#dev 2019-07-12

2019-07-12 UTC
[xavierroy], gRegorLove, [jgmac1106], [tantek], amz3, [snarfed], [schmarty], BenLubar, jeremycherfas, KartikPrabhu, vilhalmer, Kaja__, [fluffy], cweiske, dav and leg joined the channel
[jgmac1106] joined the channel
#
@JohnGunson
↩️ I'm sitting here thinking two things - one, that's a badassed Coasty jumping on a microsub like that. Two, why does he need camo? Especially green camo?
(twitter.com/_/status/1149640761790492673)
demomo[m], [kimberlyhirsh], [KevinMarks], jjuran, [tantek] and [benatwork] joined the channel
#
[jgmac1106]
btw folks look at threaderapp now: https://threadreaderapp.com/thread/1149662522770784258.html they are masking the mentions behind a button. I wonder if that is how they got proper with the ToS for the API
#
[jgmac1106]
<span class="entity-reveal-action button">mentions</span><a class="entity-mention entity-mention-first" href="https://twitter.com/TwitterSupport" style="display: none;">@TwitterSupport</a> for any @ mention though @microdotblog came through
#
@WebPlatformNews
Adding Webmention to your website: 1. Sign up on http://Webmention.io. 2. Add <link rel="webmention">to your pages. 3. Fetch webmentions from http://Webmention.io to display them on your page. 4. Use http://Webmention.app to automate sending webmentions.
(twitter.com/_/status/1149681537593614336)
[KevinMarks] joined the channel
[eddie] joined the channel
#
[eddie]
ehhh, seems more like IFTTT
[schmarty] joined the channel
#
[schmarty]
oof that hackernews thread from the main channel. only took two posts to get to talking about vouch 😞
#
klez
[schmarty], my fault :-$
#
klez
why is it bad?
#
[schmarty]
it's not bad, per se, but it gets a lot of attention despite having very few (2? 3?) implementations in the wild and almost no documented user experience.
[tantek] joined the channel
#
[schmarty]
it often comes up in the context of webmentions and dealing with spam and abuse and i feel like it tends to stop the conversation.
#
[schmarty]
vouch may be a useful tool in preventing abuse but if so i think it will be only a part of a larger group of tools and strategies.
#
[KevinMarks]
It assume the existence of blocklist and showlist functionality which few of us have built yet.
#
[tantek]
right. is anyone here moderating their mentions / replies backfeed from Twitter (via Bridgy) ?
#
klez
I'd say it should be noted in the Webmention page where it's linked from (Extensions section) that it's just *one* proposal, as it seems to be *the* proposal.
#
klez
[schmarty], ^^
#
[tantek]
so early on when Vouch was being developed, there was a lot of thought being put into it and we got some implementation momentum. unfortunately (for reasons nothing to do with indieweb), a couple of the early developers of implementations had other life stuff come up and development / interop stalled a bit since
#
[tantek]
as KevinMarks pointed out, developing Vouch revealed that there were simpler things that an implementation should do first, e.g. keeping track of who you did link to and automatically add them to an allow list
[kimberlyhirsh], gRegorLove and [snarfed] joined the channel
#
[snarfed]
tantek: i expect many wordpress users are moderating backfeed. i am. mostly automated w/akismet, but occasionally manually
#
[tantek]
ooh this is a good one (for devs in particular) - worth IndieNewsing IMO: https://inessential.com/2019/07/02/no_algorithms
#
[tantek]
what is algorithmic feed?
#
Loqi
algorithmic feed (AKA algorithm-driven feed or just algorithm feed) is a more correct term for the "algorithmic timeline" lie, and an increasingly common feature on social media silos such as Instagram, Facebook, and Twitter, where they show only some posts from your followings, as well as show some posts only hours or days after they were posted, thus not in chronological order https://indieweb.org/algorithmic_feed
#
[tantek]
algorithmic feed << Criticism: 2019-07-02 Brent Simmons: [https://inessential.com/2019/07/02/no_algorithms No Algorithms]
#
Loqi
ok, I added "Criticism: 2019-07-02 Brent Simmons: [https://inessential.com/2019/07/02/no_algorithms No Algorithms]" to the "See Also" section of /algorithmic_feed https://indieweb.org/wiki/index.php?diff=63564&oldid=51931
[jgmac1106] joined the channel
#
[tantek]
do we need to scrub webmention comments for tracking pixel images? https://mikeindustries.com/blog/archive/2019/06/superhuman-is-spying-on-you
#
[jgmac1106]
Most WP and many Known webmentions moderated through Askimet
#
omz13
Are there any statistics for the signal-to-noise (spam/not-spam) for webmentions? How much is a problem is it really?
#
[KevinMarks]
So far not much; bridged mentions from twitter can be trolling.
#
jgmac1106[m]
Think less than one can count on one hand. All hypothetical, but rather plan for moderation while the river is low ratjer than build walls for climate change after its too late
#
jacky
right - outside of someone making a page and just sending webmentions to your page; I haven't seen anything
#
jacky
I wanted to determine this kind of info when I built my webmention service
#
jacky
to help provide suggestions for moderations (known to spam over the last month etc)
#
jacky
what is limye
#
Loqi
It looks like we don't have a page for "limye" yet. Would you like to create it? (Or just say "limye is ____", a sentence describing the term)
[fluffy] joined the channel
#
jgmac1106[m]
Usually its more someone has something broken and just keeps pinging an endpoint.
#
Loqi
Webmention
#
omz13
Good to hear. I'm soon (famous last words) about to deploy my own webmentions implementation... so far I've been concentrating on catching malformed requests than spam
#
GWG
I need to work on improving moderation for backfed webmentions
#
GWG
But there's an issue
#
GWG
Most spam filtering software is not geared toward checking backfeed
#
GWG
They aren't looking for spammy web pages.
#
GWG
Which is a whole different set of signals
#
jgmac1106[m]
Askimet does well. Gets some false positives but never misses spam... And never had a spam webmention comment... Some tweets from Bridgy but they don't know they are spamming my site
#
jacky
GWG: you're saying like checking the poster from the backfed comment?
#
GWG
jacky, for backfeed, yes
#
jacky
hmm I'd want something like that personally
#
GWG
If I go by domain... I am either trusting anything from Bridgy or anything from the backfed site, let's say Twitter
#
jacky
right that's my issue right now
#
GWG
Who trusts every user on Twitter if you trust one user?
#
GWG
So, back to my statement that Akismet and other traditional spam techniques aren't geared toward backfeed
#
GWG
So, an interesting conversation is how you might do that
#
jacky
could pull out the links in a bridgy-formatted post
#
GWG
Vouch can't cover this because Twitter (still acting as my placeholder for all backfeed) isn't going to do Vouch
#
jacky
feed those to akismet
#
omz13
Talking of webmentions... when one is received and an attempt is made to get the source (to check for links), what is the best way of handling 5xx errors (my idea at the moment is to retry a few times - in case its a transient fault - before giving up and rejecting)
#
GWG
So the responsibility is on the receiver
#
jacky
to a degree
#
jacky
there could be a service that sits between someone using bridgy and themselves that does this
#
jacky
omz13: I _think_ the spec has something on this
#
jacky
but yeah doing a fall back retry (30s, 1m, 3m) is safe
#
jacky
failing closed at 5min
#
jacky
re: hn pwa apps
#
GWG
Akismet has a blog post comment type
#
jacky
is looking for one that pops out
#
sknebel
jacky: I'd maybe err towards longer retries, at least my reaction time to site outages isn't 5 minutes :D
#
jacky
lol that's fair
#
jacky
by the time an error bubbles up three times, I'd get an email
#
jacky
first time is just an logged alert
#
omz13
For transient errors (particularly 502/504 ) I was planing a retry at e^n, and give up after 24 hours... perhaps the 24 hours is too genereous?
#
sknebel
I feel like once you have the infrastructure for more than a few minutes (meaning you likely have some scheduling/storage mechanism), giving a generous time is probably not more effort on your side, so why not give it a lot of time?
#
omz13
True. I finally got my scheduler working this morning... so the effort is as simple (now) as just saying what the retry limit is before saying "I give up"
#
[KevinMarks]
fair this is a dev topic - I wonder about 2 things - mung HN format into Microsub, so all microsub apps are hn readers too
#
[KevinMarks]
secondly, look at those example apps and see what it would take to swap out the HN api for websub
[grantcodes] joined the channel
#
[grantcodes]
[KevinMarks] I had the same idea but you could to a microsub & micropub to Twitter API bridge. But then I thought it was dumb 😅
#
[snarfed]
omz13: i believe we've only ever seen *one* actual native spam webmention
#
[KevinMarks]
well, you already kind of have that - you can read twitter with granary, and post to it with silopub
#
[KevinMarks]
the escort agency on adactio?
#
[snarfed]
KevinMarks i think so! tried to find it on the wiki with a few searches but didn't. aaronpk talked about it a little at the time, maybe he could find and add it? it's history! we should document it!
#
[grantcodes]
Exactly [KevinMarks] so I think HN should be handled more or less the same way shouldn't it?
#
[snarfed]
sknebel++
#
Loqi
sknebel has 43 karma in this channel over the last year (116 in all channels)
#
[KevinMarks]
Did that site's owner go to hwc Nottingham?
[cleverdevil], KartikPrabhu and amptwo joined the channel
#
@keithjgrant
↩️ If you can figure out how to do it with a netlify function, I'd love to see it! Also, I use this to send webmentions from new posts (deployed to Heroku), though it's still lacking in documentation: https://github.com/keithjgrant/github-to-webmention
(twitter.com/_/status/1149740368885190656)
eli_oat, gxt, [fluffy] and [tantek] joined the channel; KartikPrabhu left the channel
KartikPrabhu, [mapkyca], [KevinMarks], eli_oat, j12t, [snarfed], PermawebMatrixBr, [eddie] and jjuran joined the channel
#
@iChris
↩️ I’m curious about what feels like yet another app locking your comments in like Disqus vs integrating with webmentions? https://webmention.io
(twitter.com/_/status/1149782177812492292)
#
@tomcritchlow
↩️ I am ashamed to admit I have not been able to get webmentions working :( I like the concept but.... Disqus: ad-tech funded nightmare (but nice ux!) Hypothesis: non-profit with poor UX but strong ideas Webmentions: a commune in a forest?
(twitter.com/_/status/1149784378941345792)
#
@iChris
↩️ That's probably a fair assessment. It feels like if the tech folks got behind something like webmentions, it could be cool - but I know it's a long road before that might happen. Hypothesis seems like a good middle ground.
(twitter.com/_/status/1149786294702215173)
KartikPrabhu, jimpick[m], [fluffy] and [manton] joined the channel
#
sknebel
[snarfed]: e.g. https://webmasters.googleblog.com/2019/07/a-note-on-unsupported-rules-in-robotstxt.html "Disallow in robots.txt: Search engines can only index pages that they know about, so blocking the page from being crawled usually means its content won’t be indexed. While the search engine may also index a URL based on links from other pages, without seeing the content itself, we aim to make such pages less visible in the
#
sknebel
future."
#
sknebel
(not an example, but documentation)
#
[fluffy]
[snarfed] Oh I agree that “good enough private” is good enough for most uses of privacy. There’s just been things that I’ve posted as unlisted things that I’m still worried that someone decides to share the link with others without my knowledge/approval.
#
[fluffy]
And it’s hard to know when someone’s been doing it.
#
[snarfed]
[fluffy] heh i think sknebel meant take robots.txt details here. unlisted/private is still probably ok for main channel
#
sknebel
yep, sorry
#
sknebel
wanted to get out of the way of that
#
sknebel
You are seeing this result because the page is blocked by a robots.txt file on your website. (robots.txt tells Google not to read your page). This tells Google not to read the page (which is how we generate a description), but it doesn't tell Google not to show the page in Search results.
#
[snarfed]
thanks sknebel. and kinda disappointing
#
[snarfed]
i'm curious which searches will show blocked pages then
#
[snarfed]
if it's same as usual, then yeah, robots.txt seems kinda useless
#
[snarfed]
aha, a noindex meta tag blocks int in search results entirely. https://support.google.com/webmasters/answer/93710
#
sknebel
yes, but that doesn't get seen if it is in robots.txt, which is the kinda self-defeating trap
[asuh] joined the channel
#
[snarfed]
ah ok. so then we noindex and don't robots.txt?
#
sknebel
probably? you could add another test post with that combo
#
[snarfed]
which somewhat addresses [fluffy]'s concern of robots.txt leaks. somewhat.
#
sknebel
and we check in a few weeks how far google has gotten from your public post and the wiki
#
[fluffy]
yeah meta robots and x-robots-tag are basically equivalent. I use the header in Publ because it’s easier than trying to put it into my templates.
KartikPrabhu joined the channel
#
@andybelldesign
↩️ @maxray Mate, even better you can use the page on my site thanks to webmentions! https://andy-bell.design/notes/213/
(twitter.com/_/status/1149814247314141185)
cameron1 joined the channel