The interesting thing about this essay is many people have read it and understand it but they want a language that has a big community that writes lots of Medium articles and beats the averages.
You can’t have it both ways. You can’t be safe in the herd and ahead of it.
Most startup founders aren’t looking to build an elite team of 5 that outperforms 100. They’re looking to grow rapidly and get to 100 within a year. This means looking really good (even if using misleading or inflated metrics) to your investors is what matters. Scaling a business rapidly is the objective. Technical excellence doesn’t matter until very late in the process. (Even though a small number of people can do a lot of damage, from a CEO’s perspective, this is something one can buy one’s way out of, later on. Mismanaged PR or investor relations can’t be.) Now, not only is technical excellence not necessary, but it’s risky in its own way. It takes time to find great people, great CTOs, and to build great products (noting that great engineers usually hate “get this done by X or you’re fired” deadlines and just refuse to work that way). Just as 9 women can’t have a baby in 1 month, technical excellence can’t be rushed although the rest of the startup process pretty much has to be, in order to keep the attention and love of short-attention-span investors.
Paul Graham wasn’t wrong about Lisp. (I might make the argument about Haskell, instead.) The game changed. Investors aren’t looking for technical excellence. They don’t care. They want unicorns. Unicorns don’t need great people; they need to scale up quickly with average people. Consequently, you want to use average languages.
That said, the average may be better in some ways. The average programmer seems to be worse, thanks to Agile Scrotum practices and the bottom 85% of the dev bootcamps, but the go-to average language is Python, which is an improvement over Java for the “what the average people use” language. (For anything where Python’s too slow, you’d go to Haskell or C before Java. Hence, I’m comfortable arguing that Python > Java.) Python I can actually stand to program in.
Part of this is the simple fact that most startups aren’t really tech, as you’re well aware. They’re just businesses that operate on the Internet. You can get away with average talent.
I’d argue the permissive nature of the Web and the general public’s acquiescence to broken software is also part of it. Things not working? Did you try reloading?
Man, I miss the days when Paul Graham wrote interesting articles about technology instead of selfishly monetizing his long-ago-earned reputation and sending his goons (to wit, Paul Buchheit and Dan Gackle) to wage petty feuds that shouldn’t even exist.
The interesting thing about this essay is many people have read it and understand it but they want a language that has a big community that writes lots of Medium articles and beats the averages.
You can’t have it both ways. You can’t be safe in the herd and ahead of it.
Most startup founders aren’t looking to build an elite team of 5 that outperforms 100. They’re looking to grow rapidly and get to 100 within a year. This means looking really good (even if using misleading or inflated metrics) to your investors is what matters. Scaling a business rapidly is the objective. Technical excellence doesn’t matter until very late in the process. (Even though a small number of people can do a lot of damage, from a CEO’s perspective, this is something one can buy one’s way out of, later on. Mismanaged PR or investor relations can’t be.) Now, not only is technical excellence not necessary, but it’s risky in its own way. It takes time to find great people, great CTOs, and to build great products (noting that great engineers usually hate “get this done by X or you’re fired” deadlines and just refuse to work that way). Just as 9 women can’t have a baby in 1 month, technical excellence can’t be rushed although the rest of the startup process pretty much has to be, in order to keep the attention and love of short-attention-span investors.
Paul Graham wasn’t wrong about Lisp. (I might make the argument about Haskell, instead.) The game changed. Investors aren’t looking for technical excellence. They don’t care. They want unicorns. Unicorns don’t need great people; they need to scale up quickly with average people. Consequently, you want to use average languages.
That said, the average may be better in some ways. The average programmer seems to be worse, thanks to Agile Scrotum practices and the bottom 85% of the dev bootcamps, but the go-to average language is Python, which is an improvement over Java for the “what the average people use” language. (For anything where Python’s too slow, you’d go to Haskell or C before Java. Hence, I’m comfortable arguing that Python > Java.) Python I can actually stand to program in.
Part of this is the simple fact that most startups aren’t really tech, as you’re well aware. They’re just businesses that operate on the Internet. You can get away with average talent.
I’d argue the permissive nature of the Web and the general public’s acquiescence to broken software is also part of it. Things not working? Did you try reloading?
Oh, good ‘ole blubby. So true, yet so hard to convince anyone of it.
Man, I miss the days when Paul Graham wrote interesting articles about technology instead of selfishly monetizing his long-ago-earned reputation and sending his goons (to wit, Paul Buchheit and Dan Gackle) to wage petty feuds that shouldn’t even exist.