tesseract/src/lib/blacklists.ts at main · asimons04/tesseract
github.com
Contribute to asimons04/tesseract development by creating an account on GitHub.

cross-posted from: https://lemmy.dbzer0.com/post/72685299

I discovered this after upgrading our instance yesterday, which also upgraded all frontends to their latest versions as well. After I did, our Tesseract frontend stopped working and I noticed immediately as it’s been my primary for a while. Initially I thought it was an API version mismatch, but no, it was much worse.

Someone pointed out that the developer of the frontend added an explicit hidden and unmodifiably blacklist which includes any and all instances to the left of Kissinger. There’s a commit which also explains their specious reasoning about our instance specifically, as it seems we’ve been on their shit list for a bit longer than that.

This instance and its admin staff encourages identity politics, groupthink, mob mentality, and extremist solutions to societal problems. Users who advocate violence are not moderated so long as the admins agree with the target. Caution and critical thinking are advised when interacting with this instance or its users.

When you use tesseract to connect to one blacklisted such instance , you just get a message informing you that Tesseract is “incompatible with that instance” which leads one to think of a technical issue, like an API mismatch, rather than the dev being an opinionated coward.

Isn’t it funny how all the software developed by turbolibs, like Piefed and Tesseract, end up with hidden control mechanisms from developers who think they know better than everyone else? That they don’t just think they deserve to tell you what you should think, but they should manipulate you to think it? Isn’t it funny how libs go on about how bad it is to support lemmy due to the ideology of the devs behind it, and yet lemmy has 0 opinions as a software? It does make one think…

Anyway, I forked - it as one does - and disabled the blacklist, but since this is a massively ideologically compromised software, I’ll doubt I’ll keep this frontend up after Lemmy 1.0. I think we’ll bring up mlmym again now that someone’s maintaining it again.

  • QuentinCallaghan@sopuli.xyzEnglish
    58·
    1 day ago

    A frontend should be a frontend. The decisions of what to blacklist should be up to the frontend’s user. Stupid decision on the developer’s part.

  • Zak@lemmy.worldEnglish
    43·
    1 day ago

    Opt-in denylists with pre-populated categories would be a reasonable feature. A frontend author trying to trick users into thinking there’s a technical issue with servers the author doesn’t like is presumptuous and disrespectful of the user.

    People did similar stuff with Mastodon clients when the far-right microblog service Gab migrated its backend to a Mastodon fork and I didn’t like that either.

      • Axolotl@feddit.itEnglish
        1·
        9 hours ago

        Really? I was thinking of switching to piefied (i was already skeptical because it’s made with python) but now? Hell nah, i’il stay on lemmy or use mbin

        • OpenStars@discuss.onlineEnglish
          1·
          2 hours ago

          I am not saying that you should switch to use or even like PieFed, but afaik there is no feature that is “trying to trick users into thinking there’s a technical issue with servers”, or at least not one that Lemmy and Mbin do not identically share. I would be happy to read about one though if someone wants to point me to it?

          e.g. just a few hours ago I recall reading a report where someone was confused why a post did not show up for them on PieFed but it did on Lemmy. The reason turned out to be that the user has been banned on PieFed.social. The key here is that the latter is an instance (a server machine) whereas PieFed is software.

          Lemmy instances also ban people too btw. Some Lemmy instances are in fact quite (in)famous for this.

          But that is not “trickery”, that is simply PieFed (or Lemmy, or Mbin, whatever) working as designed, someone’s lack of understanding notwithstanding, and likewise for people who want to cast aspersions onto their tribal “opponent”.

          All that said, PieFed does have a weird way of blocking all users on an instance. If the devs of Lemmy, PieFed, and Mbin could get together an agree on a protocol whereby a signal could be sent that all software platforms would understand and know how to handle… then such maneuvers would not be necessary. Instead, legacy software like Lemmy are not always going to be able to keep up with PieFed’s rapid pace of development - e.g. PieFed offers numerous things that Lemmy simply has to shadow-block (with no report of having done so), including polls, user and post flairs, various special features of communities (e.g. automatic deletion of submitted images after a certain time period, usually one year), etc. To be clear this is not “nefarious” activity on behalf of the Lemmy devs, it’s just that the software doesn’t know how to handle such content, nor do the devs make it a priority to make it do so (why should they?)

          And likewise, the ability of Lemmy uses to reply to PieFed users that have instance “blocked” them is one such source of friction between the two software offerings.

          If you are thinking of creating a community, it matters far more what instance admins you choose to interact with than what software that instance is running. Especially as some (like quokka, and reportedly soon-ish sh.itjust.works) switch around, while others (like .zip and .world) offer literally both options (Lemmy & PieFed).

    • Carl N. Yon@lemmy.worldOPEnglish
      131·
      1 day ago

      Every account on PieFed has a Social Credit Score, aka your Reputation. If your account has less than 100 reputation and is newly created, you are not considered “trustworthy” and there are limitations placed on what your account can do.

  • poVoq@slrpnk.netEnglish
    14·
    1 day ago

    As long as Lemmy doesn’t support password-less and revokable logins, no one should be using alternative web frontends hosted by someone other than their trusted server admin. And as this issue clearly shows, when an admin hosts it themselves the dev can’t hide it and it is easy to circumvent.

    • Axolotl@feddit.itEnglish
      2·
      9 hours ago

      the dev can’t hide it and it is easy to circumvent.

      That’s if the Admin check the code of a famous and very used front end

      • poVoq@slrpnk.netEnglish
        1·
        9 hours ago

        Well, in the case the OP referenced, it was impossible to miss, but I grant that the other dynamic blocklist that was found after I made the above comment is more difficult to notice as an admin.

        That said, Tesseract was forked from Photon for very unclear reasons, and that alone was a big red flag from the start that made me personally stay away from it.

        • OpenStars@discuss.onlineEnglish
          1·
          2 hours ago

          There is a lot of misinformation about this issue.

          • The (first) list is NOT hard-coded, but toggleable via the toxicity filter setting - this is the exact polar opposite of “hard-coding”?
          • I am not sure about the second list, it definitely seems unmodifiable at least
          • the assertion that it bans all leftist instances seems (at best) specious when it also bans conservative and other instances, plus it does NOT ban some leftist ones

          Tbf, Tesseract reporting the message that it is “incompatible with that instance" seems valid criticism, that should definitely have been handled MUCH better.

          Whether the OOP might have been right or wrong remains highly unclear when most of their claims have ended up disproven, which just makes finding facts that much more difficult. That said, I do sympathize narrowly with the frustration they must have felt to have to have discovered the issue the way they did, with no announcement having been made.

          Thank you for offering greater precision here in your own responses.

  • OpenStars@discuss.onlineEnglish
    728·
    1 day ago

    When asked about this HARD-CODED feature that was barely announced when it was slipped into the code, the authors’ response was:

    If you dont like it, fork it. Stop bothering us about it, we will never fully remove the slur filter.

    - source

    Oh wait… my bad, that was Nutomic, speaking about Lemmy. Tbf, after a huge outcry, they did later relent.

    I guess software having ideology injected right into it is par for the course these days. It takes an ENORMOUS amount of effort to find any workable solutions otherwise. Compromise is an art, and all the purity testing among leftists works against it (which btw is why conservatism and fascism is winning across the globe right now, at least initially, bc they are willing to set aside their differences in order to achieve a common objective).

    Fwiw I would suggest removing this particular feature from a front-end UI alternative and leave such matters to the back-end. Being able to personally block instances is superb, and offering a visual guide to which ones that someone might want to block would also be great, but hard-coding a dedicated list is ineffective at best, especially when the instance admins simply remove it from the code as they run the software without it.

    • Cowbee [he/they]@lemmy.mlEnglish
      161·
      1 day ago

      You’re equating Lemmy’s now disabled slur-filter to a hard-coded censor list of queer and leftist instances and users. Are you aware of how that comes across?

    • FleetwoodLinux@lemmy.zipEnglish
      151·
      1 day ago

      Ah yes, a slur filter to prevent people from saying particularly messed up stuff, definitely the same as “I’ll block all of my enemies for everyone”

      • OpenStars@discuss.onlineEnglish
        33·
        17 hours ago

        I happen to disagree - they are not the same thing at all, especially given the time differential. It’s a good thing I never said that they were.

        There are similarities though. See if you can spot them. Hint: their both having been HARD-CODED is one of them…

        • FleetwoodLinux@lemmy.zipEnglish
          5·
          10 hours ago

          But OpenStars! You’re commenting using the HARD-CODED frontend that talks to the HARD-CODED backend!!!

          How disingenuous can you be?? People are clearly upset about the combination of it being a personal blocklist/vendetta/whatever and it being forced on all users, not solely the fact that omg it’s HARD-CODED. While some people may have disagreed with the slur filter it didn’t ultimately prevent interaction, enabled or not; unlike this.

          • OpenStars@discuss.onlineEnglish
            22·
            8 hours ago

            “it being forced on all users” and it being “HARD-CODED” are one and the same thing, that’s what hard-coding means.

            Hard-coding is the least transparent, least friendly, least documented, and least simple-to-understand form of putting something into computer program code. If instead there had been a variable like if(yes_filter_toxic_mode) then (implement these policies) then people would not be upset anywhere remotely close to the same degree, as when that feature is unable to be turned off (readily) due to its having been hard-coded.

            Edit: the accuracy of the OP has now been called into question - apparently this issue is NOT hard-coded after all (reportedly, though I have not confirmed this directly myself). People can still be upset that a “toxicity” filter exists, and how it was implemented, but the fact that it is not shadow-filtering without the express consent of the instance admin makes this a much different discussion than if that filter was forced (via hard-coding) upon people unawares. Consent makes all the difference in the world!!! (The jury is still out whether this rose to the level of “informed consent”, but at least that is a documentation issue, not like the dev acting with nefarious intentions as people have been claiming.)

            The facts truly do matter here, and it seems that unless someone is willing to do a deep dive into the code, everything said about this issue is pure speculation. Which so far I’ve only seen one person willing to do, which yielded this (from link I mentioned above):

            from src/lib/policies/system.ts:

            export async function updateSystemFilterPolicy() {  
    if (SBDisabled && get(userSettings).enableToxicMode) {  
        SYSTEM_POLICY.set(BLANK_SYSTEM_FILTER_POLICY)  
        return  
    }  
   // auto-update of blocklist continues below
            • FleetwoodLinux@lemmy.zipEnglish
              2·
              8 hours ago

              “it being forced on all users” and it being “HARD-CODED” are one and the same thing, that’s what hard-coding means.

              That’s what I was saying after I was being a little hyperbolic, I used a definition and your exaggerated capitalization of hard-coded interchangeably

              • OpenStars@discuss.onlineEnglish
                11·
                3 hours ago

                I have been known to be a little hyperbolic myself - it does get in the way of clarity, unfortunately. The social media style of communication lends itself more readily towards “slamming” one’s “opponent”, as opposed to calm rational dialogue to arrive at a truth.