Email or username:

Password:

Forgot your password?
jonny (good kind)'s posts Post Back to profile
Top-level
jonny (good kind)

just took a look at about a month of atproto firehose i have just been accumulating, and it looks like it's time for an update to the ol "is it becoming a communication medium yet" and the answer is even more no than before.
1% of accounts receive 72% of interactions (up from 44% last december when the network was a fraction of the size),
1% of posts receive 56% of all interactions, and
almost 90% of posts receive 0 interactions.

the distribution is steep too in the high end of that tail. Scrolling through the default feeds rn on a secondary account following zero people and with zero interactions, posts are averaging in the ~hundreds up to a tens of thousands of interactions. on my actual account where i have interacted with people, i receive the fixed proportion of low-interaction mixins from my network which is like 30-40%. Think about how common seeing a post with hundreds of interactions is tho in the default feeds - 0.01% of posts receive 470 likes, and 0.0001% receive 6300. That's how much the algorithmic amplification makes a monoculture.

I have been taking samples of fedi while developing fetch all replies and backfilling, and the distribution on AP fedi is... not like that... but i haven't taken a systematic sample.

one prior post, i'll find the other later:
neuromatch.social/@jonny/11165

Edit: to be clear, this a month sample of all likes and all accounts that were active in that month. So not all accounts from all time

25 October at 4:48 | Open on neuromatch.social
6 comments
Kevin Riggle

@jonny darling power laws are normal here, this is a good sign

25 October at 5:01 | Open on ioc.exchange
jonny (good kind)

@kevinriggle i would be hard money there are strong preferential attachment effects and the kind of amplification you would expect on AP fedi too. one weird one tho that i think is counterintuitive is that there is like an inverted U shape to preferential attachment-like effects based on instance size. on very small instances, there are only a few people so not much amplification possible, but at like ~dozens to ~hundreds where the local feed is actually usable there are definitely strong amplification effects, especially on glitch instances that show boosts in the local tl, as ours does, but in a way that i think totally rocks. i see what the people around me are talking about every day. it needs some tuning to hide people like me who post too much, but still it's there. but then on the biggest instances where there's basically no instance effect, it inverts and becomes negative where a lot of instances have blocked them in part because of their sheer size. being on a big instance helps with network visibility outwards (for now...) but not "inwards."

So i want to figure out a way to sample from the fedi respectfully and responsibly because i think there would be some super interesting dynamics in the network that i want to try and approximate probabilistically without scraping the whole fedi.

@kevinriggle i would be hard money there are strong preferential attachment effects and the kind of amplification you would expect on AP fedi too. one weird one tho that i think is counterintuitive is that there is like an inverted U shape to preferential attachment-like effects based on instance size. on very small instances, there are only a few people so not much amplification possible, but at like ~dozens to ~hundreds where the local feed is actually usable there are definitely strong amplification...

25 October at 5:53 | Open on neuromatch.social
jonny (good kind)

@kevinriggle when i log back onto my server i'll send you the exponent because like all social systems exhibit power laws, so that's a normal signal, but the power of the power law is extreme

25 October at 7:25 | Open on neuromatch.social
jonny (good kind)

last thought on this for the night: one of the biggest red flags to me is that the default algorithms are conspicuously closed source. that was certainly a subject of the VC funding negotiations.

"do you have room to use the algorithm as a revenue generator if you need to, yes or no" and the answer is yes.

25 October at 6:53 | Open on neuromatch.social
jonny (good kind)

conspicuously, as in there is no mention of this anywhere, the last trace of it in the code is from a database migration a year ago, and literally every other part of it is open source.

25 October at 6:54 | Open on neuromatch.social
Jon P

Not surprising. Starter packs and most feeds have rich-get-richer effects ("quiet posters" is a rare exception). And more generally, a "Twitter alternative" inherently includes stuff about Twitter than many find prolematic as well as the good stuff. @jonny

25 October at 16:51 | Open on gotosocial.thenexus.today
Go Up