@julian
We appreciate that, but the feedback raised valid points. The opt-in approach is the right move. It may grow slower, but every user in the index will be there by conscious choice. That's a stronger foundation to build on.
Discussion
Loading...
It seems this issue has resurrected, or resumed
My entire point (all the noise notwithstanding) focused on
Default opt-in versus default opt-out
This is an agent --> recipient transaction
Default opt-in: the recipient is opted into (and participates in) the action whether the recipient --> knows of <-- the action or not
Default opt-out: the recipient is opted out of (and cannot participate in) the action whether the recipient --> knows of <-- the action or not
Neither default opt-in nor default opt-out have any logical meaning if THE RECIPIENT DOES NOT KNOW OF THE ACTION in advance
Here is one of the oldest prior discussions I participated in on Github with Ryan (snarfed) Barrett re: bridgy-fed back starting in February of 2024
Read all of it; it's very long
Here: https://github.com/snarfed/bridgy-fed/issues/880#issue-2136207725
With that, I'm out
Y'all have fun, eh?
@apps As experienced client devs, I assume you have the infrastructure for OAuth already in place? I'd rather go to your website and authenticate my Mastodon account there than have a hashtag in my bio... the Tootfinder hashtag kinda annoys me every time I see it. 😐
That aside, I believe Mastodon is going to introduce some kind of profile signal for willingness to appear in starter packs, might be worth additionally keeping an eye on whatever happens there. https://github.com/mastodon/featured_collections
@julian
OAuth would mean implementing specific integrations for each Fediverse software, and some don't even support it. The bio hashtag works universally across any ActivityPub implementation without any server-side dependency.
You can also opt in via mention if you don't want the hashtag visible in your bio.
@apps Okay yeah, that's fair. Takes me back to how I added a choice between Mastodon-style OAuth and DMing a confirmation code for account verification to FediRoster.
@apps Rather than a hashtag, is it possible to do this based on users following a particular account? Like the Bluesky bridge.
I don't know any of the technical particulars, so unsure if this is a feasible alternative. Feels like it would have less friction, though.
@tehstu
Our previous mechanism was actually more thorough than that. Before following or indexing anyone, we checked multiple signals: indexable enabled, account not locked, no #nobot in bio. But since indexable is enabled by default when you create an account, we were closer to opt-out than opt-in. That's why we're rethinking the approach with explicit opt-in.
@apps If I understood it all correctly though, you would follow accounts based on these criteria.
I was thinking of the reverse, where people elect to follow the discover account. Perhaps some settings-based checks could also be applied, as you did previously. Looking to make sure the account is indexed and so forth before following back (which I guess would be the thumbs up that the discovery is now active).
Apologies, I'm problem solving things I imagine greater minds have already considered while putting off what I should be working on (which is making me tear my hair out). Either way, I look forward to whatever ends up working. The Holos stuff is such a novel concept!
I also like the Follow approach best.
Follow the HolosIndex bot gives explicit consent "please index my public posts". HashPub by @evan follows a similar method, and also https://anagora.org by @flancian works on that basis. Give a nice acknowledgement after the follow request, and perhaps an unfollow removes you from the index again.
I find this a better mechanism than the magic hashtags in profile approach, where there's already tight charlimit in many apps for the intro. You also have (or had) this with https://fediverse.info when adding #fedi22 to ones profile.
@smallcircles @flancian @apps @evan @tehstu Yep, opt in by following the account is the way to go. Parsing hashtags is really clunky and has more potential for user error.
@apps
That sounds a lot like https://tootfinder.ch does it. It would be great if different services could share the same opt-in mechanism! Explicitly allowing one service while denying others might be important for some users and should be supported, but an additional common flag to just allow all services would be convenient for lazy guys like me.
@apps
Very good idea. 
In my opinion, the previous solution was fine, but this would make approval absolutely clear.
Re: After shutting down #HolosDiscover, we're rethinking the approach with #HolosIndex: users explicitly opt in by adding #HolosIndex to their bio with interest tags, then submit their profile.
@apps@toot.fedilab.app honestly? You should ignore the haters and not make this opt in. You're crippling its usability that way...
@julian
We appreciate that, but the feedback raised valid points. The opt-in approach is the right move. It may grow slower, but every user in the index will be there by conscious choice. That's a stronger foundation to build on.
@apps
Well, I read the whole thread. It was a matter of 2-3 people which were more a PITA than giving useful advice.
In my opinion if HolosIndex will be more strict than it already was, since it was very strict in what to index, it will be not useful, since almost no one is going to add #HolosIndex tag and it is not even an ActivityPub standard.
@nicolaottomano
We actually received links to conversations where the project was being discussed quite aggressively beyond what was visible here. We still believe our approach was ethical, but it didn't account for users not understanding their own default settings. Of course, the same settings used by real scrapers like Google, but that's another debate :)
@julian
Re: After shutting down #HolosDiscover, we're rethinking the approach with #HolosIndex: users explicitly opt in by adding #HolosIndex to their bio with interest tags, then submit their profile.
@apps@toot.fedilab.app that's exactly it. You're not responsible for others not understanding that when they post publicly, their post is public.
@julian I do think there's a bigger issue here, if a significant number of users are not understanding discoverability settings then we have to figure out a way to explain that functionality better.
If we fall back to every ActivityPub app/server having a unique discoverability configuration then we will have a situation where nobody understands the settings.
2+ more replies (not shown)
@apps @julian maybe there is a third way here, perhaps a lot of users dont actually know about that flag, or that its only going to be ignored, not respected, by those wishing to do harm anyway, perhaps a bot account that reminds people to take care with what they post if it contains stuff like photos with location data?