https://www.biorxiv.org/content/10.64898/2026.11.06.764567v1

(Found with Aftergleam, a private reader for new research papers.)

Decentralized social media for environmental research

A. Chai-allah, A. Ghermandi, N. Fox, D. Van Berkel bioRxiv · 10.64898/2026.11.06.764567v1 · ecology · submitted 2026-11-12

Abstract

Social media data are reshaping environmental research. However, recent ownership and policy changes on centralized platforms like Twitter/X have motivated many users to seek new alternatives while simultaneously accelerating data restrictions for scientists. Decentralized social media, particularly Bluesky and Mastodon, are emerging as alternatives for users seeking greater control over their data and as a more stable and sustainable source of data for researchers. This rise may be instrumental in revitalizing the role of social media in environmental research. This study aims to provide a first-of-its-kind analysis of the potential of these new sources for environmental research, demonstrating their potential usefulness through the examples of climate change discourses and people's appreciation of wildlife. We find an increased adoption of Mastodon and Bluesky, with high linguistic diversity, even if dominated by English. Wildlife image content reveals broad taxonomic diversity, with plants, birds, and insects dominating the records. We find this pattern broadly consistent with that observed on the popular citizen science platform iNaturalist. Climate change discourse is found to broadly reflect the scientific consensus on both platforms, with contrarian claims remaining uncommon. This suggests that researchers may benefit from less polarized opinions and misinformation, arising from interactions that are less driven by engagement maximizing algorithms implemented in other platforms. Our study illustrates the promise of decentralized social media as a new data source for environmental scientists and paves the way for their broader use in environmental research.

the system time is also incorrect.

my region just shifted daylight savings time this past weekend and only the tv has wrong time, 1 hr mismatch.

updated system, rebooted multiple times. still no content in timezone settings panel.

steam deck (yeah arch user, btw 😋) was also wrong by same time amount. but i was able to fix it by changing the timezone in system settings.

i guess it's lazy of me not to check git issues myself but i assume Ubuntu and kde have dozens of possibly related repositories. that would take me days to wade through.

i think i stink too but I'm used to my own smell.

but it took me a long time to realize smelling others' (mostly strangers') breath and b.o. has been causing me background stress and using up my spoons, probably since birth.

especially if someone acts insensitive or straight up assholian, even breathing the same air as them feels demeaning and like defeat.

so I'm finding it hard to take my mask off permanently (or semi-perm) even though i have all the vaccines that medicine allows.

the problem is that i can barely breathe with n95 (or kn95) and it makes everything physically hard. but not wearing it makes everything mentally harder.....

should i talk to my therapist about it? med Dr? or both?

edit: forgot to mention, cologne/perfume, lotion, fabric softener, etc also bother me. certain shampoos seem ok, i don't know why. sandalwood (like east Asian incense) is ok too.

https://xcancel.com/dicehitgames/status/2086845323805159777

had anyone else noticed YouTube front end apps/sites unable to get updated lists of videos or any lists at all? I haven't checked all the symptoms to be able to guess the root cause.

i assume YouTube itself hasn't gone down. are they blocking front ends again, maybe in a new way?

to catch up those who aren't aware, the overwhelming majority of native Japanese speakers use 12-key flick input to type on smartphones, as it evolved from the input on flip phone (garakei) hardware keypads. it also requires less keystrokes => faster than typing romaji, once accustomed.

for context, the most widely recommended Japanese keyboard on android has been Google Gboard, MS Swiftkey, etc. tbh, these are very good as a keyboard, but unfortunately also great at collecting personal private data. the previous open source choice Mozc (created by a Google from a gentler timeline) has been outdated since before the pandemic.

thankfully, now there are at least 34 open source projects working on east Asian language (primarily Japanese) flick keyboards for mobile devices:

スミレ (thanks to stanman for the link) is privacy-respecting and has the essential features one would expect. also available via play store if you prefer.

FCITX, is presumably working on it.

FUTO keyboard, and Florisboard

~~so far, i think FUTO seems the furthest along, with ~~ For FUTO fans, there's a nightly build for android available already. if you try this, remember to import the kanji conversion dictionary file linked in this comment since it was excluded from the apk to reduce file size.

this is very exciting, as only a year ago, there was hardly any mention of work being done. the only solution i could find for modern smartphones was by a Graphene OS forum user who had recompiled the latest binary of Mozc (from 2019!) while splicing in modern components for those that had become outdated. the result was janky to say the least.

edit: updated courtesy of stanman

if anyone knows other open source Japanese keyboards (that are up to date, i.e. not original Mozc, and privacy-respecting, i.e. not Gboard), especially with 12-key flick, please share!

previous related post

This could also help minimize effect of "Nicole the fediverse chick".

even using FreeTube, i have to change audio track and subtitle settings almost non-stop...

it's slightly less annoying to see wrong/unnecessary subtitles. but it's pretty confusing/tiring when the algorithm interprets sound effects or music (instrumental even) as Hindi or Uzbek or something and i have to turn off subs even if there is no talking.

audio on the other hand 🤬👿👹

even if I watch a video not in either of my 2 fluent languages, i don't ever want to hear any dubbed/ai-gen audio track.

any idea if this is in YouTube's issue tracker?

regardless, in the absence of fix from YouTube side, any clue if this can be improved on the client side?

PS - any responses telling me to just watch peertube instead will be downvoted and user will be blocked. fyi i don't reject watching peertube but my chosen antidote for the corporate streaming disease is 🏴‍☠️

edit: forgot to mention the video title auto translation presenting any and all languages' videos as if they're in my language 👿👿👿

To what extent will these changes, and potential future changes along the same trajectory, affect the use of un-substitutable apps on Graphene OS?

Play Integrity API - 2024 Dec

The transition to the new verdicts will reduce the device signals that need to be collected and evaluated on Google servers by ~90% and our testing indicates verdict latency can improve by up to ~80%.

the huge ratio of reduction suggests to me that the attestation is being offloaded from Google servers to on-device AI, but maybe i assume wrong. my instinct tells me Google would always make this impossible for 3rd party OS to implement anyway.

Hypothetically, If implementing that AI in Graphene could allow most attestation-requiring apps to install and run normally, is that something the Graphene devs would do? i know it would have to be secure and private, so assuming there was a way...

I'm not pro AI, not the surveillance big data capitalism kind at least. just wondering about the scenarios and what prep or extra work i would have to do.

In my understanding, the options need to be customized for each machine, as well as the fact that making packages for tons of distros can be a lot of work for solo or small team devs, and that's why some software is provided as .tar only.

but it seems like the install process on the user side could be automated to a single command or drag-drop, as long as the script would throw informative alerts for any errors and the user is prepared to take over manually.

does something like this exist in a standalone form that's not bundled like snaps or flatpaks?

if not, is there a broadly-applicable reason (security, damaging OS, etc) that makes it a terrible idea? or simply that no one has gotten to it?