• 0 Posts
  • 197 Comments
Joined 1 year ago
cake
Cake day: March 14th, 2025

help-circle

  • I met my best friend in an online open-source game that was pretty popular at the time (so far as foss games go). We were both into making games of our own and got talking because of that

    He taught me the bits of programming I found particularly challenging, and we generally just chatted a lot and shared and played what we were making or came across online. He’s now a full-time game developer at a tiny company (so he does everything from coding to sysadmin to customer support) and I’m doing IT security consultancy where customer contact and an understanding of how networks and systems run is also core to the job. Kind of surprising how close they are, but also different enough that I’m sometimes surprised what he doesn’t know about security (not things he should have known necessarily, but still) and conversely I find the insights about game development still very interesting. Multiplayer lag hiding, as a random example. We’re a super good match. He lives about a day driving away so we see each other every few winter/summer holiday seasons since I was ~17, at which time we met up for the first time at the only LAN party ever organised for this game (where I had the honor of making the website and meeting all these friends I hitherto only chatted with, still probably the best weekend of my life). He lives in like the most beautiful mountainous region that never gets old to this flatlander. I’m so lucky to have a friend like this, never judging, always available, understands me exactly




  • That I’m still paying the rent amount from when we moved in nearly ten years ago. Comparable places cost twice as much when we last looked in 2024. We’d actually like to move to a more modern place but at these rates it’s extremely hard to justify a nicer and/or nearer place

    The landlord is super lazy, they do the bare bare minimum for maintenance and repairs so we just do the small things instead of trying to motivate him to do it or send someone, but also, bless this landlord lol!








  • I tested it and there’ll be months where I do 400 queries and months where there’s maybe 200. Since it doesn’t carry over, I’ll simply have to use alternatives on a regular basis. Especially across all devices (all my work is on a computer, most of the time with unfamiliar technologies as the customer has a new project for us to pentest), I’ll be out of queries most months. My employer is too scrooge to pay for our primary internal communication tool (we pay for various crap where it’s mandatory, even if we barely need it, but not for a donation-run hosted service that we heavily use). They’re never going to subsidise the searches I do during work time so long as competitors are “free”. So I’ll be stuck with other engines regardless, or have to pay as much as for a streaming service. It just seems like a lot, especially when part of it goes to Russia and burn-the-world AI bros


  • Before even being allowed to queue, security cared a lot that I present some email of the company I’m flying with. I had just sent the QR code to my phone, but the person doesn’t have a scanner (so what is this going to prove?) and so she requires seeing the email for the privilege of confiscating toiletries. Of course the airline later scanned it just fine but I’d not have gotten on the airplane that day if I had given my laptop away as checked luggage (didn’t have email on phone back then)

    Of course she didn’t more than glance at it, no identity card check to compare the name or anything. An invoice from my mom’s dentist would probably have done the trick. Balancing a laptop on a knee, trying to two-handed type a password that you only know as muscle memory, is the sort of reason they require you to plan two hours “just in case we come up with more bullshit than expected for you or a passenger ahead of you”

    And yes, my backpack with electronics needed to be flagged, bomb tested, and the last sip of water needed to be finished because the bottle was theoretically capable of holding an amount over the limit. Makes sense.



  • Yes? Who hasn’t?

    Changing (not just translating) the title is sometimes clearly necessary to preserve a joke or reference for people who don’t speak the original language (otherwise they’d be reading that version already), but in most cases I find it unnecessary and just makes it hard to talk to others about it or find more info because you know it under completely unrelated names. The Wikipedia article translations about the author will sometimes mention them in a recognisable way, but sometimes not even that helps…


  • If pitch made differences in meaning at the word level, it would be called a tonal language afaik

    English has pitch for other aspects of speech, such as indicating emphasis or what sentences are questions. Try saying “you did it” with an emphasis on the first word, without changing the volume or tempo: it’ll change pitch. If you raise the “it” instead, you get a question sentence (https://en.wikipedia.org/wiki/Rising_declarative). The pitch could make a difference for how you understand a word, like you’d not think the person said “what’s that cent?” (same set of base sounds but, due to the question pitch, it’s more likely “… scent?”), but it’s not a fixed property of the word like in a tonal language

    Variants of “mhm” are also tonal to me. I don’t know if this works the same in any English-speaking region but, where I’m from, “hm” with

    • rising pitch is “I don’t understand”
    • high-low-high is enthusiastic agreement
    • low pitch is like acknowledgment that you’ve heard the person (and don’t voice an objection), and
    • low-low with a pause between means “no”

    Not sure that’s considered a word, though

    Unrelated but this reminds me of a thing that happened earlier today while we were playing a board game. European languages aren’t considered click languages, but we still have them. One player asked a question of another, and someone from Turkey tsk’d. They made no other sound or head movement. Apparently this meant “no”! Native DE/NL speakers won’t understand that this is an answer to a question. We do use tsk here, just for other purposes. Was interesting to notice the culture difference!


  • Buying a Pixel to install GrapheneOS is not the only reasonable alternative to mainstream Android.

    Each person has their own threat model. My priorities will be very different from a journalist in China, a whistleblower, or even my own mom. Some are more at risk of ransomware and a device with root is risky. For myself, I’d incur that risk as a power user and would rather my phone can be backed up so I don’t have to worry about it breaking. GrapheneOS devs/followers believe that security is the only thing you must ever want, nothing else matters, and everything else is irrelevant, as though Google or street thugs will come to your house and exploit your bootloader if you don’t give them data voluntarily. They just don’t realistically question the statements the project makes, like what threats are worth mitigating (ubiquitous tracking? Who cares, I’ve got security patches and a locked bootloader!) and what options they’re giving up by using GrapheneOS that’s about as locked-down as the average Samsung phone. The marketing game is super strong because it’s hard to argue against at scale: everyone’s situation is different and “don’t you want to be safe?”


  • I remember reading something about LLMs not being able to learn “x is y” equivalence relations. Can’t find it now but limitations like this are what make differences clear between what humans do and what we’ve managed to teach the neural network (which will be used to iterate and improve the model further, of course)

    In the Chinese box analogy, this would be like them knowing cats are considered cute but not whether considered-cute animals include cats (if I remember the limitation type correctly). If you happen to slip the right instructions/questions, something they’ve seen before or something they’re capable of extrapolating, then nothing seems off; but if someone can say in one paragraph that cats are cute but they know of no cute animal, you’d not think they’re understanding what they’re saying, and so don’t really understand the language even if they give you plausible words in all other cases

    (For cats it’ll work because there’s a billion example sentences out there. LLM vendors are also trying to sidestep such problems by having it generate a bunch of tangential text (in which it might happen to regurgitate the tokens it needs to piece together the answer) before answering the prompt, but that’s still not being able to apply logic)