Need to let loose a primal scream without collecting footnotes first? Have a sneer percolating in your system but not enough time/energy to make a whole post about it? Go forth and be mid: Welcome to the Stubsack, your first port of call for learning fresh Awful you’ll near-instantly regret.

Any awful.systems sub may be subsneered in this subthread, techtakes or no.

If your sneer seems higher quality than you thought, feel free to cut’n’paste it into its own post — there’s no quota for posting and the bar really isn’t that high.

The post Xitter web has spawned soo many “esoteric” right wing freaks, but there’s no appropriate sneer-space for them. I’m talking redscare-ish, reality challenged “culture critics” who write about everything but understand nothing. I’m talking about reply-guys who make the same 6 tweets about the same 3 subjects. They’re inescapable at this point, yet I don’t see them mocked (as much as they should be)

Like, there was one dude a while back who insisted that women couldn’t be surgeons because they didn’t believe in the moon or in stars? I think each and every one of these guys is uniquely fucked up and if I can’t escape them, I would love to sneer at them.

(Semi-obligatory thanks to @dgerard for starting this.)

  • BigMuffin69@awful.systemsEnglish
    7·
    4 days ago

    Remember how OAI claimed that O3 had displayed superhuman levels on the mega hard Frontier Math exam written by Fields Medalist? Funny/totally not fishy story haha. Turns out OAI had exclusive access to that test for months and funded its creation and refused to let the creators of test publicly acknowledge this until after OAI did their big stupid magic trick.

    From Subbarao Kambhampati via linkedIn:

    "𝐎𝐧 𝐭𝐡𝐞 𝐬𝐞𝐞𝐝𝐲 𝐨𝐩𝐭𝐢𝐜𝐬 𝐨𝐟 “𝑩𝒖𝒊𝒍𝒅𝒊𝒏𝒈 𝒂𝒏 𝑨𝑮𝑰 𝑴𝒐𝒂𝒕 𝒃𝒚 𝑪𝒐𝒓𝒓𝒂𝒍𝒍𝒊𝒏𝒈 𝑩𝒆𝒏𝒄𝒉𝒎𝒂𝒓𝒌 𝑪𝒓𝒆𝒂𝒕𝒐𝒓𝒔” hashtag#SundayHarangue. One of the big reasons for the increased volume of “𝐀𝐆𝐈 𝐓𝐨𝐦𝐨𝐫𝐫𝐨𝐰” hype has been o3’s performance on the “frontier math” benchmark–something that other models basically had no handle on.

    We are now being told (https://lnkd.in/gUaGKuAE) that this benchmark data may have been exclusively available (https://lnkd.in/g5E3tcse) to OpenAI since before o1–and that the benchmark creators were not allowed to disclose this *until after o3 *.

    That o3 does well on frontier math held-out set is impressive, no doubt, but the mental picture of “𝒐1/𝒐3 𝒘𝒆𝒓𝒆 𝒋𝒖𝒔𝒕 𝒃𝒆𝒊𝒏𝒈 𝒕𝒓𝒂𝒊𝒏𝒆𝒅 𝒐𝒏 𝒔𝒊𝒎𝒑𝒍𝒆 𝒎𝒂𝒕𝒉, 𝒂𝒏𝒅 𝒕𝒉𝒆𝒚 𝒃𝒐𝒐𝒕𝒔𝒕𝒓𝒂𝒑𝒑𝒆𝒅 𝒕𝒉𝒆𝒎𝒔𝒆𝒍𝒗𝒆𝒔 𝒕𝒐 𝒇𝒓𝒐𝒏𝒕𝒊𝒆𝒓 𝒎𝒂𝒕𝒉”–that the AGI tomorrow crowd seem to have–that 𝘖𝘱𝘦𝘯𝘈𝘐 𝘸𝘩𝘪𝘭𝘦 𝘯𝘰𝘵 𝘦𝘹𝘱𝘭𝘪𝘤𝘪𝘵𝘭𝘺 𝘤𝘭𝘢𝘪𝘮𝘪𝘯𝘨, 𝘤𝘦𝘳𝘵𝘢𝘪𝘯𝘭𝘺 𝘥𝘪𝘥𝘯’𝘵 𝘥𝘪𝘳𝘦𝘤𝘵𝘭𝘺 𝘤𝘰𝘯𝘵𝘳𝘢𝘥𝘪𝘤𝘵–is shattered by this. (I have, in fact, been grumbling to my students since o3 announcement that I don’t completely believe that OpenAI didn’t have access to the Olympiad/Frontier Math data before hand… )

    I do think o1/o3 are impressive technical achievements (see https://lnkd.in/gvVqmTG9 )

    𝑫𝒐𝒊𝒏𝒈 𝒘𝒆𝒍𝒍 𝒐𝒏 𝒉𝒂𝒓𝒅 𝒃𝒆𝒏𝒄𝒉𝒎𝒂𝒓𝒌𝒔 𝒕𝒉𝒂𝒕 𝒚𝒐𝒖 𝒉𝒂𝒅 𝒑𝒓𝒊𝒐𝒓 𝒂𝒄𝒄𝒆𝒔𝒔 𝒕𝒐 𝒊𝒔 𝒔𝒕𝒊𝒍𝒍 𝒊𝒎𝒑𝒓𝒆𝒔𝒔𝒊𝒗𝒆–𝒃𝒖𝒕 𝒅𝒐𝒆𝒔𝒏’𝒕 𝒒𝒖𝒊𝒕𝒆 𝒔𝒄𝒓𝒆𝒂𝒎 “𝑨𝑮𝑰 𝑻𝒐𝒎𝒐𝒓𝒓𝒐𝒘.”

    We all know that data contamination is an issue with LLMs and LRMs. We also know that reasoning claims need more careful vetting than “𝘸𝘦 𝘥𝘪𝘥𝘯’𝘵 𝘴𝘦𝘦 𝘵𝘩𝘢𝘵 𝘴𝘱𝘦𝘤𝘪𝘧𝘪𝘤 𝘱𝘳𝘰𝘣𝘭𝘦𝘮 𝘪𝘯𝘴𝘵𝘢𝘯𝘤𝘦 𝘥𝘶𝘳𝘪𝘯𝘨 𝘵𝘳𝘢𝘪𝘯𝘪𝘯𝘨” (see “In vs. Out of Distribution analyses are not that useful for understanding LLM reasoning capabilities” https://lnkd.in/gZ2wBM_F ).

    At the very least, this episode further argues for increased vigilance/skepticism on the part of AI research community in how they parse the benchmark claims put out commercial entities."

    Big stupid snake oil strikes again.

    • aio@awful.systemsEnglish
      1·
      4 days ago

      That o3 does well on frontier math held-out set is impressive, no doubt

      I think there is plenty of room for doubt still. elliotglazer on reddit writes:

      Epoch’s lead mathematician here. Yes, OAI funded this and has the dataset, which allowed them to evaluate o3 in-house. We haven’t yet independently verified their 25% claim. To do so, we’re currently developing a hold-out dataset and will be able to test their model without them having any prior exposure to these problems.

      My personal opinion is that OAI’s score is legit (i.e., they didn’t train on the dataset), and that they have no incentive to lie about internal benchmarking performances. However, we can’t vouch for them until our independent evaluation is complete.

      (emphasis mine). So there is good reason to doubt that the “held-out dataset” even exists.

  • bitofhope@awful.systemsEnglish
    6·
    4 days ago

    Starting to think we’re about at the point where you could make the best search engine on the market in these three easy steps:

    1. Search Wikipedia for whatever the user typed and show the top result first.
    2. Check if <search keyword> dot com, org, and net exist and show them in the order of popularity.
    3. End of page.
  • Sailor Sega Saturn@awful.systemsEnglish
    6·
    5 days ago

    I read about this gross Robo Anne Frank LLM by a company called “School AI”: Bluesky post (looks like via an activitypub bridge, but I can’t be bothered to find the canonical link), News Article, School AI’s website.

    Gee it sure is weird how all these digital clones the AI companies keep coming up with all have the exact same (lack of a) personality.

  • sc_griffith@awful.systemsEnglish
    4·
    4 days ago

    trying to write a thread about polytopia but my images won’t upload >:(. idk what i’m doing wrong, i’ve tried on both my desktop and my phone

    • self@awful.systemsEnglish
      5·
      4 days ago

      it should be fixed… again. for some reason our image cache keeps getting into a state where it either stops accepting uploads or stops accepting requests at all. I plan to upgrade us to the latest version soon, but it’ll unfortunately involve a little bit of downtime: to upgrade pict-rs to a new point release, you have to run the migrate command, but it only works for the previous release. we’re two releases behind, so I have to custom package the in-between release just to get us there.

      • sc_griffith@awful.systemsEnglish
        3·
        4 days ago

        i see! thanks for all your work <3. I think i’ll just write the thread after the upgrade, i got partially done and it started eating my images again so maybe this just isn’t the moment

        • self@awful.systemsEnglish
          3·
          4 days ago

          of course! re the images: uggh hell with it, I’m scheduling the maintenance and I’m gonna spend some time in the lead-up isolating a root cause for our breakage just in case the upgrade doesn’t fix it

  • Sailor Sega Saturn@awful.systemsEnglish
    4·
    5 days ago

    (oh no it’s politics)

    Trump’s new cryptocurrency scheme is surprisingly forthright about being a pump & dump:

    CIC Digital LLC, an affiliate of The Trump Organization, and Fight Fight Fight LLC collectively own 80% of the Trump Cards, subject to a 3-year unlocking schedule. CIC Digital LLC and Celebration Cards LLC, the owners of Fight Fight Fight LLC, will receive trading revenue derived from trading activities of Trump Meme Cards.

    Essentially according to their own website, they started by selling 20%* of the tokens to the public, and over the next few years will… sell another 80% of the tokens to the public. To the moon!

    * half of that they describe as “liquidity” instead of public distribution – whatever that means.

    • YourNetworkIsHaunted@awful.systemsEnglish
      3·
      5 days ago

      My gut says that liquidity in this context means “making sure that there are tokens available to purchase for initial buyers” or in other words listing them on the market instead of distributing them at initial purchase price.

  • blakestacey@awful.systemsOPEnglish
    17·
    6 days ago

    So, the Wikipedia article about “prompt engineering” is pretty terrible. First source: OpenAI. Second: a blog. Third: OpenAI. Fourth: OpenAI’s blog. ArXiv, arXiv, arXiv… 43 times. Hop on over to the Talk page, and we find this gem:

    It is sometimes necessary to make assumptions to write an article (see WP:MNA).

    Spoiler alert: that link doesn’t justify anything. It basically advises against going off on tangents: There’s no need to rehash the fact that evolution is a fact on every damn biology page. It does not say that Wikipedia should have an article on some creationist fantasy, like baraminology or flood geology, based entirely on creationist screeds that all cite each other.

  • o7___o7@awful.systems
    15·
    11 days ago

    Read a rumor that zuck’s marriage is falling apart, which scans.

    A second divorced man is about to hit the tower.

  • swlabr@awful.systemsEnglish
    14·
    10 days ago

    Did my regular check in of a q-pilled family member’s facebook page. Zuckerberg’s new fash turn is not being received well as he is being read as the worm that he is. i.e. they are still mad about the anti-vax fact checking.

    • imp_poe@awful.systemsEnglish
      9·
      8 days ago

      fucking hell. well i guess moving all my shit elsewhere has become my midweek project instead of my procrastinate-project.
      EDIT: [rereads and comprehends the tuta addendum] well. cool. i think i’m going to reschedule that “move everything” to after my “scream and cry into a pillow”. anyone have any other recs?

      • khalid_salad@awful.systemsEnglish
        7·
        7 days ago

        I am also in the “scream and cry” phase of this. However, I have already decided on the following replacements for non-email:

        1. Mullvad VPN
        2. Ente for photos and photo backup
          2a. their authenticator also slaps
        3. Filen for cloud storage[1]
        4. Bitwarden for password management
          4a. Keyguard is great if you are on Android. I am looking into[2] other (non CLI) Bitwarden-compatible password managers for Desktop I should not have suggested Keyguard, since I am not aware if it has been audited. It was probably a mistake for me to use it at all. Sorry everyone.

        edit: Should mention that I am also looking into a calendar replacement.

        edit2: maybe mailbox.org? They are recommended on PrivacyGuides.org.

        1: This is more like a stopgap for me until I confirm they are worth sticking with or find someone better. The 29.99€ 100GB lifetime storage may be worthwhile for this (though I cannot say whether I am confident they will be around in 5 years, this can at least serve as a short-term solution).
        2: lassitude

        • Mii@awful.systemsEnglish
          5·
          8 days ago

          Filen does photos too, so you could consolidate 2 and 3 into one. I went and bought 500 GB lifetime storage on their Black Friday offer. Let’s see if they stick around. So far the experience has been smooth.

          And if you mention mailbox.org, let me throw Posteo.com into the mix. Basically the same offer, but where mailbox presents itself a bit more business-like, Posteo seems to market itself towards individuals. I moved my mails there and are quite happy with it, but you have to roll encryption yourself. No auto solutions like Proton or Tuta.

          • khalid_salad@awful.systemsEnglish
            5·
            8 days ago

            Last time this came up, I considered Posteo but decided to hold off (I think it was actually you who suggested it ❤). There was a concern I had about deleting an account due to inactivity, but I think I just misunderstood or misremembered something on PrivacyGuides. This might be a good choice.

            As far as Filen for photos, I am reasonably confident that Ente will stick around (they seem to take sustainability seriously). Stuff like this worries me about Filen (emphasis mine):

            What else is planned for Black Friday this year?
            As every year, we’ve planned a special surprise to give back to our loyal users this Black Friday. Just like in previous years, we won’t be revealing any details just yet. All I can suggest is to check in with us around November 18th—you won’t want to miss it!

            Are lifetime plans stackable?
            Yes, we’d like to emphasize once again that all types of our plans can be easily combined. Subscription plans only expire based on their individual purchase dates.

          • khalid_salad@awful.systemsEnglish
            5·
            8 days ago

            Thanks! I actually plan on putting a couple of things on Mataroa:

            1. “Here is a list of shit you can do to make your computer suck less (would contain the above)”
            2. “Here is how to make your Pixel phone less frustrating in 2024 (basically, how to use user profiles to create a dumb phone without the massive security issues or headache of ‘oh, actually I do need that app occasionally’)”


            (Aside: if anyone can recommend a better blogging platform, I would appreciate it. I like Mataroa for its privacy focus, minimalism, and low cost-of-leaving. The only issue I have is lack of KaTeX/MathML/however-the-fuck-omg-why-is-TeX-still-trying-to-hurt-me-in-2025 support.)

            • self@awful.systemsEnglish
              4·
              8 days ago

              Aside: if anyone can recommend a better blogging platform, I would appreciate it.

              so I can’t promise better, but you’re eligible for an account on the awful.systems blogging platform if you want one. it has MathML! it has jank! you can modify the frontend by PRing into a repo!

              it hasn’t seen much use yet, but I was planning on posting a short story I’m writing there when it’s done. I can also say that as much as I don’t like WriteFreely on the web, its phone app is surprisingly ok

              • khalid_salad@awful.systemsEnglish
                4·
                7 days ago

                I have the outline of a write-up about about category errors and measuring proxies (e.g. reported sexual orientation). I think I am a poor writer but the only critique I’ve ever gotten was from a “colleague” who writes like he gets paid by the word1. I will consider sharing the post once it is finished. ❤

                1: one rejection on a collaboration with this guy was essentially “have you nerds heard of ‘brevity’ or ‘wit’?” — from computer scientists 😭

    • self@awful.systemsEnglish
      8·
      8 days ago

      proton’s ceo is, at this moment, trying to post through it on multiple official company accounts

      thank fuck I switched to tuta

      [I hold up a hand to my earpiece and make a horrified face]

      a second milkshake has hit the duck (cw: fetishized racism, you know the one)

      I’m waiting to hear back from tuta about what the fuck but this might be the reason why I hop providers again

      image description (but see cw)

      so somebody at tuta had the idea to write this post on their official account on Twitter:

      Don’t get cucked by Big Tech’s invasive policies. Keep your private data safe and secure with state of the art end-to-end encryption with Tutanota.

      and then the image attached to the post is a cuckold porn meme I started to describe here but won’t??? because describing the image put me in a really terrible mood??? it’s SFW but trust me you don’t want it. the Black men in the picture have had their faces covered by logos for Microsoft, Amazon, Google, Apple, and Facebook in that order. the white woman in the picture does not have her face covered. impact font bottom text: “YOUR PERSONAL DATA”

      • khalid_salad@awful.systemsEnglish
        8·
        8 days ago

        Here is their response:

        @mav Yes, this was bad. I did not take a screenshot back then (the post was deleted after 30 minutes), but I scrolled all my way through Mastodon to find our apology:

        https://mastodon.social/@Tutanota/108910936764865962

        https://mastodon.social/@Tutanota/108910937813834878

        This was a mistake, we apologized & we made sure that we would never post something similar again.

        At Tuta, we foster honesty, respect & diversity.

        We are here to fight for privacy & against Big Tech surveillance. We want everyone to get the respect they deserve.

        and this:

        @shalf We set up guidelines that all team members on social media duty need to adhere to. We also created a social media review group where we post & discuss every proposal before actually publishing. And it has worked fine ever since. :)

        Content of Apology

        Dear Privacy Fans,

        Last week, on Friday, a post was made on our social media profiles which goes against our core values as team members and as a company. This post was made in poor judgement, without stopping and examining the underlying racist and sexist problems posed by this meme template, and it does not represent the culture and environment of the Tutanota Team. After a period of approximately 30 minutes the offensive post was removed from all platforms and a “brief” apology was posted.

        This brief apology does not go far enough to address this mistake and we would like to update everyone on how we are working to address this issue internally and how we will make sure that this does not happen again. As a company we are working with all of our team members to foster a culture that respects diversity in all forms. This is both on a personal level between colleagues, but also at an institutional level by working to create structures which allow diversity to flourish.

        • self@awful.systemsEnglish
          3·
          8 days ago

          I mean… it’s an apology, I don’t know what I was expecting. this still feels like a bigger, redder flag than the one cop who called them a honeypot (and at the same time didn’t seem to know what tuta is) — is this really a service I feel safe recommending marginalized people use? probably not, they should use signal. is it even suitable for the “grandma & drug dealer” use case? that question’s a bit more difficult.

          could they really have said or done anything to fix this? shit, I don’t know. maybe I need to dig a lot more into who and what tuta actually is. I ran into one of their (former?) developers on mastodon and they seem to outwardly be marginalized and antifascist. if that’s what tuta’s composed of internally, then I’m a lot more able to trust them. until I do that checking though, I don’t think I’ll recommend tuta to anyone who might need it — the opsec risk of trusting your keypair to a company run by assholes is very high, especially in the current climate.

          • khalid_salad@awful.systemsEnglish
            5·
            8 days ago

            I agree. Currently I can’t recommend any email providers. Hopefully Posteo works out, but at this point I am strongly considering just starting my own competitor in the vein of Signal messenger. Basically “we think of email as a public good.”

            • self@awful.systemsEnglish
              3·
              8 days ago

              I’d love to work on something like that. have you checked out any of soatok’s work on federated key infrastructure? I can dig up some links if you haven’t and it seems interesting; I understand soatok is developing it with the possibility that it could be an enabling technology for federated end-to-end encrypted email in mind.

      • BlueMonday1984@awful.systemsEnglish
        6·
        8 days ago

        Shit’s so diseased its a sequel to Tuskegee. They cooked this cack with CIA-grade crack.

        (I have no rational way to respond to this - I’m just trying some Dracula Flow-type shit and hoping it lands.)

      • khalid_salad@awful.systemsEnglish
        4·
        8 days ago

        Well, this Andy dipshit gave an absolute dogshit apology on reddit.

        Snippets below:

        sophistry and bullshitting

        First, while the X post was not intended to be a political statement, I can understand how it can be interpreted as such, and it therefore should not have been made. While we will not prohibit all employees from expressing personal political opinions publicly, it is something I will personally avoid in the future. I lean left on some issues, and right on other issues, but it doesn’t serve our mission to publicly debate this. It should be obvious, but I will say that it is a false equivalence to say that agreeing with Republicans on one specific issue (antitrust enforcement to protect small companies) is equal to endorsing the entire Republican party platform.

        andy doesn't know his head from the DNC mascot

        Second, officially Proton must always be politically neutral, and while we may share facts and analysis, our policy going forward will be to share no opinions of a political nature. The line between facts, analysis, and opinions can be blurry at times, but we will seek to better clarify this over time through your feedback and input.

        The exception to these rules is on the topics of privacy, security, and freedom. These are necessarily political topics, where influencing public policy to defend these values, often requires engaging politically.

        The operations of Proton have always reflected our neutrality. For example, recently we refused pressure to deplatform both Palestinian student groups and Zionist student groups, not because we necessarily agreed with their views, but because we believe more strongly in their right to have their own views.

        • Soyweiser@awful.systemsEnglish
          2·
          6 days ago

          Going from, i will not make any more political statements to

          I lean left on some issues, and right on other issues

          Is funny, also which issues, come on be specific.

          • khalid_salad@awful.systemsEnglish
            3·
            6 days ago

            also which issues

            I know this one!

            1. lefty on weed or sex work or w/e
            2. but fuck identity politics and what is this ladder doing here
  • swlabr@awful.systemsEnglish
    13·
    8 days ago

    This is extremely tangential to the areas of sneer interest, but seeing as this is the only technology related community I am in, I’m putting it here.

    This song has been making the rounds on the charts/social media and I refuse to believe that it isn’t about the package management tool apt

  • froztbyte@awful.systemsEnglish
    13·
    8 days ago

    some of the first research science on promptfondlers and model-affine dipshits is starting to see the light of day and, in what will surprise probably 0% of our regulars, it confirms some things

    (I have grumped about their desire for outsourced thinking in the past myself)