Ivan Tanev

@xoofx That's not what Matt is saying. He's saying to delete duplicate knowledge, not unique knowledge. Also, his work setup deliberately creates docs (CONTEXT.md, ADRs) for the stuff that code cannot capture.

This is such a bad take. Docs -and comments- compress knowledge and should explain the *why*: decisions, constraints, trade-offs, assumptions, and behavior that code alone cannot express clearly. They are excellent context for driving LLMs more effectively. It is the same false binary as "comments are unnecessary because code should be self-explanatory." Yes, code should be clear. No, that does not make documentation useless. This mindset often feeds a toxic culture where insiders hoard context and dismiss everyone else with: "It's in the code, can't you read it?" They turn themselves into knowledge bottlenecks to preserve their influence. These are some of the worst behaviors to work with.

@kettanaito @Zepp1978 Example of the above: "Protagonist video-called Character Name. Their sister picked up on the third ring." A reader is more likely to remember the protagonist had a sister that was mentioned before than remember the sister's name.

@Zepp1978 One effective way of introducing such characters is through other characters. Not only does that solve "here's John, remember him cause you'll see him again in 200 pages", but it proves for a masterful way to build anticipation in the reader's mind.

I’m completely new at this, so it’s probably wrong, but something I’ve been telling myself is to trust your reader. That said, my cast is very small. Easy to do when not writing epic fantasy. So YMMV on that one.

@Noahpinion Saving this one for the near future.

"AI is a bubble" is looking less and less likely by the day

@garybernhardt fwiw, I made a second account where I only followed different painters and artists, only liked art, clicked not interested on anything politics, and the "for you" page is quite nice. 90%+ great stuff.

"For you" shows me huge amounts of engagement farming AI slop videos. I've "not interested" several hundred of them. No change. Is there a way to make this nonsense stop?

@garybernhardt @willtcarey When it gets something wrong, I try to stay away from guiding it like a person. Instead, I say what's to be done in imperative voice. A guide is necessary to persuade a human, and what we reach for because it mimics a human conversion, but is wasted on a machine.

@willtcarey That's basically what happened. It insisted that it was useful for two turns. Then I said something like "If we're already verifying against an HTTP root, and the mailto etc. links don't have HTTP protocol, then that excludes them right?" and suddenly it agreed.

Common refrain: "you're prompting it wrong". Here's something that GPT 5.5 xhigh just planned. It wants to special case "mail links" and "JavaScript links", but the first bullet excludes those. It planned the useless guard anyway. You can't prompt these logical errors away.

@kettanaito @fardarter * Raid 1, and 2 bay NAS, I meant for cheapest option.

@kettanaito @fardarter Raid 5/6 Synology NAS. Raid if you wanna go for the cheapest option

Got a bunch of stuff to backup. Any recommendations for a drive to use?

@kettanaito Can't wait for my kids to be old enough to watch this. Also Avatar: The Last Airbender.

Gravity Falls is a phenomenal show. One of the three pieces of media I rewatch every year just to remind myself of good storytelling and get inspired from the passion so many folks put in to make it possible. But it has one fatal flaw, sadly...

@mattpocockuk Good courses build up from mark making up, into simple shapes, into shape intersection, into construction, etc. It went: do 3 character designs per day for 30 minutes :D

@mattpocockuk It also didn't build up exercises in a good way, even though it referenced courses by proko and drawabox I had taken.

/teach now creates reusable components which get improved/developed as you learn more More token-efficient, and builds momentum Look at these sweet cubes:

@kettanaito @kentcdodds Same. AI has utility, but having to verify all behavior drastically reduces any possible utility, and no one should trust unsupervised AI to do absolutely anything of import.

@kentcdodds I have innate (and I want to believe healthy) distrust for AI. I don't trust it to make a fact check, why would I let it anywhere close to something personal?

If you don't have a personal AI assistant (à la OpenClaw/pi/hermes), what's stopping you? I'd you tried it before and stopped, why? List all the reasons.

@nullvoxpopuli This would probably work on me to fix a bug.

don't like that your favorite software has a bug? 😈 write a script to re-create that bug 1000 times / minute to boost your score in Sentry so that the devs will see your issue above all the others 🙃

@nullvoxpopuli I'd point to github.com/MinishLab/semb… as a thing, or a class of thing, that may stick around. Though, generally, you're correct.

And ai tooling especially is a waste of time since the next generation of models invalidates the need for any of them (on repeat, forever)

The problem with the "if it works who cares what the code looks like" mindset for agentic work is that it assumes the agent has a perfect understanding of "works." Realistically, things are underspecified, agents make bad assumptions, etc. To be fair, agents are pretty good at unit test coverage. They're pretty bad at designing human experiences (API, CLI flags, etc.), especially cohesive ones for future roadmap plans they may not have visibility into (unless your backlog is perfect and vision fully laid out, which I doubt). They're bad at knowing where performance matters and what type (CPU vs memory tradeoffs). They're bad at where compatibility matters and where it doesn't (and tend to err on the side of preserving it without further guidance). Etc. Unless you have this ALL specified, you can't possibly claim "it works" without taking a look and thinking about it.

@davis7 What does Effect add here? You're paying the full effect runtime cost to run a single promise, without augmenting the error channel in any way, so it's equivalent to a normal promise rejection. Removing Effect.runPromise() from that code is strictly better.

100% the effect http client is really good, and now that models are writing most of the code there's a lot of stuff like this that would've been annoying before Cloudflare is another great example of something too annoying before that I'm starting to heavily use

I think it really depends on your engagement with the art community. In different fandom or niche art circles, I see AI art being constantly used for, frankly, scamming people. Mostly in the form of taking commissions or creating patreons. The algorithm really rewards consistent posting, and most real artists cannot create a new piece every day, while that's mostly trivial for AI, even with advanced prompting. So you get a loop where the algorithm promotes new consistent artists that have a """good""" style and do not disclose AI. It is a real menace for small semi-pro and professional artists that rely on commissions. For small artists, the only way to build a following is through social media, and it's being crowded out by scammers. Back to my original point about your examples - if you label your stuff as AI, than I think it's mostly fine! Then at least the audience can make an informed choice that aligns with their morality about the issue. You can even make the argument of letting your limited artist resource focus on the most important parts. However, that argument only works if you're truthful about what you used.

I feel like the most cringe and bad examples of "ai artists" (like grifters on twitter calling themselves artists for generating stuff) are mostly not really doing anything anyone cares about tbh? Or at most they do slop that's spammed on social media, which is kinda bad but also predates AI so kinda whatever. Im not more offended by pixar fruit cuck dramas than i was by elsa getting pregnant with spiderman. Meanwhile use of ai gen seems mostly minimal to cut corners in some places, which maybe is bad but I could see a positive argument for it (let's human artists focus on more important shit). If you need 100 simple icons an ai can make maybe your artist gets to spend more time on cooler stuff. And then there are some heavy uses of AI that seem artistically valid, things like gossip goblin and angel engine come to mind. Those series have creative ideas and dialogue that are genuinely good and the AI really enhances a production that otherwise could not really be anything more than plain text due to economic limitations. My hope would be if AI gen gets more accepted that it's mostly used for indie productions and to increase the efficiency of human artists. And I feel like in that world you could get this effect where instead of artists being replaced the bar for art quality and demand for it just increases, so the market stabilizes in a way that doesn't negatively affect artists much. Similar to how ai coding agents might affect programmers short term but also increase the demand and quality bar for software which in turn balances the market again.

Regardless of my stance on the topic, it's beyond annoying how people who don't know how LLMs work are trying to anthropomorphize them and claim that there is no difference between them "learning" and humans learning and applying their knowledge. For some reason it triggers me even more than people who understand the difference, but just don't care as long as this tool is beneficial to them.

Leaving aside the intellectual theft angle for a moment, it's also a question of assigned value. There is a good reason that AI 'artists' lie about using AI - because they want recognition equal to someone who put in the effort to achieve mastery without putting that effort in. For creative fields, the scorn for anything AI (that is not explicitly labeled as such) is deserved. With your collage example, usually it's clear from the style that it's a collage, and you can appreciate the artistry on that level. Selecting the images yourself and arranging them in a way that gives new meaning. Meanwhile, with AI art, the output is usually explicitly prompted in a way that imitates genuine human output. There is a gray zone where people mix the two - some human art with eg generated backgrounds, or ai enhanced shading or something of the sort. That can be more defensible as being closer to collage art, though labeling probably should still be applied.

A funny counter I see here, if one is ok with their stuff also being used in AI training is it ok then? You can just hold the view AI training is allowed but other forms of piracy are not. I think in general a lot of pro-ai views here kinda boil down to the perception that taking all copyrighted materials out there to make an AI doesn't really feel the same as taking someone's art and reselling it, cuz there's a pretty big transformation process involved that also outputs novel stuff, and that also kinda can't realistically work otherwise and would be a shame to prevent existing given it's utility. AI seems to enable novel stuff being made (as in not direct copies), akin to some very complex and efficient collage machine. A collages seem intuitively like kind of an ok thing, if someone makes a collage of tons things it doesn't feel the same as if someone just took an existing thing and said it's theirs. I also feel like a lot of issues people truly have with AI image gen legally are more so related to the economics of it. If AI got super good and replaced artists with no piracy involved I think artist would be virtually exactly as mad at AI because it affects their jobs. Same way taxi drivers would probably hate robotaxis regardless of whether their algorithms were trained on taxi driver data or some other data, and they'd think legally society has to help/protect them somehow.

@realChakrawarti @kettanaito @joelhooks Thank you, that is genuinely encouraging to hear!

@VanTanev @kettanaito @joelhooks You are awesome! 👊🏼

if nobody sees your content does your grift really exist?