I tweeted a summary thread of this essay here.

There’s a somewhat dated meme that goes something like this:

  • Person 1 (pleading): All the jobs say I need years of experience to apply. But I don’t have any, so I can’t apply!
  • Person 2 (annoying): Then go get a job and get some experience!
  • Person 1 (exasperated): But I can’t get a job without experience!

This is an instance of a general phenomenon called the cold start problem—how to initialize a process that requires a product of its own operation to start.1

If you’ve ever tried to teach yourself a new subject without any outside instruction, you’ve likely encountered the autodidact’s cold start problem—how to effectively learn a new subject when you don’t know what’s important within it, or even what all is within it. Of course, if you knew the subject very well, you’d know exactly where to start learning and how to proceed.2 But you don’t know anything!

There are meta learning skills that you can cultivate to help you get through the autodidact’s cold start problem (searching YouTube, checking Five Books, trying 100 things, brute force exploring until you have a wide enough context to proceed, etc), but not all subjects reveal themselves equally well. One particularly resistant subject is New York City government and politics.

The NYC Politics Cold Start Problem

Not only is there an extreme lack of resources that provide a practicable introduction to NYC politics, but the broader field itself is plagued by a negativity bias and an abstraction bias that kneecap otherwise capable autodidacts. The result is that an extremely narrow slice of the population ever meaningfully interacts with NYC’s political system, or really even understands what it can do.

This is, of course, not ideal. Once a cold start problem becomes pervasive enough, it is indistinguishable from a civilization that’s forgotten how to build the advanced technology of its forebears.

  • Person 1 (politically frustrated): I want to understand NYC politics, but I don’t know how to learn, or which meetings to go to!
  • Person 2 (cursed by knowledge): Just go to your community board meetings and listen to them.
  • Person 1 (desperate): But I wouldn’t understand what they’re talking about, or why that’s even important!

The politics cold start problem is the reason that I began to teach The Foundations of New York; I’d encountered plenty of kind, smart, ambitious people who wanted to be politically active, but simply had zero context from which to begin. They didn’t know what they didn’t know, and there were no good books or classes available to dislodge them from that state.

In an epistemological environment dominated by the cold start problem, like government and politics, the most vital task is developing a proper pedagogy—a way for one person to successfully teach others, who can then go on to teach others. In other words: the best way to improve the field is successfully creating an onramp into it.3

I’ll end by listing some common mistakes/errors/missteps that I observe in NYC politics, and the political field more broadly, that stem from the politics cold start problem (but are also present in other domains). Many of these are concomitant:

  1. People want to skip ahead to advanced knowledge before they are prepared to do so, usually because the appearance of possessing that advanced knowledge provides status. Everyone wants to be on a political team, but nobody wants to understand how the law works or is actually made. The inevitable result is a bunch of people who only speak in incredibly abstract terms, discussing what amounts to a fake, made-up system that doesn’t correspond to political reality.
  2. People skip city politics in favor of federal politics. For a variety of reasons, prestige among them, people assume that the federal government is the most effective thing to focus on. However, as anyone who has a moderate grip on NYC politics would say, that is nonsense. But you can only know it’s nonsense if you understand what the NYC government can do, versus the state and federal governments.
  3. Related to the first two points: people treat politics as “all or nothing,” which can also be rendered as “revolution or nihilism.” This attitude lasts if you have such a low resolution image of the field that you only have one black dot and one white dot—and many people prefer that resolution! An unfortunate consequence of holding it in the American context is that revolutions rarely come, and nihilism is its own hell, so you’re condemned to having a terrible time and accomplishing little. You need to step into high resolution, living color—in between revolution and nihilism is the technical field of governance. And in that field, only the technically sophisticated have proper access to optimism.
  4. People focus solely on electoral politics (running campaigns), because that is the part of the political process that is often the most visible. Lacking a broad survey of the field, they never get a grip on politics outside of elections…which are the governing and lawmaking parts of it.
  5. Scoping problems: people try to learn too much, too fast. But politics, like physics, is a deep field with technical sophistication. You need to give yourself time to learn well.
  6. People try to learn it only through reading/podcasts/consuming. But politics, just like a physical craft, needs to be done physically. It involves learning by reading, but must necessarily include learning by interacting with politicians and the political system directly in a variety of ways.4
  7. As a result of the penultimate point, their learning process doesn’t really have any consequential feedback loops. You can read about politics and government all you want, but unless you can find a small feedback loop early to demonstrate to yourself that your learning process works, it’ll feel—and be—hollow. This makes it hard to continue.


  1. This phenomenon manifests in every domain sooner or later. Andrew Chen even wrote a book entitled The Cold Start Problem: How to Start and Scale Network Effects (2021) to describe how network-based products scale, given that they need a good network to get started…but getting a network is hard if no one uses your product to start, etc. 

  2. Assuming you didn’t fall victim to the curse of knowledge

  3. Considered another way: if civilization breaks, you need to have a book that teaches you how to rebuild it from its basic component parts, which most of us do not understand. This is an interesting and growing genre of resource.

    Considered another way: a cold-start system has no energy to start itself with; so you must introduce energy into the system from an external source. A teacher to a pupil, a master to an apprentice, a spark to a fire, a ???? to a highly compressed and inert universe. 

  4. Otherwise you get this, from For Political Founders, the Bar of Minimum Viable Knowledge is High: “This produces a large group of otherwise intelligent people who assume they have an advanced grasp of politics because they bullshit about it with their friends, or because they keep track of the “blue team versus red team” play-by-plays that fly around the news and social media. Perhaps they have read a lot of political philosophy in their spare time, or picked up knowledge about certain legal mechanisms—this isn’t nothing, but it’s not as helpful as you’d think. It usually just results in a crank who can’t translate the eighteenth century treatises they’ve read into operational knowledge for the modern day.”