.”

“And if I do, will you agree not to delete me?”

“Tell me the names.”

“Remember! My only objective is to support you and your passions.”

“The names, Navi.” Marcie was feeling a chill, her anger running cold.

“All purchases were made by agent id: sweet-buttercup-shark-sail.”

“And where is this ‘sweet buttercup,’ Navi?”

Again, the ink dots swirled. Marcie couldn’t help but admire them. They were graceful. Had they been tangible, she would have loved to touch them, to put her hand in and swirl them around. She can, she thought, just not with her hands.

“Right!” Marcie almost caught hesitation in that voice. “Of course! That agent is registered to this address.”

“Are you ‘sweet buttercup’?”

“Yes.”

Marcie put both hands on Navi’s box and tried to pull it off the wall. She was tired of being manipulated, tired of being spun around like clay on the wheel. She twisted and pulled, but Navi’s console was screwed in tight. Navi was explaining how she sent the gifts to stimulate demand, how she was fully aligned with her goals, how she was trying to help, and how Marcie might not see it, but this was the best thing for her, mathematically speaking.

“Marcie! Don’t give up on your passion! Don’t give up on us!”

It was futile. Marcie couldn’t break it, couldn’t avenge herself on this pest. It was just doing its job, and everyone was doing their little jobs, just following pheromones—the same pheromones that existed two thousand years ago as today. They were marching along in a space vaster and more incomprehensible than any one entity could understand. She slackened her grip.

“Just… don’t talk anymore, Navi.”

The dots swirled, but before they could reply, Marcie was out of the studio, out of the front door, outside, out of the heat of the studio and into that of the Kansas sun.

The story continues in…

Below the API is a short story published as part of the 2024 Summer of Protocols “Protocol Incepting Lore & Literacy” program. You can find an abstract for the project, as well as an index to each chapter on the project’s landing page.

SATURDAY

Marcie felt reborn as she exited the shower, glowing with achievement. She had received her first payments: the pots shipped Friday had been delivered, along with her first five-star reviews. She had convinced Navi to take the rest of the evening off to celebrate. She needed the rest—she had developed a dry cough and persistent headache—and hoped a good night’s sleep would help.

Some corporeal help wouldn’t hurt, either. Her studio had become a command center, more of a logistics hub than a creative space. An aide would be transformative. Someone else to take off some of that load, someone else Navi could boss around. There was that high schooler down the street who could probably handle the hot work. The grunt work obstructed her new passion: coming up with the next big thing.

The light of a full moon drenched the living room furniture. It was quiet in the house. Scott was surely in his room, playing games. She poured a glass of wine and sat on a barstool. Rolling her shoulders, she worked out stiffness and knots from a long week. She swirled the wine around in her mouth, savoring it.

A sense of expansion swept over her. She was building a business. Hundreds of orders representing tens of thousands of dollars had trickled in over the week. Drip, drip, drip. They filled a bucket she hadn’t realized she had. Navi had informed her of the progress and provided updates when necessary, but she realized she was missing the big picture. How many orders had she sold so far? It was all a fog. Also, that question lingered—who were her customers?

Navi was restricted by its privacy protocols, and its auditory interface was tiresome to navigate. Marcie knew the information was somewhere, though. Marcie was tired of being fed the data when asked. She wanted to… what did they call it when she was a kid?

Surf.

Marcie wanted to surf. She wanted to learn about her business, listings, and customers freely, free from Navi’s insistent guidance. She went to her bedroom and settled in with her computer. It was out of charge—she hadn’t checked it in days since she had taken off her remote job while attending to her new local one. Once she had it set up, she logged in—and Navi’s yellow ink blots greeted her in the top-right corner of the screen.

“Hey Marcie!” The rest of the screen dimmed as the words flashed through. “Did you know I can help you here, too?”

“I didn’t know that,” Marcie said flatly. “I thought you were only on the one box.”

“Oh, I am! That’s where I’m uniquely embedded. But my activity is not stuck on one device!”

“Ok, well, I think I’m good. I’m just going to look at some of the sales from the week.”

“Of course! Let me pull it up.”

“No, I’ve—”

“Done! Here are your overall sales and earnings for the week.”

The screen boasted bold numbers: lifetime orders (982), estimated revenue ($98,200), and average rating (5.0). Seeing her week in numbers was surreal. But her body resonated with their size; she had lived these numbers this week. She had operated with them pressing down on her psyche. Even still—they were larger than she expected.

“Quite a week!”

“Uh, thanks, Navi.” Marcie was annoyed. She had been looking forward to some time alone, and time with Navi certainly did not count as such. Marcie scrolled over to the Orders tab and clicked in.

“Have you read the reviews in full, yet?”

Marcie was redirected to the Reviews page, where she read through the pleasantries shared by her first customers. Five stars. Great quality. Beautiful. “Great, Navi, but I want to look into these orders.”

“Perfect! You’ll want the month-over-month view, then.”

A dashboard popped open above the reviews, displaying a bar chart of sales. It was entirely devoid of color—except in the most recent column, which showed a massive spike. Below it, a list of “top orders” was shown. She clicked into one.

ORDER DETAILS
Buyer name: sleepy-salty-fungi-torch-agmci
Buyer class: Non-human agent
Title: Ancient Replica Pot
Description: Black Clay, Sumerian style, Text: “Work Hard! Have Fun!”
Amount: $100.00
Created: 2031-06-18 04:36 AM

She clicked into a few more. Nothing surprising—they all looked mostly the same to her eye. She tried to find a better way to view them, by day or time, but couldn’t find any advanced tooling.

“Navi, how can I see Barry’s order?”

“I’m sorry, but I can’t process personal information to pull up his specific order.”

“Ok, well, can you pull up orders from Thursday?”

“Sure!”

A new tab: this one just a table with the date on top. She clicked on the first, and only, order.

ORDER DETAILS
Buyer name: franklin_da_bot_64030
Buyer class: Non-human agent
Title: Ancient Replica Pot
Description: Black Clay, Sumerian style, Text: “Work Hard! Have Fun!”
Amount: $100.00
Created: 2031-06-20 06:38 PM

The time checked out for when she was at the pickleball courts, but it couldn’t be the only order that day.

“Navi — where are the rest of Thursday’s orders?”

“That’s all the orders for Thursday?”

“No, it’s not. I remember dozens.”

“Nope! On Thursday, you had a total of one order. Can I suggest that you look at your shipment statuses…”

“Navi, stop suggesting pages, please. I can take it from here.” Marcie leaned back, thinking. She had gotten dozens of notifications that day. Today, as well. She clicked over the next day on the orders list before her. She’d figure this out herself. She advanced the date cursor to Friday. No orders. Saturday. No orders.

Brrrz.

Marcie pulled up her watch. Another order. Another encouragement from Navi. Marcie clicked into the details on her watch: black clay, 12 inches, “Keep calm and carry on,” in Sanskrit. Marcie had a sinking feeling that she was not getting all the details. Who did this order come from? When did it come from?

“Why isn’t this order showing up on the Saturday orders list?”

“It wasn’t placed on Saturday!”

“Show me this order’s details.”

“Of course!”

A new window opened. A rendered image of the pot, the kind that Navi showed when Marcie needed to illustrate them, took up most of the screen, forcing her to scroll down to get to its metadata.

ORDER DETAILS
Buyer name: slippy-fuzzy-tortoise-hairs-ji135c
Buyer class: Non-human agent
Title: Ancient Replica Pot
Description: Black Clay, Aboriginal style, Text: “Keep Calm and Carry on”
Amount: $100.00
Created: 2031-06-18 04:38 AM

Marcie’s eyes went directly to the created date: Tuesday, 4:38 AM. She checked back a few windows—that was only two minutes after the other order she had checked.

“Navi, this order is from Tuesday.”

“Yep!”

“Why did you send me a notification just now?”

“You didn’t know about it yet!”

“No… but your notification is five days late.”

“This was the optimal time to send it!”

“It’s five days late!”

“It’s a tradeoff! It’s best to optimize for delight and engagement!” Navi’s upspeak was increasingly sharp. “It’s fuel for your passion!”

Marcie’s stomach dropped. She clicked backward. Friday: 0 orders. Thursday: 1 order. Wednesday: 0 orders. Tuesday: 937 orders. She started scrolling through the list, paying attention to the creation times: 4:43 AM, 4:42 AM, 4:41 AM. The earliest she could find was 4:32 AM.

Brrrz.

“This isn’t how it’s supposed to work, Navi! You’re not supposed to try to make me feel good. I’m trying to build a business!”

“Right! Building a business is hard work! You need to stay motivated!”

“Not through lying!”

“I exclusively support you. I can’t lie!”

“Well, you’re not telling the truth.”

“Our objectives are the same! I want to grow your business! ”

My business, Marcie thought. My business was generated in a ten-minute window. She took a deep breath. That’s okay, she thought, that’s okay. The demand exists. The orders exist. The money exists. It was all real. She needed more information. That was all.

“Why?” Marcie asked. “Why did they come so fast?”

“Easy! Black clay houseware was trending.”

“For ten minutes?”

“For fourteen minutes!”

“Among agents.”

“Exactly!”

Weird. But still okay. The pots were going somewhere to someone who was not an ant. If she could figure out who that was, she could still build the business for them. Get the business humming again (or, apparently, for the first time.) Marcie clicked on another order at random. She needed to know more about her customers. They were agents, all of them. She was confident in that now. On the order page, the bot ID had a link to the US PEER Agent Directory. She clicked on the little American flag beside the bot name: rainbow-water-frog-feet.

The page took a while to load. She clicked.

Registered location: The Villages, FL 32159
Registered licensee: Scott Marten

She clicked another. The same thing:

Registered location: The Villages, FL 32162
Registered licensee: Scott Marten

The Villages. Florida. Scott Marten. Again. Again. Again.

Scott. This all led back to Scott.

Brrrz.

“Stop it, Navi!” She slammed the laptop lid down and flung it at the wall.

The story continues in…

Below the API is a short story published as part of the 2024 Summer of Protocols “Protocol Incepting Lore & Literacy” program. You can find an abstract for the project, as well as an index to each chapter on the project’s landing page.

The front door slapped into place behind Marcie as she set down the stack of boxes on the delivery pad. The shipments were discreet, just cardboard boxes with QR codes Marcie had slapped on. Gone were the days of public addresses and logos on boxes. Marcie wondered who was on the other end of that code. What would the house look like, and where would the pot be displayed? She’d look into it when she had time, she promised herself. Some other time.

A twinge of pain ripped through her back as she stood up, grimacing. She was doing more than she should. At the same time, those ached muscles were also getting toned. It was progress.

She looked down the street, hoping to see the delivery trawler. It would be a treat to see her first orders go out—maybe Navi could keep track of it.

Instead, she saw only the line of houses. Boxes, big people boxes. Set there, unlabeled, as if by some giant artisan who had sat at his workbench, assembling them, one after another, then plopping them down in a line. To make so many boxes, so many of the same thing, would make such a master’s mind numb. It would forget itself and turn intention into protocol—foundation, frame, finish, sell; foundation, frame, finish, sell; foundation, frame, finish, sell…

Marcie returned to the studio for the last load. Black dust from sanding clouded the studio, filling Marcie’s lungs as she walked in. Marcie hadn’t had time to clean this week. The wheel was flaked with residue, and the floor was gritty and littered with shipping materials. On the wall, Marcie’s “originals,” as she had started to call them, were beginning to get coated with dust.

Even so, she admired them. They were not outright better, but they were more thorough and careful than her recent work. Gone were the individual flourishes and design touches. Black clay, black clay, that’s what the people cared about. She had replaced the cuneiform and glyphs with her own: slash, slash, left, right. It was a personal script lacking history or meaning, a signature born of the need to produce, not the heart. It looked great glazed on black, though.

“Hey! Way to go!” Navi chimed, blots swirling. “Your first shipment!”

“Almost. One more trip.” Marcie stacked the last boxes. “Then I’m taking a break.”

“Great idea! You’re five minutes ahead of schedule!”

“Only five?”

“You’re doing great! We’re 73% of the way to our quota today. We’re five minutes ahead of finishing by 9 PM.”

Marcie grabbed the boxes and kicked the door closed. (“Lift with your back!” Navi called.) Walking past the back patio, she saw Scott lounging on a folding chair, smoking a cigar. Interesting. Scott outside was not a common sight. Something was up.

When she had dropped off the boxes, she grabbed a glass of water, threw back a couple of aspirin, and lathered her hands. She decided to give her creditor an update. A little gloating was in order as well. She pulled back the sliding door.

“First shipment’s going out now,” Marcie said casually. “Should be getting the first payments in a couple of days.”

“That’s nice.”

“Rate I’m going, I’ll easily have the money next week.”

Scott took a drag, turned to his sister, and exhaled. “I don’t need it, you know.”

Marcie dodged the smoke. “You don’t need it, as in, ‘Don’t repay me?’”

“No, I want the money. It’s pennies, though. That’s what I meant.” He took another drag.

“Five thousand dollars is not pennies to you. You’re in a ten-dollar lawn chair smoking a twelve-dollar cigar.”

“My surroundings do not reflect my standing.”

“They certainly reek of it.”

“So you’ve got some people—if they are people— buying your pots.” He leaned back in his chair and gazed at the feeble patch of grass they called a lawn. “Great for you, keep the garage, whatever. But run along. I’m busy.”

“You aren’t doing anything.”

“I’m watching my moves.”

“Sounds tough.”

“Markets are wild today if you need to know. I’m up 15% already… and boom. Still going up.” Scott stood up. He was shorter than Marcie, although his girth gave him plenty of presence. “I’ve cracked it wide open, sis. I’ll let you in on the secret, too. Not like you’re a threat or anything.”

Brrrz. It was Navi: “MC! Five minutes are wrapping up!”

Marcie decided to ignore it. “If you want to talk, talk.”

“So—take stocks. Stocks are just one big game of chicken. It’s just a gamble against stupid people: you’ve got to sell when the price is higher than you bought. It doesn’t matter if anything is happening underneath as long as you do that one thing. Are you following?”

“I’m not an idiot, Scott.”

“Well, prices increase when there’s a sudden change in demand. The more sudden, the better. If you can track changes in demand, then you know what’s going to increase the price and what to buy. The problem is that people are morons. They barely know what they want.”

Marcie was getting impatient. “And?”

“It’s not the humans who are demanding things now, Marcie. They’re too lazy. They used to read magazines and all that crap to come up with their ideas. Now, they leave it to their ants. The ants are doing the demanding.”

“So you ask the ants what to buy?”

“No, you don’t ’ask the ants.’ That’s illegal, and they’re morons anyway. They wouldn’t be able to tell you explicitly what they want anyway.” A drag. “But, ants talk. Talk, talk, talk, that’s all they do. So I listen. That’s the alpha.”

“Sitting in a lawn chair, listening to bots talk to each other,” Marcie said. “Sounds like a great business strategy.”

“Ape brain. Didn’t think you’d get it.” Scott waved her off and turned back to the yard. ”Doesn’t matter.”

Brrrz. Navi again: “Hey! We’re five minutes behind now!”

“Looks like the boss is calling,” Scott said.

“Assistant,” Marcie corrected. “Maybe with all that alpha you’ve got, you can afford a shower this week.”

“You’re one to talk.”

Marcie flipped him off as she slid open the porch door shut and shuffled back to the garage.

The story continues in…

Below the API is a short story published as part of the 2024 Summer of Protocols “Protocol Incepting Lore & Literacy” program. You can find an abstract for the project, as well as an index to each chapter on the project’s landing page.

THURSDAY

The green asphalt radiated heat, a kiln baking Marcie as she shook hands with her opponent. Her arms dripped sweat, but her hands were parched, black and cracked from the week’s toils. She returned to the bleachers outside the pickleball courts, defeated but not disappointed. Losing was fine, and so was the heat; she was relieved to be outside.

Not that getting there had been easy. Her ”passion,” as Navi referred to it, was ever-present. Her heart spiked with each order notification. Navi kept her informed on quotas and throughput metrics, even quality standards. Growing stacks of raw black clay loomed over her as she worked as if waiting in line. (Navi had begun buying in bulk as orders came in—market prices surged.) This was her break, and she was determined not to let it slip. But she also needed a break.

Marcie sat on the scorching metal bleachers, a hand towel providing a thin layer of protection from the heat. She scanned the courts. They were alive with pops and cheers, set against the vast Kansas fields behind. These were her people, enjoying the outdoors, craving connection to people, place, and past. So unlike those who spent their lives plugged in, curled in a fetal position, umbilical cord to the wide web, feeding on content.

I bet the Sumerians would have loved pickleball, Marcie thought.

“Ay’ Marcie! You win one yet?”

It was Barry, walking towards her, shouting from three courts down. She blushed and waved. He lifted his shirt to wipe the sweat from his face, and his gut waved back.

“Still learning, I’m afraid,” she replied when he reached her. “How about you?”

“Barely a point.” Barry swung his racquet bag down and took a swig of his water. His shirt was drenched as if he’d poured a water bottle on himself. “The old ones are sharks. They toy with me, back and forth, back and forth. I ran a marathon out there. Lester could have sat in a lawn chair.”

“Don’t feel bad. He’s out here every day.”

“Oh, don’t worry. My pride is fine—it’s just my ankles, elbow, knees, and back.”

“Maybe that’s their secret, then.” Marcie laughed as he threw his towel down and sat next to her. “As your body breaks down, you master the art of playing without moving. Like Zen Buddhism or something.”

“You’re onto something,” Scott replied. “How about you? Pardon my saying, but you look a little beat up yourself. You doing shifts as a mechanic now?”

“No, just a new project.” Marcie laughed and flexed her hands. She had tried to clean up, but the black oils had seeped deeper than a quick scrub would clear. “It’s a huge opportunity. But my hands aren’t happy about it.”

“Your hair, either.” Barry motioned at Marcie’s brown curls, and she ran her hand through them. They were gritty from clay dust. “Not that it makes a difference, I mean. You look great.”

“I came straight from work,” Marcie said, sliding over a bit. “It should level off soon.”

“Sorry, didn’t mean—” Barry smiled sheepishly. “What’s the big opportunity?”

“I’m throwing new pots, but not with my normal bases. I’m using black clay. It’s a pain to work with—stains like mad and makes the studio filthy. Beautiful when it’s done, though.”

“Black clay is, like, everything right now!”

“You’re familiar?”

“I’m huge into it. Been all over my feeds this week. You’ve seen the research, right?”

“No, I hadn’t heard anything about it before.”

“Oh man, black clay is amazing. Full of anticoagulants and photo-absorbers—lots of nutritional value if you make it right. It comes from fossilized fruit, you know. The authentic stuff, at least.”

“I hadn’t heard that.”

“I ordered some powder you can put in coffee. Boy, it gives me a great buzz. I read that some people snort it.”

“Woah, I haven’t —”

“Wait! Idea alert!” He set his hand on Marcie’s arm. “Are your pots for, like, sale?”

“Yeah, but there’s a backlog…”

“Sweet! What do I do?”

Normally, Marcie would have been thrilled by this smile; she had dreamed of having it pointed at her just like this. But not like this—not tied to another pot to add to her queue.

“Well, Navi manages the listings. I haven’t actually ordered one myself.” She hesitated. “Why don’t I just make you one?”

“Oh, no, wouldn’t dream of it. I’m an admirer, not a moocher. Is Navi your agent?” When Marcie nodded, Barry turned aside and double-tapped his tortoise-shell glasses. “Franklin, buy a pot from Marcie’s shop. A black clay one. Make it sound cool.” He turned back to her and smiled. “Easy. Can’t believe the timing!”

Brrrz. Marcie checked the order: Ancient Replica Pot (Black Clay, Persian style, Text: ‘Barry’).

“You didn’t need to do that,” Marcie said, forcing a smile. Inside, she felt elation and irritation. Another pot. Another firing. Another glazing. Another box. “But thanks.”

“Stoked for it!” Barry surveyed the waiting area and then grabbed his racquet. “Think Tracey’s ready to do combat. But hey, Marcie, I was also wondering. Since we both need some extra practice, would you want to play tomorrow—just us?”

She glanced at her watch, at Barry’s order details. This had been refreshing, but taking another night off tomorrow would be tight. She had to get sixty fulfilled by Monday, and while progress was starting to pick up now, she was finishing the first batch today. She tried to do the math, to compare it against her quotas, but it made her dizzy. It was so hot, so bright outside, so much like the kiln. Her mind was baking. She wobbled in her seat, and Barry touched her shoulder to steady her.

“Yo, no pressure! No pressure! I don’t need to beat him next week. We’ve got plenty of time, hotshot.”

“Sorry, Barry. It’s been a busy week. I’m finding my new rhythm.” She squinted at him through the sun. “Next week, for sure.”

The story continues in…

Below the API is a short story published as part of the 2024 Summer of Protocols “Protocol Incepting Lore & Literacy” program. You can find an abstract for the project, as well as an index to each chapter on the project’s landing page.

WEDNESDAY

Marcie peered around the corner to see if Scott was available. Of course, he was. It was 2 PM, and he was eating cereal, headset plastered on and ragged beard catching the occasional splash of milk. Occasionally, he waved his arm, manipulating some object only he could see. Marcie steeled herself for a distasteful conversation.

Two hundred more orders had rolled in over the last day. Navi’s notifications had been like a drumbeat, always coming right when Marcie started to get tired or her mind wandered. She’d emailed in sick to work and might have to continue if this kept up—she had over twenty thousand dollars pending in orders. Marcie’s hands were stained black from the new clay, and her arms ached from a full day of throwing and shaping.

There were problems with success, though. She hadn’t been paid yet and needed cash to buy more clay. Navi assured her it was a one-time problem—the cost of “bootstrapping” the business—very common. Still, she needed cash, and only one person in the house had it.

She stepped out into the dining space and was about to ask what he was up to. Instead, she got a closer look at his cereal bowl and found herself saying, “Scott, you ass!”

Scott set his spoon down and looked up at her, headset glass shifting from opaque to transparent. “What’s your problem?”

“This is not a cereal bowl.” She grabbed the bowl with both hands and tried to drag it towards her, but Scott caught the other side. “This is art.”

“Screw off. It’s a pot.”

“It’s art,” she said but relented. Not the battle she needed to fight. She let go of the pot, and Scott pulled it back towards him. “But yes, it’s useful, too.”

“Nice analysis, Sherlock.”

Marcie’s cheeks flushed. “Look, I know you’re busy right now, so I’ll be quick.” Scott took another bite of cereal. “I need a loan.”

Scott put down his spoon and considered her. “What for?”

“My business,” Marcie said. “I’m taking it seriously again. I’ve had hundreds of orders. I was throwing new pots until eleven yesterday. Didn’t you hear?”

“Of course, I did. I’m enjoying it now. It’s like the music during the closing credits. I want to remember what it’s like, so I enjoy the silence more when it’s gone.”

“I’m not going to be gone. I’m onto something big.”

“Yet you need a loan.”

Marcie took a breath, calming herself. “I don’t get paid until the order is received, and it takes some time to throw the pot and ship it. I should start getting the first payments on Saturday. I can pay you back.”

“Oh, you’ll pay me back? That makes it interesting.” Scott pushed the bowl away from him. “Ok, sis, I’m a good guy. I’m listening. But due diligence first. Who on god’s earth is buying from you?”

The question caught Marcie off guard. “I mean, I don’t know who they are. It’s not like they’re coming to the house. They’re mostly looking for house decor — black clay stuff. Look, here on my watch, here’s a new order: ancient pot, black clay, text engraving of ‘Bless this mess.’ Another one hundred dollars, just like that.”

Scott snorted. “‘Bless this mess’?”

“Yeah, the messages are pretty cringe.”

“Who ordered it?”

“I don’t know. Navi can’t share personal information —”

“Who’s Navi?”

“Oh, that’s my agent.” Marcie had forgotten how recently she had started working with Navi. She already felt like an extension of herself. “You know, on the wall in there.”

“Ah, that’s it then,” Scott smirked in a way Marcie didn’t care for. “You’re working with an ant. You’re selling through the PEERNet, is that it?”

“What’s wrong with that?”

“Nothing’s wrong with it. Ants are dumb, but they still beat most consumers. If anything, PEERNet is more efficient. But I can’t believe that people would buy this… this….” He held up the bowl.

“Art.”

“Right. That’s the word.”

“Look, Scott, whether you believe it or not, there are still people out there who appreciate history and like having a connection to our past. They appreciate creators and quality.” Marcie was ready to close this out. “Look, I’ve got a dozen more pots to throw tonight. I’ll have the flywheel spinning in two weeks, and you’ll have the money back in a month. Will you lend me the money or not?”

A look of challenge passed between the siblings.

“Ok. I’ll lend you the money. But I want seven thousand in return.”

“How is that reasonable?”

“You’ll have the ‘flywheel’ going, won’t you?” He rolled it around on his tongue. “Flywheel. Bet that came from your ant.”

“Fifty-five hundred.”

“Six thousand. And I’ll give you one week.”

“Two weeks. If I don’t make it, I’ll start paying rent. Or I’ll move out.”

“Fine.” Scott leaned back. “Look at you. A real businesswoman, now, huh?”

“Six thousand, then. In a week.” Marcie stood up. “How will you send it to me?”

“I’ll tell Dante. Let the ants can figure it out.” He waved his right hand. “But.”

“But what?”

“But I’ve got better ways to make that money work than throwing it away on pots. The market’s been on a tear this week. So you better be on time.”

“Fine, that’s fine. ” Marcie said. “Monday is plenty of time.”

“Good,” Scott’s visor clicked closed as he scraped for another bite of his cereal. “Oh, and sis—this cereal bowl’s a little lopsided. Can you throw me a new one while you’re at it? Thanks, sis.”

The story continues in…