The density dependence curve and startup location
Update on 2023-03-06: While I won't edit the below, I do have an important update on it. The world is just a colder place for startups than I had realized; cities are generally further to the 'left' on the curve I describe below than I had previously assumed. The below stands, but I now think that its implications -right now- are not what I believed two months ago. The most popular options have a good chance of being the best ones, if they are still located to the left of the inflection point on this curve.
Decades of research suggest that the survival chances of organizations partly depend on the local density of similar organizations. The chances that a startup survives depend in part on how many other startups there are in its vicinity. Yes, I haven't defined all those terms, and you can throw 'density dependence organizations' into google scholar any time you like in order to retrieve them ;)
Now, if you're interested in startups, you've probably heard of the idea of hubs or clusters before. Among startup folk, the notion that hubs confer a survival improvement is likely to seem obvious. What you probably didn't know is that this effect doesn't increase linearly as you add startups to a cluster.
Instead, it actually goes into reverse. When there are relatively few other startups in your 'niche', adding startups improves your chances. When there are relatively many, each extra startup instead hurts your startup's chances. Now, estimating where exactly a given city or region is on this curve is hard, and it can probably only be done conclusively in retrospect. But there are signs, because the drawbacks of having too many similar companies are observable: overheated job markets and office markets, and the local prevalence of chancers and snake-oil salesmen, to name just a few.
A long time ago, I was interested in this and some related topics, ones that still had some open questions left in them to research (this one is pretty well studied). It's always occurred to me that there is a location strategy element: you should actively avoid locating your startup in an 'overcrowded' city, but it's also hard to be one of the few beleaguered souls in some far-flung outpost. What you really want is to locate in a city that's in the early stages, and on the way up; a city where there are plenty of startups that might overlap with you (knowledge-sharing, camaraderie, collectively attracting specialized talent and capital), but not so many that the drawbacks have materialized too badly (overheated housing and job markets, to start with).
Side-note: locating somewhere if you're not sufficiently connected to all these other startups can certainly expose you to the drawbacks -- but let's not forget that it doesn't automatically garner you any benefits.
To illustrate, imagine that you show up, as young people from around the world sometimes do, at the Caltrain station in Palo Alto.
If you do this for the first time, you'll probably marvel at the way the Silicon Valley 'cradle of innovation' is served by a train line that seems to belong in the first half of the twentieth century. My vague understanding is that local politics prevented the region's BART (somewhat more modern) metro system from extending that far south. Better to rent a car, if you can.
Back to your imaginary visit: there's really no guarantee that you'll learn anything of interest or value to your budding startup in the course of a wander around, or (without prior arrangement) that you'll even come across anyone who could be of use to you. You need some kind of local network access too. That's easier in some geographic areas than others - to be fair to Silicon Valley, it's probably more permeable than most in this respect.
So, to recap: it's clear that to benefit from density dependence, I should pick a city where I have at least some hope of making useful connections in my startup's field. But rather than defaulting to the most popular options, I should try to consider where is both early enough in the density dependence curve, and moving positively along it (so I keep benefiting, at least for a while, from the growth in similar local startups).