An introduction to some terms and influences. Primitives for higher education and beyond.

Institutions earn and sustain the legitimacy to do hard things by first doing easy things visibly well. In practice, that's the opposite of how many institutions behave--particularly public-serving institutions--and the gap between the prior statement and the practice is where a lot of public trust goes to die.

A dominant theory inside most large organizations, public and private, is that legitimacy is built through process. Stakeholders are consulted. Frameworks are adopted. Steering committees are convened. Roadmaps are published. The implicit promise is that if the process is rigorous enough, the eventual deliverable will land on a foundation of broad assent, and the institution will have earned the standing to act.

In a 2019 Michigan Law Review essay called The Procedure Fetish, Nick Bagley argued that the federal administrative state has been operating on this theory for 50 years, and it is no longer working. Procedures have multiplied; trust in agencies hasn't followed. Alon Levy makes a parallel case about urbanism in Streets Before Trust: years of community meetings about bike lanes don't produce communities that trust the transportation agency. What produces trust is a bike lane that exists, not one that was promised years ago and never delivered.

Simply put: for most people, legitimacy is downstream of outcomes. You earn the right to act in the future by delivering something that works today, and then you get to do the next thing.

This approach has something to do with how complex systems can more reliably improve, I think.

Economists have a term for this type of work: marginalism. (Hat tip to Tyler Cowen's excellent blog, Marginal Revolution, which has been one of my gateways to this type of thinking.) The discipline is to find the next best move at the margin and make it. You can't fix a whole system in one pass; you can't see the best end state from where you stand, and pretending otherwise produces plans that are confidently wrong. Engineers have a similar discipline: solve the problem in front of you; don't generalize prematurely. Biologists have the most general version of this, borrowed from Stuart Kauffman: every system has an adjacent possible, the set of moves available from its current state, and the only way to reach a more distant state is to traverse the adjacent possible one move at a time. You can't skip.

There's a reasonable case to be made that Charles Babbage was further beyond the adjacent possible than almost anyone in history. In the 1830s he designed the Analytical Engine, a fully programmable mechanical computer with memory, a processor, conditional branching, and punched-card input, more than a century before the first working general-purpose computer. Modern analysis confirms that Babbage's design was sound (which is why it's a useful example in a way that speculative science fiction, or general futurism, isn't). The machine would have worked. It just couldn't be built. The precision metalworking, the materials, the funding, and the manufacturing tolerances required to assemble tens of thousands of gears into a reliable apparatus didn't exist in Victorian England, and wouldn't exist for generations. Babbage had leapt past the adjacent possible. His design was correct and useless at the same time.

The iPhone, which often gets framed as a transformational innovation, is the opposite story. Almost nothing in the original iPhone was new. Capacitive touchscreens existed. Mobile phones existed. MP3 players existed. Mobile internet existed. Cameras existed. Apple assembled a set of components, each of which had matured to the point of being ready, into a configuration none of them had previously inhabited. The iPhone was an act of integration across a dozen mature technologies, every one of which sat squarely in its adjacent possible. That's why it worked, and why a company could actually build it, and why it shipped.

Apple produced a working machine; Babbage produced schematics. Both required ambition. Babbage's ambition was prescient. The problem was that his ambition outran the adjacent possible.

Steven Johnson's book, Where Good Ideas Come From, was my original introduction to the adjacent possible. It's where I learned the Babbage story, and it lays out a range of patterns that produce innovations like Babbage's or the iPhone. Ben Thompson's blog Stratechery opened my eyes to another: the creation of primitives. A primitive is the smallest thing that does real work and can be recombined with other things to do larger work. Cloud computing was built using primitives; arguably so was the iPhone.

It's also a fair way to understand what the organization at which I am executive director, Modern States, does, why it works, and why it sits at a meaningful angle to other reform and improvement efforts. The primitives lens is what the rest of this series is for, and I won't go deeply into it here. The argument here is narrower: the unit of institutional progress is the smallest concrete move that ships, works, and reshapes what becomes possible next. Quick wins are an important way that the long arc actually moves.

This is the under-appreciated half of the quick-wins argument. Solving the problem that exists now is how the long arc advances. Every concrete delivery does two things at once. It improves outcomes today, and it reshapes the set of moves available tomorrow. The institution that ships a working tool learns things about its users that the one writing a strategy memo never sees. The agency that fixes the small thing develops the relationships and the credibility that make the contested thing possible. The team that completes a project builds capacity that planning, however careful, doesn't produce. None of these gains was on the original roadmap, because none of them could have been. They exist only because something got done.

Three implications follow.

First, the comprehensiveness of a strategy is a weak signal of its quality. A plan that addresses every priority is usually a plan that hasn't been pressure-tested against any of them. Asking what can we deliver in a year and then doing it is how you find out whether the long plan is real.

Second, the right unit of analysis is the move. The destination is downstream of a sequence of moves, most of which you can't specify in advance, because each one reshapes what comes next. A strategy that tries to lock in the destination is doing the wrong work (whereas a strategy that expands options for the future is multiplying its value).

Third, the failure mode is real but solvable. A series of disconnected small projects that never compound is not any better than a coherent long plan that doesn't go much of anywhere. What makes quick wins compound is the discipline of putting each one on the path. The test for any specific quick win is whether shipping it expands the adjacent possible in the direction you actually want to go.

This is something that separates institutions that get to do consequential work from institutions that spend a decade preparing to. The consequential ones are delivering something this quarter. They understand the long arc as a thing you build one move at a time, each move reshaping what becomes possible. Babbage's machine was right. Apple's machine got built. The difference is the whole game.

Trust follows streets. So does everything else.

Best,

JP

I’m using this space to think through ideas publicly, hopefully with others. Feedback welcome; comments are available.

I’m also using AI extensively to be able to connect ideas and speedrun my way towards clarity. Errors and omissions are mine.

This piece sets the frame for a series on primitives, the adjacent possible, and the institutions that build them.

Sources & references

On legitimacy through outcomes

• Nick Bagley, The Procedure Fetish, Law and Political Economy Project blog (short-form summary of the longer Michigan Law Review essay).

• Alon Levy, Streets Before Trust, Pedestrian Observations, December 31, 2020. Applies the same argument to urban planning and transit.

On marginalism and the adjacent possible

• Steven Johnson, Where Good Ideas Come From: The Natural History of Innovation (Riverhead, 2010). Popularized Kauffman's concept and applied it to innovation, technology, and institutional change.

On Babbage and the Analytical Engine

• Analytical Engine, Wikipedia. Notes that the first general-purpose computer came more than a century after Babbage's 1837 design.

On primitives

• Ben Thompson, Stratechery. The contemporary articulation of how primitive-building shapes platform strategy. Specifically informs the primitives framing across this series.

On the broader abundance / proceduralism critique

• Ezra Klein and Derek Thompson, Abundance (Simon & Schuster, 2025). Frames procedural overgrowth as a central failure of contemporary American liberalism, a theme for the future.