This is where someone usually says, “Look, this may work for some people, and that’s great. But that’s not the kind of life I want to lead.” To which I

This is where someone usually says, “Look, this may work for some people, and that’s great. But that’s not the kind of life I want to lead.” To which I reply… “Okay—then I guess you don’t want it that bad.” And then they get upset. They think I’m attacking them personally, calling them frauds or failures. Because they “think” they want it that bad. They’re talented writers… passionate consumers of pop culture… intelligent readers and viewers… and they “think” they want to be a professional writer. But the truth is: they don’t. This worldview is very interesting, and more similar to the Valley ethos than one expects. Because I used to think that software engineering was unique in that you could stare at a screen and intently tap keys to create works of great value to others. No natural resources required. But of course writing books and especially writing screenplays is similar. Yet the political outlooks of your modal Hollywood screenwriter and your modal Silicon Valley startup engineer could not be more different. The former is usually quite liberal, while the latter probably leans libertarian on average. Perhaps this is because software engineering is more about "right answers" than screenwriting. The empirical test in screenwriting is whether the screenplay gets optioned, and whether it makes any money at the box office. These would be analogous to investment and ultimately profits in a startup. Absent, though, is any obvious analog to compiler errors or smaller scale empirical tests for a screenplay. There is spellcheck, but it's at the level of syntax highlighting, not a REPL. If screenwriters could write in emacs or vim and predict the sentiment response to regions of text by highlighting the region and sending it to a subprocess, that would be interesting. I suppose this is technically possible today if you piped the text to Mechanical Turk or ran some sort of automated sentiment analysis on it. Hmmm. It would be sort of interesting to backtest this on existing screenplays to see if MTurk or automated analysis could find predictors of box office hits. With the right featurization, would be surprising if there was no signal there at all.