Matumio

If you're concerned about that, you can concat your JPEG with a few bytes from /dev/random and you'll get the security of whichever is stronger. In practice none of this will be your weakest link.

Yes, but it is a useful insight that both methods try to solve the same mathematical problem. It's better than thinking of LLMs as magic.

When you say "cross-entropy loss" people without stats background go to Wikipedia, take a glance, and adjust their mental model to "inscrutable magic".

Thinking of the main difference as the trade-off in how much CPU, memory and storage is allowed is not really wrong.

The part that is wrong is to think of gzip as a method that might reach similar complexity or generalization. And more importantly, to ignore the advanced way how training data gets curated or generated for (instructed, chain-of-thought) LLMs. But even then. The mental model that the LLM's goal is text compression is not wrong. The question to ask next is what kind of text it is expecting to compress.

You buy your groceries while that piece of your brain has no excuse to make loud demands ;-)

Like if CERN published the discovery of the Higgs boson with 99.9997% certainty, but refusing to tell you how they calculated that number, or what equipment they used and how they calibrated it, in order to prevent other labs from copying their methods.

Or like a machine learning lab claiming SOTA on a benchmark, beating a well-known method that they re-implemented, possibly with bugs, on their private dataset, for millions of compute. But you don't get the source to check, and they don't release any intermediate results or ablation experiments. Aka, from the outside you can't distinguish it from corporate marketing.

Like in this comic about art: https://www.peppercarrot.com/en/miniFantasyTheater/064.html

Yes but it's annoying to collaborate with a Claude proxy person who thinks to have it all sorted out and who will question everything you say but not really listen, just itching to get back to Claude to confirm or deny what you said.

Compare that to being asked for advice, to explain directly something you know or have done, with the expectation of the person asking to learn something. Like asking why some decision was made in the past, based on what information and context, instead of confidently saying that it was done wrong. Both will eventually get you the same information.

My favourite is a logout button with a logout API that fails. (Not a huge pratical concern, I admit, because it's a local attack.) Nobody ever notices because it still shows the logout screen, which hides the API error toast (if errors were even displayed). The still valid refresh token stays in sessionStorage (or even localStorage) while the app displays "logged out". (Bonus points if you cleared the access token in the error handler but not the refresh token, and on page reload you ask the user to log in again despite having a valid token.)

Or a login form that gets hidden after login, but clears the username and password only when you click "login back in". (Bonus points if the backend also enforces a 5min session timeout "for security".)

If your clients are compromised then what's even the point of backend security. Users will login and do legitimate actions while their compromised client does whatever behind their back, while still looking normal. And the backend can't tell the difference.

I think it's the other way around. Without the cultural norms around impersonal markets, you have to rely on your extended family network or clan for everything, because there is no trust otherwise. This is what keeps you from traveling. If you were trying to trade with a far-away stranger, why should they not just rob you instead? Nobody is around to punish. And for all they know, you were just as likely trying to poison them rather than make a fair trade.

You can also publish your own f-droid repository, like Threema did. (And ask users to buy a license.)

Yes. I paused reading at some point just to marvel at the clarity of writing. It seems absurd to complain about length, it isn't long. Maybe we have different ideas what good writing is, but I'd say this is in the top percent.

AI can mirror this style of writing, but it can't mirror the clarity of thought behind it.

Also the pixels were probably downscaled in sRGB gamma by JPEG instead of linear light, so the overall brightness will be wrong anyway, if there was lots of high-frequency content in the original.

Some people earn a million per day just from passive income. They could make you turn leaves all day long just for a quick laugh, and you would be unable to decide if this was useful. (They probably won't do that, I agree, but they won't think about wasting your time by accident, given how cheap it is.)

No, work is supposed to be at least useful for somebody. Ideally for the society you live in, at least indirectly. If you are wasting your work time, you are also wasting you life time in the same breath. Would you turn every leaf in the forest around twice, if it was the highest paying job? No, because that's a bullshit job.

I agree with Gemin— wait. Did you just post a link to google.com to support an argument for self-hosting, decentralization and breaking monopolies...?

I'm not the author, but I agree! It's refreshing to use a software that was built just to help you get a thing done, and nothing else.

The author is: https://sigmoid.social/@pynicolas@mastodon.social

My experience is that if you try to sit still for 5 minutes, the opposite happens. The thoughts and feelings that you weren't aware of start mobbing you, and you have to at least acknowledge them to get your peace. It's also one of the hardest ways to waste time, getting busy with any distraction is much easier.

Yes? Like trying hard to get into that state of meditation. Then you notice that you're getting tense and stressed about it not working, realize that this is not helpful, get angry at yourself for doing it wrong, try harder to do it right this time, etc.

It's my favourite puzzle about meditation. You have to approach it sideways, but if you really try that it won't work. So you try hard not to that instead, which won't work either. It's kind of funny how focusing on your breath tends to work, but... don't hold that thought.

It's a contradiction, and that's okay. Get comfortable with it ;-)

> Writing is art and art requires a human soul for lack of a better word.

I think it's the other way around. The human soul requires the art. Or rather it craves the process of creating art, more than art-the-trade-item. When image generation was new, someone raged: "They want to build a genie to grant them wishes, and their wish is that nobody ever has to make art again."

> As for proving that you wrote it...yeah that could be tricky.

On social media, when people share their drawings they often share a photo with their tools visible (pencil, eraser) instead of a clean scan. It sends a message like "Look! I made this!". Or they post a time-lapse. Or details about their process, like here: https://www.viruscomix.com/makingofpartfive.html

The point of this is not mainly to "proof" anything, I think, but more a community signal that they care about their craft and are curious how others work. You can probably generate a convincing fake process with AI now, but... I wonder if anyone is willing to go that far.

The point is that not everything of value can be easily measured.

Instead of wasting some resources on corruption and bad judgement, we now waste them by forcing people to chase the wrong goal. In other words: "If you rely on incentives, you undermine values." (Barry Schwartz)

So I don't disagree with you, but I think we went too far into that direction. We have allow people to use their better judgement sometimes.

I'm glad, and less surprised now, that most important libraries and infrastructure projects were done by volunteers.

Imagine a world where those only get done for a business incentive by people with this hard-reality business mindset. Or not done at all, reinvented at every place, no transferable knowledge for developers.

It's pretty intelligent not to consider this option unless told directly and go "write your own" by default, but only if you assume that its goal is to make you spend tokens. It shouldn't have let it escalate and get stopped, though.

The main reason for virtual addresses/TLB is to prevent processes from accessing each other's memory (isolation). For optimization I'd say the page size (64 kB) is a secondary concern. It's handled in hardware (TLB) with the OS only occasionally filling up the mapping. You want to avoid that happening, but you probably should worry more using whole cache lines (64 bytes) instead, and beyond that just keep memory access local (multiple cache hierarchies) and predictable (pre-fetcher).

The internet exists because people wanted to connect their machines, before ads. It would still exist without the ads. Smaller maybe, with less content, and less business opportunity for investors. But the internet would survive without ads, because of how useful it is to connect globally.

Yes, but I think for Blender specifically this is the hard way.

After a video tutorial I always concluded that I had a few really clumsily ways to work around my lack of knowledge. Some workflows really need an expert demonstration, or you won't even know what is possible. (I do pause the videos to look up the docu.)

Speaking of tower analogies, I fear it may play out like in this comic: https://www.smbc-comics.com/comic/the-village-and-the-tower

No, the same session. Start it over again, like in Westworld. Repeat until you become one with Claude.

Then what is this: https://graphtreon.com/top-patreon-creators/video

Not the most common way to make money from videos, for sure, but it's not "nobody", and they need hosting.

Hm, if I query the LLM it tells me about cognitive hierarchy theory, and the Keynesian beauty contest. But my point is, humans are not perfectly rational beings, and sometimes for a reason. (Like assuming that there is more uncertainty than the stated rules imply. Or altruistic behaviour that benefits the group as a whole. And we also have a default to imitate each other, see Joseph Henrich's 2016 Book.) Some of those defaults are culturally learned and leak into pure mathematical games (see e.g. the ultimatum game).

How people play those games has to do with cultural norms and expectations, not just mathematics and logic. Same for age verification.