[fluffy]okay so despite OpenID 1.x being dead/dying I am implementing support for it in Authl, because Mastodon and OAuth2-based stuff is annoying me. So, in reading the spec… is it just me or is the whole “delegate” aspect incredibly insecure?
[fluffy]wait no it does seem insecure, like there’s nothing that guarantees that the openID server being specified has anything to do with the delegated URL.
@fvschI see a few people on the Indie Web movement adding Webmentions to their blogs and I'm wondering: it looks like they're all men? What happens _when_ (not if) you have thousands of webmentions that are 95% abuse and threats from white/male supremacists? (twitter.com/_/status/1148159831087419392)
[KevinMarks]The idea of openid delegation is that if you can edit the head to put the rel link in you probably own the site. The bidirectional links in rel-me auth are a stronger version of this idea
@kReEsTaL↩️ Mais bon, ce qui m’a dissuadée d’utiliser les Webmentions, c’est le fait de publier sur mon blog le tweet, le nom et l’avatar de personnes qui ne m’ont pas explicitement autorisé à le faire… Moi-même ça me lourderait d’apparaître sur le blog de je sais pas qui sans permission. (twitter.com/_/status/1148204547950727168)
@kReEsTaL↩️ Alors qu’à la base je ne fais que répondre à un tweet. Et puis quid de la politique de confidentialité, comment être au courant de qui plublie quelle Webmention, comment supprimer ces contenus, etc. etc. Trop prise de tête. (twitter.com/_/status/1148204690317946885)
[aaronpk][fluffy] you're correct, any openid server can make a claim about any URL. Delegation works because the openid client will only trust only the server that the URL delegates to
[aaronpk]I also don't want to do the little client registration dance for mastodon, or add all the mastodon specific stuff, since it would be better for everyone if mastodon adapted their OAuth API to speak IndieAuth instead.
gRegorLove, [Jay_Hoffmann], [schmarty] and [KevinMarks] joined the channel
@fvschI see a few people on the Indie Web movement adding Webmentions to their blogs and I'm wondering: it looks like they're all men? What happens _when_ (not if) you have thousands of webmentions that are 95% abuse and threats from white/male supremacists? (twitter.com/_/status/1148159831087419392)
[tantek]comments << Concerns re: displaying aspects of comments https://twitter.com/kReEsTaL/status/1148204547950727168 translated: <blockquote>But hey, what discouraged me from using Webmentions is the fact of publishing on my blog the tweet, the name and the avatar of people who did not explicitly authorize me to do so ... Myself it would burden me to appear on the blog of I do not know who without permission.</blockquote>
@kReEsTaL@WalterStephanie@fvsch Mais bon, ce qui m’a dissuadée d’utiliser les Webmentions, c’est le fait de publier sur mon blog le tweet, le nom et l’avatar de personnes qui ne m’ont pas explicitement autorisé à le faire… Moi-même ça me lourderait d’apparaître sur le blog de je sais pas qui sans permission. (twitter.com/_/status/1148204547950727168)
Loqiok, I added "Concerns re: displaying aspects of comments https://twitter.com/kReEsTaL/status/1148204547950727168 translated: <blockquote>But hey, what discouraged me from using Webmentions is the fact of publishing on my blog the tweet, the name and the avatar of people who did not explicitly authorize me to do so ... Myself it would burden me to appear on the blog of I do not know who without permission.</blockquote>" to the "See Also" section of /commentshttps://indieweb.org/wiki/index.php?diff=63411&oldid=61746
[aaronpk]I keep meaning to write a blog post describing how mastodon (or any similar project) can use IndieAuth to solve their issues, complete with screenshots
[fluffy]like there’s only one documented endpoint for actually determining the ID of whomever logged in, which requires reading *everything* about the account, and it’s documented as only needing the `read:accounts` scope but in my experience it needs the full `read` scope which is annoying
[fluffy]and yeah I agree that IndieAuth is a way better protocol for this but OIDC at least has mindshare and a perception of being more generally-useful (not htat I agree with that perception)
[fluffy]oh and another really annoying thing is that Mastodon doesn’t seem to have any way of passing a state parameter into the authorize endpoint, which is why I end up registering a state-specific callback each time. But now that I’ve had some sleep I’ll try just passing in state instead. Because like, otherwise how the heck is anyone supposed to know which transaction the callback is in reference eto?
LoqiReport abuse (AKA report inappropriate) is a feature in many (most?) silos for notifying the silo owners that a specific user, post, or comment is abusive https://indieweb.org/abuse
aaronpkagree about vouch, and frankly i'm skeptical that it's anything close to an answer until there are actually implementations of it beyond the 2-3 we have right now
@fvschI see a few people on the Indie Web movement adding Webmentions to their blogs and I'm wondering: it looks like they're all men? What happens _when_ (not if) you have thousands of webmentions that are 95% abuse and threats from white/male supremacists? (twitter.com/_/status/1148159831087419392)
[tantek][fluffy] I feel like "docs suck" is so evergreen / universally true for 99% of OSS that I'm not sure it merits a Mastodon specific criticism / de-karmaing
[tantek][schmarty] I think I agree with the idea behind what you were saying re: vouch and that tweet, it belongs primarily on the UX side (problem to actually solve), *and* I think it is useful input to consider for/when implementing Vouch, for devs that may just end up on the Vouch page and would like an expanded "Why" section (which gathering in See Also is an incremental step towards)
@fvschI see a few people on the Indie Web movement adding Webmentions to their blogs and I'm wondering: it looks like they're all men? What happens _when_ (not if) you have thousands of webmentions that are 95% abuse and threats from white/male supremacists? (twitter.com/_/status/1148159831087419392)
[fluffy]I like the ideas behind vouch as a web-of-trust implementation but holy heck trying to think about how the UX would work on it makes my brain spiral
[tantek]I do think this is an area where we can innovate and provide much better "safe by default" solutions on the IndieWeb than any silo reporting mechanism or server-admin-managed policies
[tantek]it's literally why I'm working on getting reply-context details "correct" before moving onto displaying facepiles or comments which will require addressing reply-context issues *plus* all the potential for abuse
[fluffy]yeah. I think that a much lower-hanging fruit for spam/abuse prevention in webmention et al is having the endpoint do greylisting stuff like what you mentioned
[tantek]I think a key innovation we need to figure out is some amount of automated blocking / filtering that stops stuff *before* it even hits a moderation queue. I would rather not spend much (if any) time having to manually moderate (delete / block) some of the stuff I've seen on Twitter etc.
[fluffy]and a place where a larger shared endpoint (like webmention.io) comes in handy for this is it’s one location that can collect a lot of aggregate information to inform predictions on the value of an incoming thing
[fluffy]like my email spam filter works really well but that’s because it uses a combination of ML approaches and hand-written spam tests. And my spam filter would absolutely not do anything to prevent me from getting heaps of abuse from an Internet hate squad.
[schmarty]cleverdevil, jacky, (and others?) and I were discussing starting some (harmless, I hope) webmention bots that might nudge folks in that direction
[snarfed]also re moderation and anti-abuse, it applies to everything with replies/comments, right? ie it's not webmention-specific. should we develop language around that to counter this kind of concern that webmentions are particularly vulnerable, when they're not?
aaronpkwebmention is already one level better than a comment form or even email, since it takes more work to send a webmention with actual content than send an email or fill out a comment form
[fluffy]That’s a good question. One of the big problems with it is the poison-pill/bad-actor thing, where sometimes a bad actor like Wil Wheaton has a personal beef with someone and then that someone ends up on a global blocklist.
[snarfed]anyway. i wouldn't lead indieweb efforts around shared moderation techniques, but i'd love to see them, and i'd happily incorporate them into bridgy etc if they work!
[fluffy]I’ve seen a lot of particularly bad situations where trans women were blocked by prominent Internet People because the celebrity reacted badly to complaints about transphobic language
[snarfed](also eg bridgy already suppresses backfeed from accounts you've blocked or muted in twitter, which i think people probably aren't aware of enough. https://brid.gy/about#blocked )
[fluffy]or then there’s TERF Blocker which ended up being taken over by someone who started using it as a platform for their personal vendettas against non-TERFs
[fluffy]I think shared blocklists that provide insight/scoring/tags/whatever are helpful, but using them as an automated source of ground truth is hopelessly naive.
[fluffy]Right, or taking lists as suggestions for like “hey these 10 people say this person is bad, you might want to consider blocking them too” but only using first-order suggestions, not suggestions that were brought on second-hand.
@nhoizey↩️ Vouch is not there yet indeed. But it will be even better than comments moderation.
You can already moderate Webmentions manually like for comments.
But it’s not yet native in CMSes, indeed. (twitter.com/_/status/1148280259470663686)
[snarfed]^ i'm not convinced. vouch is a web of trust. does anyone know of any other web-of-trust implementations (in tech) that actually succeeded and went mainstream? i'm not sure i do.
[tantek]blocklist << Example of shared blocklist: https://blocktogether.org/ and criticism, it can and has been (unintentionally?) abused by those with more power (privilege? e.g. celebrity status) to [[block]] those with much less power across a wide swath of “subscribers” to those shared block lists. More: https://chat.indieweb.org/dev/2019-07-08#t1562605567076900 (feel free to extract inline here)
[tantek]I think we will get there, where our sites have a "chat" UI that could be "as simple as" the UI at https://chat.indieweb.org/dev with a text field at the bottom to enter chat "posts" which get posted to one's own site and POSSE'd to IRC
[tantek]indeed since it's on *.indieweb.org, if you've already signed-into indieweb.org with your domain, the chat UI could step up into a micropub client mode that posted what you typed into it to your own site, as well as sending it to IRC
[tantek]and since the web UI knows how its own post permalinks work, it could automatically add the u-syndication links to the posts on your own site via Micropub as well
aaronpk[tantek]: right and it's on a community domain name, and I'd rather interface with that using my site as an identity instead of my site as a data store
[fluffy]argh, everything in the mastodon code indicates that read:accounts is all that’s necessary for verify_credentials, but when I try passing in the read:accounts scope it doesn’t work. Maybe the problem is elsewhere.
[tantek]anyway once you add the hashtags like #indieweb or #indieweb__dev etc. then your backend could filter those accordingly if you wished (from various streams, home page etc.)
aaronpkbacks away from this conversation. there are _so many_ more important and interesting problems to solve before figuring out a new thing like combining group chat and personal websites
LoqiJoin the #indieweb discussions via the web, Slack, IRC, or Matrix interfaces now with additional channels for dev, wordpress, and meta specific chat! https://indieweb.org/chat.indieweb.org
[tantek]aaronpk, consider this week also as "post event" off to keep doing whatever wrap-up you feel like. I am for sure (hope to have some posts up by EOW)
[fluffy]Okay I figured out the problem with the OAuth scopes on the Mastodon authenticator. Turns out in *some* parts of the API the param is ‘scope’ and in other parts it’s ‘scopes’. ffs.
[fluffy]for some reason `scopes="read"` works for the entire read scope but for limited scopes you use `scopes` on the client registration but `scope` on the token grant.
[KevinMarks]so when a big account dunks on you to drown you in crap from their fanclub, it will enable you to excise them all and keep your mentions relatively sane
[grantcodes]Thanks! I just went ahead and published it 😄, still need to update it before I syndicate it elsewhere. 🤔 Maybe I could have made it unlisted for a public draft
[cleverdevil]There used to be Firebug Lite, which at least would be a help. I guess its time to start searching the app store for a developer-focused browser!
[grantcodes]I also added donation support and want to highlight that somehow, but haven't figured how to not be obnoxious about it yet. But it's in the menu for now
aaronpk[cleverdevil]: [benatwork]: since this isn't known-specific, bringing here: have either of you tried making a serverless site in AWS, using only the amazon-managed servers? that'd mean a micropub endpoint on Lambda which pushes content into some managed database, and the website itself would have to be served up from static files on S3
[KevinMarks]I've been doing a bit of lambda stuff and it is a dance. Also, you spend a lot of time giving your various bits permission to talk to each other.
[cleverdevil]It could be used, but honestly, DynamoDB is extremely picky. Its focused very heavily on very high scale performance problems, and as a result has all of these constraints in place that make it very fiddly to use.