Latest #FOSSAcademic post: Researching the fediverse from the perspective of individual or instances. In which I draw on an article Christina Dunbar-Hester to talk about researchers' perspectives on the #fediverse, including my own.
@rwg Oh wow, thanks for this! I have a companion piece drafted, not sure if I will finish it, on moderation experience that accompanied this incident as I tried to get to the bottom of it (I see what you did there 😂). It is a little more instance-to-instance, tho still thru user's eyes.
Anyway agree fully abt how level/scale of where researcher looks affects analysis. (Usually in my research I've focused more on the mediating layer too, rather than end users; this had, uh, peculiar origins)
@rwg Thank you!! Back at you--grateful for your work!
(Lol, a main reason I am more submerged here is in fact the asstodon antagonist: when he reported me at work, I stopped using even my abbreviated name)
@rwg@inquiline thanks you both for this work. It is exciting to see fresh conceptual frames arise through study and experience in open social.
One thing the "lossy" concept brought up for me was the anti-lossy character of Bluesky and the crypto socials that use Merkle Trees: https://protocol.ai/blog/transcription-bluesky-blog/ That actually shocked me, that the data structure is built around cryptographic preservation, as I think of social chatter as intuitively lossy.
@ntnsndr@rwg I actually had a footnote about Bluesky but I cut it because it's really far from what I am paying any attention to (except for the broadest strokes). But yeah this is a very good contrast to draw; and blockchain is great foil because it is the fantasy of perfect, infinite, lossless communication
@ntnsndr@rwg@inquiline a fediverse server that did deliberate loss[1] of old messages would be such a fascinating thing. I think a lot of people say they prefer disappearing content, but often the revealed preference is for permanence. (I wonder if @signalapp folks has ever considered deletion-by-default.)
[1] could be just taking private older messages, or perhaps complete deletion; perhaps for all messages, or just those that fall below a certain popularity threshhold; etc.
I know folks (including Lurk folks) who argue that we really ought to remove the expectation that messages will live forever, so that we can reduce the environmental impact of the fedi.
I'm not sure if storage is the main trouble maker. (It could be.) We're currently arranging a data sprint event with data centre professionals where we want to push them, among other things, on the always-on nature of data centres. That's where the magic is. Like, shutting things down/slowing things down so that a network is good enough
Thanks for the mention. Yes for Lurk, storage is the main trouble maker and that has to do with the ever-expanding remote media cache of Mastodon. If you do not subscribe to a cloud model (that assumes storage is cheap, while it isn't) this becomes a problem. The remote cache is necessary at the same time, because that is what makes local moderation policy possible (you always respond / work on a local copy of remote content). We have a forthcoming position document on why lurk won't scale. It ties together the ecological, social and financial sustainability of the instance and the consequences that has for instance policies (fixed number of accounts that matches what the hardware supports, scepticism towards automated posts / LLM content, no federation with threads).
@stefanlaser@rra just chiming in to say, very (1) informative thread, thanks! and (2) would love to hear more about the "without commercial cloud, this doesn't work" -- I am increasingly thinking about problems where "open" assumes no-cost scaling and this is an interesting example of that problem.
Add comment