You’ve just moved through thirteen ideas that were always connected.
A hammer disappearing. A breakdown that teaches. An assumption buried so deep it felt like reality. A biologist watching frogs and discovering that perception is production. A background that has no bottom. An interpretation that was always already there. Language that doesn’t carry meaning but evokes it. Speech acts that create social reality. Commitments that constitute identity. Design that starts with care. A formalization trap with two jaws. Code that appears without the knowing that would have produced it. And a question about what knowing actually requires that the field building the most powerful tools in human history is not yet asking clearly.
These aren’t thirteen separate ideas. They’re one idea, seen from thirteen angles.
The idea is this, what makes humans irreplaceable is not what we can produce. It’s what we’ve become through the process of producing it. The background. The structural history. The accumulated breakdown that deposited knowing in a system that was present for all of it and was changed by it.
That’s the one idea. Everything else was approaching roads.
Flores and Winograd wrote this book in 1986 as a foundation for design. They wanted to change how people built tools for humans. They mostly didn’t succeed at that, not because the argument was wrong but because the commercial and intellectual pressure was entirely in the other direction and the rationalistic tradition was producing results impressive enough to defer the reckoning.
The reckoning is no longer deferrable.
We are designing systems that are being embedded into the most consequential domains of human life, medicine, law, education, governance, creative work, software infrastructure, and we are designing them from the wrong model of mind. From the model that was available, that was culturally dominant, that produced the impressive results that got us here.
Flores would say the breakdown has arrived. Not as catastrophe yet, as the accumulated friction of systems that work impressively in the center and fail invisibly at the edges. As the knowing gap opening in the next generation of practitioners. As the commitment-shaped interactions that feel bilateral and aren’t. As the background being colonized quietly by fluent not-knowing.
Breakdown is the signal that the current framework is inadequate. The question is whether we have a better framework ready.
You do. It’s thirty-eight years old. It was written by a computer scientist and a Chilean philosopher who had been a political prisoner, who had seen what happens when systems built on false models of human nature are imposed on actual humans, who understood from the inside that the gap between the model and the living reality is not a technical problem waiting for a better algorithm.
It’s a design problem. A philosophical problem. A problem about what we believe humans are and what we’re willing to build from that belief.
So what do you do with this?
The first thing is the hardest and the most important, hold the knowing gap as real. Not as a concern about AI safety in the abstract. As a specific, structural, Flores-diagnosable condition that is opening right now in domains you inhabit. In your organization. In your practice. In the next generation of people learning in the fields you care about. Name it precisely when you see it. The formalization trap. The second jaw. The background being colonized. Precise language is the first form of resistance.
The second thing is to protect your own breakdown. Deliberately. Structurally. Not as nostalgia for harder ways of doing things but as the active maintenance of the mechanism that builds knowing. Stay in contact with productive friction. Debug without assistance sometimes. Build things from scratch sometimes. Let things fail on you in contexts where you can learn from the failure. The background cannot be outsourced. Protect the process that builds it.
The third thing is to make complete speech acts. In your work, in your organization, in your use of AI tools. Make real requests with stated conditions of satisfaction. Make explicit promises rather than assumed ones. Surface the assessments that are currently private and generating invisible friction. The network of conversations is the organization. The quality of your speech acts is the quality of your thinking made visible and accountable.
And the fourth thing, which is why this series exists, is to share the framework. Not the conclusions. The framework. Because conclusions can be argued with or dismissed. A framework changes what you can see. And what you can see determines what you can design, what you can build, what you can protect, what you can demand from the systems being built in your name for your world.
Flores and Winograd gave us the framework. Thirty-eight years ago. Before the internet. Before mobile computing. Before machine learning. Before any of this was visible on the horizon.
They saw it anyway. Because they were asking the right question. Not what can computers do. Not how intelligent can we make them. But, what are humans, actually? And what does that mean for what we build?
That question is still the right question. It may be the most important question of the next decade.
You’ve been inside it for fourteen videos.
Now take it somewhere.
@ Lee Greene