Cryptography nerd

Fediverse accounts;
Natanael@slrpnk.net (main)
Natanael@infosec.pub
Natanael@lemmy.zip

Lemmy moderation account: @TrustedThirdParty@infosec.pub - !crypto@infosec.pub

@Natanael_L@mastodon.social

Bluesky: natanael.bsky.social

  • 46 Posts
  • 462 Comments
Joined 6 months ago
cake
Cake day: January 18th, 2025

help-circle
  • There’s past evidence that the brain essentially outsources whole categories of knowledge and memories and skill to its surroundings.

    You might get good at certain things and learn certain things, somebody else learns something else, and then you both learn roughly what the other knows, at which point you rely on them for questions specific to what they know, and they rely on you for your specialty.

    We do this with technology too (it’s a big part of skills involving tools), and people has been doing it with dictionaries, online searches, etc.

    But doing it so universally for everything, just because chatgpt can form answer-shaped text for anything, is just insane. Don’t you even want to have your own personal feelings and thoughts? Do you just want to become an indirect interface to a bot for other people?

    It’s like the kind of personality-less people who mold themselves after popular people around them, but they’re doing it with an algorithm instead…










  • Natanael@infosec.pubtoScience Memes@mander.xyz4D Salmon
    link
    fedilink
    English
    arrow-up
    2
    ·
    3 days ago

    I can see a few ways to nudge people, like a puzzle that’s more linear than it looks like, or selecting the tasks in the next version of the encounter based on the events in the previous version of it to match (you wouldn’t see clearly what the “enemy” is doing so you wouldn’t notice that those actions the second round weren’t fixed, the game is just trying to replicate the interactions between past/future self and the rest of the stage is completely arbitrary). Basically treating it as a choreography problem.

    Maybe even make it feel more “wild” by recording how the player controls the character during the game before that point, use something like generative ML to make the “enemy” AI act as the game think you would respond to your past self, then when you’re the future you returning to the stage the game nudges you into using your abilities in the same way (altering how you have to use your powers to solve puzzles to match how the previous “predicted AI you”, making you face other enemies and use your abilities in similar ways to win fights)


  • Natanael@infosec.pubtoScience Memes@mander.xyz4D Salmon
    link
    fedilink
    English
    arrow-up
    3
    ·
    4 days ago

    I have some similar ideas about a story driven time travel game where it’s set up so that at one point you face an anonymous enemy, which forces you to trigger certain conditions in the stage and then they escape.

    Later in the game you return to this stage at the same point in time, expecting to face that enemy again, but you are this “enemy”, and the game nudges you into replicating the exact same sequence of events so that you take the actions which the “enemy” did so your (NPC) past self can replay the exact steps you previously took.

    But the game doesn’t show this clearly to you the player until you completed the stage - while your character is shown to have noticed something strange, the game doesn’t show you the player (or you character’s party) that you faced yourself until the end of the stage, using different positions and camera angles to hide it from you and the party. And if you manage to replicate the event chain perfectly, you’ll get to see your party members being visibly stunned when you see them realize it was a time loop (your actions in the stage breaks the loop), and hopefully the player can be made to feel like they experienced a time loop, as if they really faced off with themselves twice (“how did the game know I would do it that way?”)


  • Natanael@infosec.pubtoScience Memes@mander.xyz4D Salmon
    link
    fedilink
    English
    arrow-up
    2
    ·
    edit-2
    4 days ago

    In practical terms, Braid is probably closest. You have time rewind mechanics, in some stages it’s selective where rewind applies to specific objects AND/OR specific areas (so it’s not just try/retry, but actual time manipulation and setup)

    Shoutout to Superhot, where time moves when you move (bullettime on steroids)




  • Natanael@infosec.pubtoScience Memes@mander.xyzCursed
    link
    fedilink
    English
    arrow-up
    1
    ·
    5 days ago

    It’s kinda hilarious when the best formula only handles large numbers, not small. You’d think it would be the reverse, but sometimes it just isn’t (something about the law of large numbers making it easier to approximate good solution, in many cases)