Nothing should dilute or adulterate the exalted joy of watching the chess engine you've toiled over for the better part of three weekends start to suggest moves (from a basic 3-ply negamax search with a simple point-counting position evaluation heuristic), unless it's the slight(ly overdetermined?) suspicion that you're overcompensating for something, that you've proved your point by now, that bringing yet another moderately-sophisticated side project in a not-the-most-popular programming language over the threshold of "really cool-looking proof-of-concept" isn't going to show Everyone that you are Smart and should be Respected any more than the last seven already did. Some people actually use software for something other than a trophy, to automate some aspect of the world that otherwise would have been done more poorly. So you've heard. If one were to hypothesize, for the sake of argument (but perhaps not only for the sake of argument) that there can exist diminishing marginal returns to some games, that Respect from Everyone is not a real thing that can be won, that there are treasures and masteries you'd never imagine while chasing GitHub stars, much like how you know there are treasures and masteries that you'd never imagine while chasing school marks—what strategies would that imply, now that you know there is such a thing as being strategic? And how would you tell the difference?
Leave a Reply