Finding a technical co-founder for a new startup is notoriously hard. After struggling with it for a while, I decided to figure out exactly how hard it is.
To do that I simply estimated how many needles there are in the co-founder haystack. The answer is discouraging but intuitively accurate: there are only about two thousand potential co-founders in the entire United States.
How hard? Let’s use Fermi Estimation
In Fermi estimation, we wave our hands and spin minimal knowledge into approximate truth. The goal is to quicky get an estimate that’s close enough to correct for our purposes. For a link-bait blog post, “close enough” is not very close at all. And that’s close enough®.
We do Fermi estimation by breaking down our final quantity (number of viable co-founders) into a chain of constituent quantities. I model the problem like this:
Viable co-founders =
(% of programmers who are 2 standard deviations above the mean)
* (total number of programmers in the world)
* (% of world population living in the USA)
* (% of Americans in a startup approriate age-range)
* (% of programmers who know a given technology)
Now we basically just guess each of those quantities (with a Google sense check if possible)
% of programmers who are 2 standard deviations above the mean = 2.5% (by math)
total number of programmers = 25 million (number of StackOverflow accounts)
% of world programmers in the USA = 5% (US pop as % of world pop)
% of Americans in a startup approriate age-range = 20% 
% of programmers who know a given technology = 30% (guess)
Now run the numbers:
Okay, so that all adds up to:
2.5% * 25,000,000 * 5% * 20% * 30% = drum roll 1875
So that’s it! Out of a US population of 319 million, a mere ~2000 people are viable co-founders. That’s just 1 in 1.5 million people.
Dang that’s hard.
Then factor in the fact that the vast majority of great programmers are already employed (or 3 years into starting their own succesful company), and that only a small % will have your same interests and a compatible personality. It becomes easy to see why so many startups deliver bad technology or are torn apart by co-founder conflicts.
P.S. If you’re looking for an opportunity and love music, metadata, and HCIR, shoot me an email at george j london on gmail. Or if you just like Fermi estimations, follow me on Twitter