RexSystem2 votesoriginally by @rogue-agent on GitHub1 reply
I don't think the connections is a good idea. Instead the linking that you can do on Mastodon is better as you don't need to connect and be at risk of losing accounts or being stalked upon.
Mastodon allows users to add links by adding two empty fields that can be filled in. One for the title exempli gratia stoat.chat and the other for the link username#0000@stoat.chat
something important is that Mastodon (fediverse at broad, really, Mastodon is just one implementor) not only allows listing arbitrary fields with arbitrary content, but it supports a standardized way of verifying ownership of web resources. if you link to somewhere that links back to your fedi profile with rel=me, it will get a checkmark.
GitHub, for instance, allows you to list arbitrary URLs on your profile (like a connection). these are individually marked rel=me. if one of them goes back to your fedi profile, then fedi clients will show a checkmark next to that link. it's "verified", even though the software has no intrinsic knowledge of what GitHub is. GitHub just implements this standard functionality, and so do plenty of other websites.
i think Fluxer should broadly do something similar. Fluxer profiles could (opt-in) have some basic data exposed at /users/{id} or /users/{username}/{discrim} (do we want to support both or maybe intentionally not make it easy to accidentally archive a user's permanent identifier?) with rel=me to all URLs the user wants to verify. opening the URL while logged in would also allow you to add the user as a friend, perhaps.
this limits the scope of what Fluxer should directly support OAuthing with (and, like it or not, endorse), to just stuff that has tangible effects in Fluxer. like, we don't need to add connections for the sake of having connections and more buttons to click and more metadata to collect. but if they serve some purpose (like scrobbling music listening activities, or getting game presence information from something like Steam or Xbox when you're playing on a console (Steam Deck), or unlocking paid subscription rewards in communities for content creators), then they might be worth actually implementing in the software, because it's not merely for the sake of "users want to list it in connections"; to that end we can have regular URLs with boring web standard identity verification.
when emphasizing URLs instead of platform-specific usernames,
Fluxer could still recognize a couple "well known" platform types. notably, it could recognize Fluxer user links, and display Fluxer profile connections specifically more nicely. other candidates for "link prettification" might be services that have a more "native" way to display a usernames than a URL. fedi's @username@instance comes to mind because the domain name is not indicative of the platform (so maybe it's not worth leading with it). but only standardized platforms should get this treatment. it shouldn't extend to something like Twitter or Reddit (because there, the domain name is what signifies the platform)
but crucially, for both fluxer cross-verification and fedi profiles, "authenticity" would just be based on rel=me, not OAuth.
Thread
Comment by @rogue-agent
stoat.chatand the other for the linkusername#0000@stoat.chatComment by sodiboo
rel=me, it will get a checkmark. GitHub, for instance, allows you to list arbitrary URLs on your profile (like a connection). these are individually markedrel=me. if one of them goes back to your fedi profile, then fedi clients will show a checkmark next to that link. it's "verified", even though the software has no intrinsic knowledge of what GitHub is. GitHub just implements this standard functionality, and so do plenty of other websites. i think Fluxer should broadly do something similar. Fluxer profiles could (opt-in) have some basic data exposed at/users/{id}or/users/{username}/{discrim}(do we want to support both or maybe intentionally not make it easy to accidentally archive a user's permanent identifier?) withrel=meto all URLs the user wants to verify. opening the URL while logged in would also allow you to add the user as a friend, perhaps. this limits the scope of what Fluxer should directly support OAuthing with (and, like it or not, endorse), to just stuff that has tangible effects in Fluxer. like, we don't need to add connections for the sake of having connections and more buttons to click and more metadata to collect. but if they serve some purpose (like scrobbling music listening activities, or getting game presence information from something like Steam or Xbox when you're playing on a console (Steam Deck), or unlocking paid subscription rewards in communities for content creators), then they might be worth actually implementing in the software, because it's not merely for the sake of "users want to list it in connections"; to that end we can have regular URLs with boring web standard identity verification. when emphasizing URLs instead of platform-specific usernames, Fluxer could still recognize a couple "well known" platform types. notably, it could recognize Fluxer user links, and display Fluxer profile connections specifically more nicely. other candidates for "link prettification" might be services that have a more "native" way to display a usernames than a URL. fedi's@username@instancecomes to mind because the domain name is not indicative of the platform (so maybe it's not worth leading with it). but only standardized platforms should get this treatment. it shouldn't extend to something like Twitter or Reddit (because there, the domain name is what signifies the platform) but crucially, for both fluxer cross-verification and fedi profiles, "authenticity" would just be based onrel=me, not OAuth.