Reeks and Wrecks does not have to be the future
Kurt Vonnegut’s 1952 novel, Player Piano, feels terrifyingly prescient: a dystopian society where machines and computers have replaced most human workers, leaving only engineers and managers with meaningful roles and pay while everyone else lives on basic welfare, “working” at the Reconstruction and Reclamation Corps ("Reeks and Wrecks").
Those who failed aptitude tests end up at Reeks and Wrecks to perform make-work, deliberately inefficient manual labor designed to give the economically displaced masses something to do. Their jobs include tasks like road repair, construction projects, and land reclamation, but these activities are intentionally done without modern machinery to create employment rather than accomplish necessary work efficiently. They know their work is unnecessary busywork, which strips them of dignity and genuine purpose.
Vonnegut's critique of a society that can't imagine any solution to technological unemployment beyond creating artificial inefficiency resonates. But we have agency to make it less prescient. The dystopia emerges not from machines themselves but from failing to reimagine social structures for a new world.
1️⃣ Do not compete with the machines. It sounds obvious but it bears repeating: they are coming, so let’s focus on human-machine collaboration.
2️⃣ The land of the uniquely human is not shrinking as much as it is revealed. Humans value human nature, even if we can’t define it with the same rigor as we can prove a theorem. Let’s build around the uniquely human. As an example, player pianos can only replicate an existing performance in great detail but not deliver on an unseen piece. Efficiency is not all.
3️⃣ Preserve human agency. We need a say in our fate, need democratic participation in how AI gets deployed, what gets automated, and what remains human-centered by choice. This includes protecting certain domains for human involvement not because machines can't do them, but because we value human participation.
4️⃣ Invest massively in education and adaptation. Not just narrow technical training but cultivation of uniquely human capacities like creativity, ethical reasoning, interpersonal connection, and critical thinking. Education should prepare people for multiple transitions throughout their lives, not single careers.
https://link.springer.com/article/10.1186/s13321-025-01099-w
Nice explainer: link
Beyond performance: how design choices shape chemical language models
High variance kills traditional SaaS unit economics
That’s a section of the super clear and insightful article by @anjali Shrivastava, “A broken pricing paradigm”. It was published more than 2 years ago and yet it is more relevant than ever in the “agentic era”. We have witnessed rate limits and skyrocketing costs at the big model labs (Anthropic, OpenAI) because a token is neither an atomic unit of cost nor an atomic unit of compute. “Traditional SaaS pricing mirrors the physics of stable software,” but AI introduces high or even infinite variance to the pricing equation: if task requirements AND number of tasks per day are high variance (fat tails: a small number of tasks and/or users dominate the costs) economic theories fall apart with possibly unlimited risk.
The shock to the SaaS model requires a complete change in the pricing mindset. It is not like SaaS companies that have been thriving can now compete without AI. Gone is the zero marginal cost. Plus, the cost distributions per user are power laws with long, long tails.
Now, layer on top of that the new agentic paradigm, even in its simplest form, one agent with multiple tool use. The agent must discover the tools it needs, find out where the data it needs might be and in what form, explore an undefined number of dead-ends and spend an unpredictable number of tokens. Imagine what happens if it must interact with an unpredictable number of other agents that will also consume an unknown number of tokens and possibly interact with more agents, and turtles all the way down. The product of possibly many, possibly infinite variance distributions does not bode well for pricing.
AI agents, for all of their promises, have a number of issues to address (security, privacy, quality assurance, stability, determinism, ...) but perhaps the biggest of all is the unpredictability of resource use. Some early adopters discovered that letting agents loose could lead to outrageous invoices. The current solution is to have limits on how much an agent can use, but predicting resource utilization is a necessity.