[adamlgerber]

i feel like the underlying thesis of this is maybe wrong. someone closer to the methodology would know better but here is what i see:

(1) Meta and Google have seen their growth slow (not shrink) because they reach virtually the entirety of the online population, especially in the US. Meanwhile their time spent metrics continue to rise.

(2) Reddit is called out as a modest grower but its usage has more than doubled in the US since 2021 from 90M to 170M (according to emarketer).

Doenst mean the conclusions are wrong (i agree with it on polarization) but the growth measures seem to not reflect reality.

[jeffbee]

I think you are right to suspect the methods and the results. If you look at the paper's github, the python notebook was clearly written by a chatbot (the comments are all in the second person). So what you have here is a monograph, unreviewed, unpublished, based on GPT-level understanding of a survey that might not even apply to this subject.

[chr15m]

And worse, it says something smart people wish was true.

[throwaway94275]

Meta and Google time-spent growth is probably people watching Reels and YouTube. They're both becoming Tiktok and most of the accounts on Tiktok when I was on it for a while did not look like people's real name. So with regard to Meta/Google "growth" idk if there's anything too social about that.

[adamlgerber]

sure, call it entertainment rather than social. very fair comment, but that is not a distinction this paper is making. the paper is also talking about tiktok too which falls into the same entertainment category.