Programmer and sysadmin (DevOps?), wannabe polymath in tech, science and the mind. Neurodivergent, disabled, burned out, and close to throwing in the towel, but still liking ponies 🦄 and sometimes willing to discuss stuff.

  • 6 Posts
Joined 1 year ago
Cake day: June 26th, 2023


  • In 2023, Reddit decided to start charging exorbitant amounts for API access, making it non-viable for free 3rd party apps to access its content, citing things like AI crawlers “stealing” their (users’) content.

    In 2024, Google announced an agreement with Reddit to access the API, citing things like enhanced up to date search results. I don’t recall having seen whether they pay for it, or how much, but possibly they do.

    It would stand to reason, that if Reddit has managed to get a single dime for API access, and they keep thinking free access to their users’ content is “stealing”, then Reddit would be interested in making it as hard as possible to access the content without paying.

    Could be threat of punishment: “You give us exclusive crawl access, or we don’t crawl you”.

    That could’ve been part of the agreement: “You give us cheap/free API access, or we don’t crawl you”.

    Reddit tightening things down while trying to sell API access, just happens to benefit Google.

  • Client-side is not cloud.

    Yes, you can keep client-side reasonably secure. You can’t send the data for cloud processing and seriously expect much security or privacy… for now. Encrypt client-side and use cloud as storage… maybe; encryption algorithms also have a “best by” date.

    My point is:

    • “Cloud hosted” can not be fully “secure and private” right now.
    • “With cloud storage”, has a “best by” date.
    • “Not cloud”… well, is not cloud 🙂

    Letting anyone with the ability to switch the software without you noticing, anywhere near the keys controlling some Bitcoin funds, is a really bad idea.

  • Welcome to currency exchanges.

    If you want to buy a sandwich in Indian Rupees… you either find someone selling sandwiches for Indian Rupees, or you have to exchange them to whatever the seller will accept (USD? CAD? AUD? EUR?..)

    Yes, FOREX has some swings, it’s not for everyone. Bitcoin may swing more ore less than other currencies, depending on the day.

    Does the cheese-sandwich-maker have to look up the rate this instant and calculate a spot price for me?

    If you travel around, you’ll find countries where sellers do exactly that: they pull out a smartphone, check the spot price, add some margin, and tell your the price in USD for whatever you’re trying to buy.

    Alternatively, you can swipe a credit card that will do all of that automatically. BTW, there are BTC backed credit cards too.

    buy cheese from someone who takes BTC […] just kicks the can further down the road.

    That’s how money works: you kick it all the way around the street, over and over, changing from can, to stone, to ball, to… etc. Those who manage to start with a can and end up with a Lambo, win. Those who end up with a single grain of sand, lose.

  • cloud-hosted, secure, and private

    Until homeomorphic encryption becomes a thing, cloud can’t be secure or private.

    every exchange I’ve looked at holds the keys to your account

    Exchanges, are not wallets. You’re supposed to move the coins out of the exchange for safekeeping. If you can’t, then it’s not a crypto exchange, it’s an ETF peddler.

    how a user can actually buy bitcoin using Proton Wallet.

    Wallets, are not exchanges. They can link to exchanges, like Metamask does, but their core function is to hold your keys.

  • breaking things down for an audience understanding neither the technical nor artistic aspect…

    Not a reason to misrepresent things. Reminds me of the animistic fallacy, if they even understand what’s really going on themselves.

    As for text, I’ve seen the MS generator spit out decent text, at least in titles and logos, and some AI art with full legible sentences.

    Unless you start off training by feeding the model 3d data (say, voxels) alongside 2d projections

    Some time ago already, there was an SD fork with bounded box support, and a ChatGPT preprocessor prompt template to do the layout. Object permanence in this case is as simple as continuing with the lower layer once the upper one is finished, maintaining object continuity in the lower layer. It’s reasonable to expect this to go from bounded boxes, to freehand layers for each object. Since an LLM has been shown to be a good preprocessor to set the layout, some more integration between both, with object feedback from the SD to reduce the layer bounding box, would do wonders. Adding an opacity mask could be a bit harder, but sounds doable.

    I don’t see the need of much higher abstraction to address this issue. Rendering videos of translucent objects, might need it, though.

  • Some are very easy to spot, like a shadow of a character, that’s missing a limb on the shadow, or has different placement or pose. Illumination or parallel surfaces where they vary in shadowing without a reason, is also a dead giveaway. But the mist damning evidence is having one scene, then a slightly different scene in a reflection.

    There are reasons for human authors to do any of these on purpose, but unless that purpose is part of the work, they’re most likely AI mistakes.

    Of course it’s kind of funny how there is already a large overlap between the best AI art, and the most senseless “modern art”.

  • Yes, my comment applied more to photorealistic AI images.

    Illustrations are a different beast, where people have much more creative freedom… and that video is reasonably good at explaining that, but I find it falls short at some points:

    1. AI image generators don’t “consult” source images to generate an output. At training time, they extract patterns from the training set, which is never again used for generation, only the extracted patterns are.
    2. Modern AI generators are increasingly good at generating text. They still struggle a bit, but compared to a year ago, they can now generate headlines and large text correctly, while the mess gets shoved into smaller and less important text. This isn’t all that different from human artists adding “filler gibberish” text.
    3. Layers. While a naive (and cheaper) approach to AI generation doesn’t use layers, there are generators which do use layers, and can keep object consistency across obscured or cut-off sections.

    As AI generators advance, all these differences are likely to disappear… by following this same criticisms to fix things.

  • It’s not sbout monopolies, it’s about survival of Google’s core business model.

    All other browsers whose businesses are based on selling ads, face the same risk. They’re ALL between a rock and a hard place:

    • On one side, the EU and other countries want to push privacy laws that protect their citizens from getting casually spied on by foreign entities
    • On the other, Google’s core business model relies on spying on users and reselling the use of that data to the highest bidder… many of them being foreign entities to the targetted people

    If both Google/browsers/Ad sellers, and Ad purchasers, don’t come up with something that is tracking, but cuacks like privacy, the whole Ad ecosystem is at risk.

    FLoC is an attempt at compromise, by having an intermediary (the browser) who gathers full tracking data, but only sells a “reasonably anonymized” version.

    Of course Ad purchasers see that as an inferior product, so they aren’t keen to jump onto it… but if they all don’t get something like that going on, then everyone’s going to get shut down, with Google standing to lose the most.

    From the end user’s perspective, their failure would be slightly better, but otherwise worse than the current state of things:

    1. Less tracking on sites that didn’t rely on it in the first place
    2. More paywalls on sites that lose Ad revenue
    3. More sites asking people to enable full tracking in order to access their content

    IMHO, stuff like FLoC would be a better solution.