How to destroy a world
Refute the narrative! Attack the leaders! Dilute LinkedIn feed quality! Er...
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Sports World, amplifying coaching and player exploitation scandals; for Harry Potter World, exposing its author as outdated and regressive.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