#metabrainz

/

      • CatQuest
        for situs like this, it would be swell to just lookup without having to do a lot of digging
      • ruaok
        690,463 accounts with no email before jan 1 2016.
      • those can all go away.
      • CatQuest
        when was critiquebrainz started?
      • ruaok
      • 2016-03-01
      • Freso
        "System" user was created Dec 18, '15, blog post was March 1st, https://blog.musicbrainz.org/2016/03/01/say-hel...
      • Well then.
      • pdf's Discourse account was created on Feb 17, '16, so that's probably when I started inviting moderators from the old forums to join.
      • Based on https://community.metabrainz.org/admin/users/li... I think March 1st looks like a decent "cut-off" date. The only people registered before then are fine anyway.
      • ruaok
        Feb 17 sounds like a good date to use for this.
      • 547,607 users with bios over 200 characters.
      • 376,207 with bios over 500 characters.
      • CatQuest
        hmm that sounds susspicious
      • ruaok
      • same of accounts with bios > 500 chars.
      • sample
      • CatQuest
        probably low on anyone's radar, but test is 502ing again
      • reosarevok
        I wonder how many chars mine has :D
      • ruaok
        602
      • ruaok deletes account
      • CatQuest
        hmm i have soem random crap on mine
      • a quick human afirmation will tell most spammers bio from not spammers
      • Freso is at 663 with all formatting and spaces etc.
      • how difficult woudl it be to include a copule of sentences fro mthe bio in the report?
      • and instead of users removing thme manually, just ticking a "this is a spammer" tick o nthe report.
      • ruaok
        easy
      • CatQuest
        thne the automated removal of all those
      • tnhe only the difficult cases are left to actually manually inspect
      • Freso
        The "rob" account has 892 characters. Obvious spam. :)
      • ruaok
        first I think we should nuke all the of the accounts from the suspect domains.
      • then see who is left over.
      • khan____ has quit
      • UmkaDK has quit
      • UmkaDK joined the channel
      • kyan joined the channel
      • hibiscuskazeneko joined the channel
      • hibiscuskazeneko has quit
      • Gazooo joined the channel
      • arbenina_ has quit
      • hibiscuskazeneko joined the channel
      • bitmap
        ruaok: I'm around now
      • ruaok
        moin!
      • zas, yvanzo?
      • and reading the backscroll would be useful.
      • tl;dr: it is not the templates that make us slow, but massive amounts of spamming.
      • bitmap
        huh
      • yvanzo
        actually both but we will take action against spam at first
      • ruaok
        yes. but once we fix spam problems, the slow aspects of templates may simply not be worth fixing. :)
      • drsaunders has quit
      • Mineo joined the channel
      • zas
        5 mins plz
      • ruaok
        k
      • zas
        ok, i'm there
      • ruaok
        right then. yvanzo and bitmap, here as well?
      • bitmap
        here
      • ruaok
        as per google doc, we should be able to get rid of a bunch of accounts really quickly.
      • #2 and #3 are pretty quick to do.
      • #1 isn't too bad either.
      • but we shouldn't do those until we've done more work on nuking spammer motivation.
      • remind me, what exactly did we decide yesterday with regard to setting bios/links?
      • I think we may need to review that in light of the latest findings.
      • bitmap
        <reosarevok> do you need a confirmed email to use CB/LB?
      • ^ was that a yes? didn't see an answer
      • ruaok
        Gentlecat?
      • I don't actually know, TBH.
      • bitmap
        disable public collections for limited users
      • BrainzBot
        MBS-9352: Disable public collections for limited users
      • Gentlecat
        ?
      • ah, no
      • unless OAuth on MB requires that
      • ruaok
        ok, we should change that as well.
      • drsaunders joined the channel
      • I don't think so.
      • bitmap
        well, we can join with editor_oauth_token to see if they've authorized anything
      • ruaok
        I think we need to be a bit more drastic in our actions, then wait and see.
      • if they worked, we can consider easing up again.
      • with that in mind, I would really like to see two things: 1) now show /user and /collection to not logged in users. 2) Do not allow limited users to set website/bio.
      • yvanzo suggested that we could not even show link/bio until they are no longer limited users.
      • thoughts?
      • CatQuest
        but implement it like a "you've leveled up" gamification
      • that's kidna important
      • to not discurage legit users
      • bitmap
        is the idea just to reduce spammer motivation, or reducing lookups on those pages?
      • CatQuest
        for now just make the wording like that
      • zas
        I agree, and restrictions need to be explained during account creation process
      • Freso
        I thought the not-showing-until-no-longer-limited was how it already worked. +1 from me.
      • ruaok
        bitmap: it is both.
      • the only thing known to reduce spammers is to kill their motivation.
      • but it may not work right away.
      • CatQuest
        and not shoing is not enough. not actually letting thmeadd it is important
      • ruaok
        for instance, spammers may still set bio/link even if we don't show them.
      • that still pollutes our DB, even though they've become useless.
      • zas
        About that, how is handled the "limited user" status for now ? a bool ? it could be useful to have more levels if we go the way to restrict features depending on user "experience" (levels!)
      • CatQuest
        +1
      • ruaok
        zas: it would be, but lets focus on that a little later on.
      • zas
        k
      • CatQuest
        this would make it much easier also to reward long standing users both auto and not who work hard on edits
      • righ
      • ruaok
        yes, but that isn't our current goal.
      • our current goal is to get rid of that useless traffic that we're working so hard to serve.
      • CatQuest
        yea i know, i jsut am typing slow becasue i am trying to not typo ;)
      • sorry
      • ruaok
        no worries. just trying to keep focus.
      • CatQuest
        i agree we need to focus on this *first*
      • bitmap
        I only see /collection/create in the top URLs, which you have to be logged in to see anyway
      • Freso
        bitmap: The /collection/ is mostly to prevent DMCA notices, I think.
      • CatQuest
        imho it's important to explain restrictions on account creating, but I also thing it's important to do it in the "you have levels and for now oyu're a newbie account and when you do good edits you'll get more abilities!" sort of way
      • Freso
        We've had a few DMCA takedown requests on user collections in the last couple of weeks.
      • ruaok
        Freso: yes, but via discouraging spammers in the first place.
      • bitmap
        right, that's what I thought MBS-9352 was about
      • BrainzBot
        MBS-9352: Disable public collections for limited users https://tickets.metabrainz.org/browse/MBS-9352
      • ruaok
        we still have a TON of collections and profiles with spam in them.
      • I'm totally open for opening things back up after we've cleaned up.
      • but for now I want to really shut things down might tightly and then clean up.
      • CatQuest
        however. it might be an idea to allow bio entering for new accoutns. *provided* it goes through a fettering process and if certain key (secret!) phrases or words are in it, just make it completely invisible and notify account editors right away
      • .. after the clean up
      • ruaok
        ok, any objections with the current immediate proposed bits? (Disallow setting bio/link for limited users, require login for /collection /user, add /collection /user to robots.txt)
      • Freso
        +1
      • CatQuest
        we should write *why* in these places too though
      • ruaok
        any other comments?
      • CatQuest
        else e are going to get loads of questions about why so and such
      • bitmap
        I don't like it, but fine with doing it for now
      • CatQuest
        +1 to the banner idea reo had
      • ruaok
        which part do you not like? all of it?
      • CatQuest
        i'm ok as long as its deff not permanent
      • Freso
        I don't like /user and /collection requiring login. But I'm okay with it as a temp. measure.
      • CatQuest
        to stop this happening right now so we cna clean up the mess witouth new mess showeling inn all the time
      • +1
      • bitmap
        what Freso said, plus not indexing collections, but I'm okay with not indexing users
      • CatQuest
        i don't see a reason to index collections
      • ruaok
        > Actions to review after a period of time
      • CatQuest
        +1
      • ruaok
        is a new section in the doc.
      • ideally collections should be publicly viewable, I totally agree.
      • CatQuest
        yea ut why does google have to index them?
      • ruaok
        ok, I'll enter tickets for these items.
      • UmkaDK has quit
      • now, onward to automatic removal.
      • CatQuest
        despite the serious amount of junk i am a littel bit sceptical to this
      • ruaok
        each of the three proposed points should have a check for edits and votes.
      • if a user has an edit or a vote, they should NOT be removed.
      • I've reordered the list in that section, so it is roughly in order of priority
      • CatQuest
        what about also querying for certain words/phrases and if they do include them they SHOULD be deleted? on top of the other checks
      • bitmap
        I think the discourse one should just be "accounts with no oauth tokens", maybe
      • CatQuest
        yes
      • lazka joined the channel
      • ruaok
        oh, interesting. yes.
      • CatQuest
        I also thing we should double check with critiquebrainz