News for hackers. Stuff that matters.

puppy_nap

The M-* series of competitions are the current major forecasting competitions in the time series world.

https://en.wikipedia.org/wiki/Makridakis_Competitions

I believe the top performing models in recent years have all been gradient boosted trees, usually then fed into some secondary ensembling scheme.

Re:Meta Just Proved People Hate Chronological Feeds on Thursday July 27, 2023 @05:02PM (#36900125)
The fact that chronological feeds drive less user activity than ML-based ranking engines isn't an interesting result. Presumably the ranking algorithms have access to timestamps and would have just learned to order posts by time if that maximized engagement.

Of course, more activity doesn't mean your user is actually having a better experience. We have all seen social media addicts who spend hours and hours doomscrolling yet are miserable the entire time. Users spending less time on Meta didn't mean they "hate" the feed, from what I've seen, people who actually enjoy social media are those who spent very little time on it.

This article brings up a good point, which is that in an environment where all the big players are willing to abuse their users, those who do try to make their platform less addictive will just end up losing their users to those who don't. The only real solution is industry wide regulation.

These papers suggest that LLMs do something a lot more specific (when asked to simulate a certain political background, they're able to give responses to questions in a way that's consistent with those political backgrounds). That's not particularly surprising to me as I would expect a human to be also able to simulate this kind of thing pretty accurately. I don't think it implies that LLMs would be good at answering typical business survey questions.