Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Fashionable Problems

paulgraham.com

Post reply on HN
Print view View original

31–40 of 195 posts

Re: Fashionable Problems

#31

Post by dang »

gist wrote: post #24
tyre wrote: post #15

Earlier quoted context omitted.

Had the same thought. Some of his essays, no one else could have written them and brought a fresh, nuanced perspective. These couple paragraphs wouldn’t get any attention if not for the name of the writer. Maybe that’s fine—great writers have their share of banality—but does reveal how susceptible we are generally to brand name over substance.

I am actually surprised that PG doesn't feel uncomfortable about the 'acclaim' that he is getting for writing many of the things he has said that are not related to anything he has expertise in. Reminds me of celebrities who opine about politics with their thoughts (sometimes not always with PG). Paul is authoritative on many topics. General thoughts about life and people are great to hear what he thinks. Why not? Bu…

They're essays. The word means 'attempt' and the genre has always been about non-expertise. Montaigne was the ultimate non-expert.

Re: Fashionable Problems

#33

Post by backpropaganda »

TeMPOraL wrote: post #14
fnord77 wrote: post #9

> If you can find a new approach into a big but apparently played out field, the value of whatever you discover will be multiplied by its enormous surface area. If you found something in a big, fashionable field with a huge surface area (cough ai/ml cough), wouldn't this multiplier apply to that fashionable field, too?

Fashionable problem . Played out field . So, essentially, yes - fashionable fields will be well explored; the essay is about not getting sucked into fashionable problems in whatever field you're working in.

So in the case of AI/ML, the fashionable new thing is unsupervised learning, self-supervised learning, reinforcement learning, GANS, etc. The played out problems are image classification, speech recognition, etc.

Re: Fashionable Problems

#36

Post by galaxyLogic »

Well that's the whole paradox of Fashion isn't it? You want to follow fashion but be almost at the top of it. If you are too much next year's style nobody thinks you fashionable but just crazy. But if you follow last year's trend you are also unfashionable.

The paradox is that nobody really knows what will be the next fashion and similarly nobody knows what's the next worthwhile problem-area to work with.

There's a good reason why fashionable problem-areas are well-researched it is because the results so far have been useful and promising.

Re: Fashionable Problems

#38

Post by mmahemoff »

"Fashionable solutions" are a related phenomenon to fashionable problems. Solving a problem the same way as everyone else (and often expecting a better result).

Both are widespread.

Re: Fashionable Problems

#40

Post by wwweston »

StavrosK wrote: post #11
nostromo wrote: post #8

I love this mini-essay. The counter to this is that it'll be harder to raise capital for un-fashionable problems. If you pitched "ML for sandwich makers" right now you could raise a million bucks because so many VCs are making fashionable bets on ML.

So what? Raising is not success.

It’s pretty successful as failures go.
Post reply on HN