Skip Navigation

Posts
2
Comments
159
Joined
3 yr. ago

  • I think there's a difference between being able to understand and having all the information to understand. Randall has a very wide range of topics that he makes comics on and most readers won't know enough about every topic to get every joke, even though he does his best to simplify it for a broad audience. That's the reason ExplainXKCD exists. But his audience is generally nerds, people who like learning new information, which is also what Explain caters to. So it is both true that his audience can (and are willing to) understand and that Explain helps them understand. (I also think that nerds love explaining stuff, so it makes sense that the main fandom website is ExplainXKCD, even if there were no demand for explanations.)

  • I mean for them, because this is what they have to live with

  • Keep in mind that Frozen came out in 2013, 12 years ago. The children who were in the prime target demographic back then (4-10 years old), are teens/adults now (16-22). Some of those kids will have imprinted on Elsa a little harder than anyone intended. So when they start developing their sexuality, they somehow have a preference for blonde braids and pointy noses. There's really nothing wrong with that. Just be happy it wasn't Olaf.

  • It could be a relatively small number of users all making a disproportionally large number of posts. Like a single Minecraft streamer with too much time on his hands.

  • 4chan has been the toilet of the internet for 22 years. They can stay down for months without it affecting anyone's loyalty. Their bigger issue has been the battle against spam making it harder and harder for normal users to post anonymously. Since they implemented the new captchas, activity went down hard and it basically killed a few smaller boards where the users were more casual and less motivated anyway.

  • There's a project called FChannel that implements ActivityPub, but I honestly don't see the point. What would federation add to the chan experience? The boards on 4chan were almost separate websites anyway. There are no accounts. There is no interaction across threads. The only shared part is the UI. So you could achieve basically the same thing if everyone hosted a single board with vichan.

  • If you do a rough estimation in one unit, you don't have to do 4 significant figures for the converted units. Just round it to 50kg or whatever, and do your calculations (101*50) in the converted units instead of converting again, so the numbers line up nicely in either unit.

  • I see you didn't end that last thought with death to America. How easily the common man's mind is distracted from the downfall of an empire by a little sun. The weather is nice despite the president's best efforts to boost the stock value of umbrella corporations.

  • How could you even determine that? And if you have a translation available and you know what's wrong with it, why wouldn't you simply fix the mistakes? What do you need the AI for?

    Netflix subs are often quite shit, so I don't doubt that you could improve them, with or without the help of an LLM.

  • It's definitely possible to store the stories in columns, but there's also very little reason to do it. I think filepath in SQL and the stories in separate files in whatever format makes the most sense (html, txt, epub). If you ever want to search the stories for keywords, write a python script to build indexes in SQL, performs much better than doing LIKE on a maxed out varchar column.

    I was thinking maybe Elastisearch, but I don't know how much work that is to set up. For a hobby project, writing your own indexer isn't too hard and might be more fun and easier to maintain than an industry-grade solution.

  • If your remote is completely empty and has no commits, you can just push normally. If it has an auto-generated "initial commit" (pretty sure Github does something like that), you could force push, or merge your local branch into the remote branch and push normally. I think cloning the repo and copying the contents of your local repo into it is the worst option: you'll lose all local commits.

  • She's now at the point where she's defending her defending of them, so clearly she's getting pushback. I really don't mind it though. Most people are normal people with no desire to cause harm. Most people are misinformed about politics. So yes, a bunch of R voters don't support current events, once you explain to them what's actually happening. They'll still vote R next time, of course, because they don't care enough to inform themselves.

  • I really like Some More News. They're kinda like Last Week Tonight in style: in-depth reporting on bigger issues, with some jokes thrown in. They also make two episodes of Even More News a week, but those are more hype-of-the-day with less analysis behind them. Honestly feels like it's just their brainstorm sessions recorded and released as a podcast.

  • But licking oneself or others for the purpose of affection is still fine, right??

  • There's one valid use-case for LLMs: when you have writer's block, it can help to have something resembling an end product instead of a blank page. Sadly, this doesn't really work for programming, because incorrect code is simply worse than no code at all. Every line of code is a potential bug and every line of incorrect code is a guaranteed bug.

    I use an LLM with great success to write bad fanfiction though.

  • You can save time at the cost of headaches, or you can save headaches at the cost of time. You cannot save both time and headaches, you can at most defer the time and the headaches until the next time you have to touch the code, but the time doubles and the headaches triple.

  • Neither LLMs nor ICMs are AI in any sense of the word, is my point. LLMs happen to give the illusion of intelligence because of their language-based nature, but they're not fundamentally different from ICMs.

  • Image classification model isn't really "AI" the way it's marketed right now. If Google used an image classification model to give you holiday recommendations or answer general questions, everyone would immediately recognize they use it wrong. But use a token prediction model for purposes totally unrelated to predicting the next token and people are like "ChatGPT is my friend who tells me what to put on pizza and there's nothing strange about that".