(ℹ) Join us now at the IRC channel | ䷉ Find the plain text version at this address (HTTP) or in Gemini (how to use Gemini) with a full GemText version.
| *FANTOM has quit (Ping timeout: 120 seconds) | Oct 24 01:02 | |
| *FANTOM (~fantom@freenode-0m9.8g1.bp0598.IP) has joined #techbytes | Oct 24 02:09 | |
| *x-amarsh04 has quit (Quit: Konversation terminated!) | Oct 24 11:02 | |
| *x-amarsh04 (~amarsh04@6mnab3uw5dgtk.irc) has joined #techbytes | Oct 24 11:06 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has left #techbytes (Error from remote client) | Oct 24 11:26 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has joined #techbytes | Oct 24 11:33 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has left #techbytes (Error from remote client) | Oct 24 13:26 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has joined #techbytes | Oct 24 13:34 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has left #techbytes (Error from remote client) | Oct 24 13:46 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has joined #techbytes | Oct 24 13:49 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has left #techbytes (Error from remote client) | Oct 24 13:55 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has joined #techbytes | Oct 24 14:37 | |
| schestowitz[TR2] | https://www.thelayoff.com/t/1m5h1ex0s | Oct 24 15:32 |
|---|---|---|
| -TechBytesBot/#techbytes- ( status 403 @ https://www.thelayoff.com/t/1m5h1ex0s ) | Oct 24 15:32 | |
| schestowitz[TR2] | " | Oct 24 15:32 |
| schestowitz[TR2] | Bob says debt is only a problem if you plan to pay it back. | Oct 24 15:32 |
| schestowitz[TR2] | 20 hours ago by Anonymous | Oct 24 15:32 |
| schestowitz[TR2] | | 4 reactions (+4/-0) | Reply | Oct 24 15:32 |
| schestowitz[TR2] | Post ID: @gs+1m5h1ex0s | Oct 24 15:32 |
| schestowitz[TR2] | +4 | Oct 24 15:32 |
| schestowitz[TR2] | @Bm. You need to understand that IBM is all about costs. Cost lowering is job one if you are in management. Do layoffs impact costs? You bet they do, so layoffs are always in the picture. NOTE layoffs usually involve business operations that IBM is withdrawing from, OR offshoring. If you are in a high cost country, you should definitely expect layoffs. | Oct 24 15:32 |
| schestowitz[TR2] | 23 hours ago by Anonymous | Oct 24 15:32 |
| schestowitz[TR2] | | 6 reactions (+5/-1) | Reply | Oct 24 15:32 |
| schestowitz[TR2] | Post ID: @g8+1m5h1ex0s | Oct 24 15:32 |
| schestowitz[TR2] | +6 | Oct 24 15:32 |
| schestowitz[TR2] | @bm There's ALWAYS layoffs. | Oct 24 15:32 |
| schestowitz[TR2] | 1 day ago by Anonymous | Oct 24 15:32 |
| schestowitz[TR2] | | 6 reactions (+6/-0) | Reply | Oct 24 15:32 |
| schestowitz[TR2] | Post ID: @fr+1m5h1ex0s | Oct 24 15:32 |
| schestowitz[TR2] | +7 | Oct 24 15:32 |
| schestowitz[TR2] | @aa it's what he does best - stealing from and plundering the company and it's assets under false pretenses and acting all sanctimonious. He should burn in Hades for eternity. | Oct 24 15:32 |
| schestowitz[TR2] | 1 day ago by Anonymous | Oct 24 15:32 |
| schestowitz[TR2] | | 7 reactions (+7/-0) | Reply | Oct 24 15:32 |
| schestowitz[TR2] | Post ID: @d1+1m5h1ex0s | Oct 24 15:32 |
| schestowitz[TR2] | -7 | Oct 24 15:32 |
| schestowitz[TR2] | So no layoff? | Oct 24 15:32 |
| schestowitz[TR2] | 2 days ago by Anonymous | Oct 24 15:32 |
| schestowitz[TR2] | | 9 reactions (+1/-8) | Reply | Oct 24 15:32 |
| schestowitz[TR2] | Post ID: @bm+1m5h1ex0s | Oct 24 15:32 |
| schestowitz[TR2] | +15 | Oct 24 15:32 |
| schestowitz[TR2] | Hire clowns, get a circus. | Oct 24 15:32 |
| schestowitz[TR2] | 2 days ago by No surprise here | Oct 24 15:32 |
| schestowitz[TR2] | | 15 reactions (+15/-0) | Reply | Oct 24 15:32 |
| schestowitz[TR2] | Post ID: @bk+1m5h1ex0s | Oct 24 15:32 |
| schestowitz[TR2] | -9 | Oct 24 15:32 |
| schestowitz[TR2] | That’s it? That’s not to bad! | Oct 24 15:32 |
| schestowitz[TR2] | 2 days ago by Anonymous | Oct 24 15:32 |
| schestowitz[TR2] | | 11 reactions (+1/-10) | Reply | Oct 24 15:32 |
| schestowitz[TR2] | Post ID: @ba+1m5h1ex0s | Oct 24 15:32 |
| schestowitz[TR2] | +4 | Oct 24 15:32 |
| schestowitz[TR2] | I thought it would be more with them consent buying companies to keep up with the other big names that show profit and in the black not the red | Oct 24 15:32 |
| schestowitz[TR2] | 2 days ago by Anonymous | Oct 24 15:32 |
| schestowitz[TR2] | | 4 reactions (+4/-0) | Reply | Oct 24 15:32 |
| schestowitz[TR2] | Post ID: @as+1m5h1ex0s | Oct 24 15:32 |
| schestowitz[TR2] | +7 | Oct 24 15:32 |
| schestowitz[TR2] | "profit" is the administration paying a billion dollars in bailout to lay off the quantum staff ("Anderon") | Oct 24 15:32 |
| schestowitz[TR2] | 2 days ago by Anonymous | Oct 24 15:32 |
| schestowitz[TR2] | | 9 reactions (+8/-1) | Reply | Oct 24 15:32 |
| schestowitz[TR2] | Post ID: @ad+1m5h1ex0s | Oct 24 15:32 |
| schestowitz[TR2] | +11 | Oct 24 15:32 |
| schestowitz[TR2] | @OP Is that what they call profit at IBM these days ? | Oct 24 15:32 |
| schestowitz[TR2] | 2 days ago by Anonymous | Oct 24 15:32 |
| schestowitz[TR2] | | 11 reactions (+11/-0) | Reply | Oct 24 15:32 |
| schestowitz[TR2] | Post ID: @ab+1m5h1ex0s | Oct 24 15:32 |
| schestowitz[TR2] | +15 | Oct 24 15:32 |
| schestowitz[TR2] | Yet that F’ing POS will still take his $30M in incentives next year for the terrible job he’s doing. | Oct 24 15:32 |
| schestowitz[TR2] | " | Oct 24 15:32 |
| schestowitz[TR2] | https://www.thelayoff.com/t/1m5b291zw | Oct 24 15:41 |
| -TechBytesBot/#techbytes- ( status 403 @ https://www.thelayoff.com/t/1m5b291zw ) | Oct 24 15:41 | |
| schestowitz[TR2] | "I am also on one in the US. Just started last week. There is absolutely no way to complete it. One of the many items on the list is to fully implement a software feature on a new product. When my team did it on another product it took almost two years. I brought this up to my manager and he ignored it. I brought it up again in the next meeting and asked "Are you or my second level going to engage with my question about the scope a | Oct 24 15:41 |
| schestowitz[TR2] | nd timeframe of this item?" His response was "We discussed it when we agreed on the PIP." No engagement whatsoever, and it seems pretty obvious they are getting ready to fire me. | Oct 24 15:41 |
| schestowitz[TR2] | On another note, does anyone know (roughly) the timeframe from failed PIP to firing? Is it pretty immediate? | Oct 24 15:41 |
| schestowitz[TR2] | 14 hours ago by Anonymous | Oct 24 15:41 |
| schestowitz[TR2] | | no reactions | Reply | Oct 24 15:41 |
| schestowitz[TR2] | Post ID: @we+1m5b291zw | Oct 24 15:41 |
| schestowitz[TR2] | 0 | Oct 24 15:41 |
| schestowitz[TR2] | @tj this should be standard protocol. If you are on PIP it will be and endless cycle- repeating. Everyone hit with an PIP should drop everything and immediately update resume and start applying to every other job you can find. | Oct 24 15:41 |
| schestowitz[TR2] | 23 hours ago by Anonymous | Oct 24 15:41 |
| schestowitz[TR2] | | no reactions | Reply | Oct 24 15:41 |
| schestowitz[TR2] | Post ID: @tp+1m5b291zw | Oct 24 15:41 |
| schestowitz[TR2] | +2 | Oct 24 15:41 |
| schestowitz[TR2] | You were PIP'ed. Even with successful completion you will get a low rating for 2026. 2027 comes if they don't believe your performance has significantly improved, you're likely to end up on a PIP again. Honestly, complete the PIP if you can and make sure you are looking outside IBM. Leaving will be the only solution to get off the wheel of death. | Oct 24 15:41 |
| schestowitz[TR2] | 1 day ago by Anonymous | Oct 24 15:41 |
| schestowitz[TR2] | | 2 reactions (+2/-0) | Reply | Oct 24 15:41 |
| schestowitz[TR2] | Post ID: @tj+1m5b291zw | Oct 24 15:41 |
| schestowitz[TR2] | +2 | Oct 24 15:41 |
| schestowitz[TR2] | @qr once you are on pip you are marked for the rest of you career. If you are able survive it and hold on - it will still linger and haunt you 15 years later | Oct 24 15:41 |
| schestowitz[TR2] | 1 day ago by Anonymous | Oct 24 15:41 |
| schestowitz[TR2] | | 2 reactions (+2/-0) | Reply | Oct 24 15:41 |
| schestowitz[TR2] | Post ID: @r7+1m5b291zw | Oct 24 15:41 |
| schestowitz[TR2] | +2 | Oct 24 15:41 |
| schestowitz[TR2] | There are two kinds of PIPs. The more common is a budgetary issue, not a performance management issue. There’s little way out of these. If you find some HR policy or employment regulation to spare you, they’ll eventually find another way to eliminate you. | Oct 24 15:41 |
| schestowitz[TR2] | The other is an actual performance issue that seems remediable (and might even be caused by short-term things outside of work) In the latter case, HR and management like to say “this is a marathon, not a sprint.” That means that if you pass then you’ll face extra scrutiny for several years. | Oct 24 15:41 |
| schestowitz[TR2] | 1 day ago by Anonymous | Oct 24 15:41 |
| schestowitz[TR2] | | 2 reactions (+2/-0) | Reply | Oct 24 15:41 |
| schestowitz[TR2] | Post ID: @qr+1m5b291zw | Oct 24 15:41 |
| schestowitz[TR2] | +7 | Oct 24 15:41 |
| schestowitz[TR2] | @OP Get the heck out of the dump called IBM. You're running out of time. | Oct 24 15:41 |
| schestowitz[TR2] | 2 days ago by Anonymous | Oct 24 15:41 |
| schestowitz[TR2] | | 7 reactions (+7/-0) | Reply | Oct 24 15:41 |
| schestowitz[TR2] | Post ID: @hx+1m5b291zw | Oct 24 15:41 |
| schestowitz[TR2] | +6 | Oct 24 15:41 |
| schestowitz[TR2] | @OP Unfortunately, I am not able to advise as it is country specific. In my case (Europe), I was lucky enough to get "garden leave" shortly after being PIPed. Good luck to you. You will be much more happier outside of this toxic company and work environment. Trust me. | Oct 24 15:41 |
| schestowitz[TR2] | 3 days ago by Anonymous | Oct 24 15:41 |
| schestowitz[TR2] | | 6 reactions (+6/-0) | Reply | Oct 24 15:41 |
| schestowitz[TR2] | Post ID: @gh+1m5b291zw | Oct 24 15:41 |
| schestowitz[TR2] | +9 | Oct 24 15:41 |
| schestowitz[TR2] | @OP Op you must understand they're PIPing you to avoid having to RA you. They're going to claim to fire you for cause. | Oct 24 15:41 |
| schestowitz[TR2] | India found a cheat code to fire Americans, it's the PIP. Unless you can document that the PIP was impossible, applied unfairly or not universally, et cetera, they're firing you. Even if you can prove all of that, you're still going to be fired, and it would take months if not years of lawsuits to get any recovery. | Oct 24 15:41 |
| schestowitz[TR2] | Hope for the best, but assume the worst. Get your resume in order and find someplace better than IBM. Literally any other big tech company is. | Oct 24 15:41 |
| schestowitz[TR2] | " | Oct 24 15:41 |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has left #techbytes (Error from remote client) | Oct 24 18:03 | |
| schestowitz[TR2] | good coments https://forums.theregister.com/forum/all/2026/10/23/20265/?_gl=1*b98ah3*_ga*MTc5ODEy | Oct 24 18:04 |
| schestowitz[TR2] | OTIzNy4xNzc4MDgyODU0*_ga_JXW44Y23NM*czE3OTI4NjA1MTQkbzUyNiRnMCR0MTc5Mjg2MDUxNCRqNjAkbDAkaDA. | Oct 24 18:04 |
| -TechBytesBot/#techbytes-forums.theregister.com | Torvalds: AI is fine for competent devs, but spare us the 'drive-by' patches • The Register Forums | Oct 24 18:04 | |
| schestowitz[TR2] | saving | Oct 24 18:04 |
| schestowitz[TR2] | they will censor some | Oct 24 18:04 |
| schestowitz[TR2] | " Even if you like the taste, it's still slop | Oct 24 18:05 |
| schestowitz[TR2] | > "All that said, when it comes to programming, the ship has sailed." | Oct 24 18:05 |
| schestowitz[TR2] | The Titanic sailed too, with a lot of people on board - not all of whom survived. | Oct 24 18:05 |
| schestowitz[TR2] | > "AI is becoming part of the programming toolkit, whether its critics like it or not." | Oct 24 18:05 |
| schestowitz[TR2] | Ah, the "it's inevitable" argument. No, friend, no technology is inevitable, even if it's in some ways superior to what came before. It's always a choice that someone makes. But if you do voluntarily choose to adopt these technologies, and thereby fund and enable the people who control them - some of the most heinous people who have ever lived - then I hope you succumb to psychosis and get put away where you can no longer hurt anyo | Oct 24 18:05 |
| schestowitz[TR2] | ne else. | Oct 24 18:05 |
| schestowitz[TR2] | > "when someone who knows what they're doing, like Torvalds, uses AI, it can be very useful." | Oct 24 18:05 |
| schestowitz[TR2] | Ah, the "it's useful" argument. The most pitiful, amoral, ethically bankrupt, argument. Yes, "AI" can be " very useful", and so can asbestos, CFCs, lead, child labour, slavery, and genocide. | Oct 24 18:05 |
| schestowitz[TR2] | I don't think that's as strong an argument as you think it is. | Oct 24 18:05 |
| schestowitz[TR2] | > "Anyone who says otherwise is just making the same mistake the Luddites did." | Oct 24 18:05 |
| schestowitz[TR2] | Bullshit. The Luddites didn't claim that mechanisation wasn't useful. They claimed that the workers were being de-skilled and devalued so that the ownership class would be ever more enriched at their expense. **And they were right**. And yes, the adoption of GenAI for coding is the exact same playbook playing out again. Workers get deskilled, quality goes down, and more of the value is captured by the ownership class. | Oct 24 18:05 |
| schestowitz[TR2] | The Luddites weren't doomed by any mistake of their own, it was always a losing fight for them because governments have always sided with the avarice of the ownership class over the needs of workers. Luddites were raging against the machine, but not literal machines. | Oct 24 18:05 |
| schestowitz[TR2] | > "In short, get over it. I don't love it, either." | Oct 24 18:05 |
| schestowitz[TR2] | "Give up, don't fight, don't oppose oppression"!? What a ridiculous defeatist attitude. With this statement you've just outed yourself as a full-throated advocate for the AI industry, whatever mental acrobatics you may try to convince yourself otherwise. | Oct 24 18:05 |
| schestowitz[TR2] | > "AI ... AI ... AI ... AI" | Oct 24 18:05 |
| schestowitz[TR2] | But in the end, if you're arguing in favour of "AI" you've already lost the plot. "AI" is, and always has been, a marketing term. It's meaningless. Using it serves only to muddle thinking and launder massive amounts of societal harm. I doubt it's coincidental that the big information polluters push the term to apply to everything they can get away with. | Oct 24 18:05 |
| schestowitz[TR2] | It's like having a conversation about "nuclear", where people are discussing banning nuclear weapons. Then weapon industry sockpuppets come along crying "you can't ban nuclear! It's really useful for radiology!" | Oct 24 18:05 |
| schestowitz[TR2] | Minimise Comment | Oct 24 18:05 |
| schestowitz[TR2] | 30 | Oct 24 18:05 |
| schestowitz[TR2] | 25 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 1 day | Oct 24 18:05 |
| schestowitz[TR2] | AdaLoveseal | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Even if you like the taste, it's still slop | Oct 24 18:05 |
| schestowitz[TR2] | When this whole bloody thing is over, we'll finally have an answer to the everlasting question: How much does all that AI stuff actually cost? Even ignoring all the ethical arguments (which we probably shouldn't), there is still no economic model for SaaS (Slop-as-a-Service). I know how much a GPU is (which is twice as much as last month) and which models I can run on it, and *that* cannot contribute meaninfully to an OS kernel. It | Oct 24 18:05 |
| schestowitz[TR2] | takes frontier models. And those... those are so heavily subsidized that when the people who use them as the basis for their work have to pay full price, we'll see a lot more "coding by hand". Which we usually refer to as "programming", thank you very much. | Oct 24 18:05 |
| schestowitz[TR2] | 31 | Oct 24 18:05 |
| schestowitz[TR2] | 2 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 23 hrs | Oct 24 18:05 |
| schestowitz[TR2] | viennaGuy | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Even if you like the taste, it's still slop | Oct 24 18:05 |
| schestowitz[TR2] | It costs a lot right now, in terms of what it takes to power and cool a datacenter, in tokens and in the knock-on cost that all the memory and GPU is having on non-AI hardware. But I've a feeling that over time this will come down. It is hard to see it not becoming more effective when governments and shareholders are jointly getting nervous about the sink cost and resources. | Oct 24 18:05 |
| schestowitz[TR2] | As for slop, I'm also hoping and expecting the quality of AI output to get better over the same time frame. | Oct 24 18:05 |
| schestowitz[TR2] | Right now we're right at the embryo stage of AI/LLM and while we're getting some value, it is nowhere close to where it needs to be. But I'm accepting that it is here to stay and that the vast majority of us will somehow adapt and accept particularly when this starts being integrated in everything to the point where it is just one more algorithm that doesn't have a visible name. | Oct 24 18:05 |
| schestowitz[TR2] | Caveat: I rather like AI and I see the potential, and I use ChatGPT personally a lot, but I accept the maturity isn't there yet. | Oct 24 18:05 |
| schestowitz[TR2] | 4 | Oct 24 18:05 |
| schestowitz[TR2] | 8 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 12 hrs | Oct 24 18:05 |
| schestowitz[TR2] | FF22 | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Even if you like the taste, it's still slop | Oct 24 18:05 |
| schestowitz[TR2] | "It costs a lot right now, in terms of what it takes to power and cool a datacenter, in tokens and in the knock-on cost that all the memory and GPU is having on non-AI hardware. But I've a feeling that over time this will come down." | Oct 24 18:05 |
| schestowitz[TR2] | No, it will not. That's literally the only scenario that can't happen. It will especially not happen if AI is really the silver bullet solution for everything which AI-pushers try to purport it to be, then demand for it will explode, while infrastructure will be limited and growing only slowly - which will limit available capacities, and that will drive prices up. | Oct 24 18:05 |
| schestowitz[TR2] | 6 | Oct 24 18:05 |
| schestowitz[TR2] | 2 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 24 hrs | Oct 24 18:05 |
| schestowitz[TR2] | amb310 | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Even if you like the taste, it's still slop | Oct 24 18:05 |
| schestowitz[TR2] | THANK YOU! | Oct 24 18:05 |
| schestowitz[TR2] | The Luddites get slung around so much in the current AI conversation and it's almost always pro-AI people treating them like they were simpletons against technology and progress when it was nothing of the sort. They were against getting screwed over by the wealthy. | Oct 24 18:05 |
| schestowitz[TR2] | 20 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 19 hrs | Oct 24 18:05 |
| schestowitz[TR2] | doublelayer | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Even if you like the taste, it's still slop | Oct 24 18:05 |
| schestowitz[TR2] | I've said it before, so to avoid getting too repetitive, I'll keep it short. The actual Luddites weren't benevolent toward the poor. They wanted their well-paid specialist jobs to continue and not to go to much poorer less-skilled workers. Yes, some of their anger was directed at the wealthy factory owners. Some more was directed to the peasants they didn't care about in the slightest. They were neither idiots nor heroes. | Oct 24 18:05 |
| schestowitz[TR2] | 7 | Oct 24 18:05 |
| schestowitz[TR2] | 2 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 9 hrs | Oct 24 18:05 |
| schestowitz[TR2] | Dan 55 | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Even if you like the taste, it's still slop | Oct 24 18:05 |
| schestowitz[TR2] | Some more was directed to the peasants they didn't care about in the slightest. | Oct 24 18:05 |
| schestowitz[TR2] | [citation needed] | Oct 24 18:05 |
| schestowitz[TR2] | 4 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 23 hrs | Oct 24 18:05 |
| schestowitz[TR2] | DS999 | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Its only slop | Oct 24 18:05 |
| schestowitz[TR2] | If poor quality code is blindly accepted. | Oct 24 18:05 |
| schestowitz[TR2] | If the top maintainers (and especially Linus over them all) maintain the same code quality standards, and only accept code they understand then there will be no slop. Or rather to say, the slop will get rejected because it either won't meet the code quality standards or it would require too much effort to be "understood" and would be rejected on the basis that it would be too difficult to maintain in the future. | Oct 24 18:05 |
| schestowitz[TR2] | Unless you decide to claim that ALL code written by AI is slop, then I guess your claim will be a tautology. | Oct 24 18:05 |
| schestowitz[TR2] | 5 | Oct 24 18:05 |
| schestowitz[TR2] | 3 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 22 hrs | Oct 24 18:05 |
| schestowitz[TR2] | Lusty | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Its only slop | Oct 24 18:05 |
| schestowitz[TR2] | “basis that it would be too difficult to maintain in the future” | Oct 24 18:05 |
| schestowitz[TR2] | I don’t disagree with your point but have to pick you up on this. AI is exceptionally good at commenting and understanding code so maintainability is far, far less relevant than it used to be. You don’t have to use it to write code but to read and understand code you’ve either not seen for a while or never seen it’s hard to argue that it’s not a great tool. | Oct 24 18:05 |
| schestowitz[TR2] | 4 | Oct 24 18:05 |
| schestowitz[TR2] | 7 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 22 hrs | Oct 24 18:05 |
| schestowitz[TR2] | DS999 | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Its only slop | Oct 24 18:05 |
| schestowitz[TR2] | I was talking about the maintainers understanding it. | Oct 24 18:05 |
| schestowitz[TR2] | Yes AI is better than humans with comments. So sure you might understand WHAT is being done thanks to the comments, but WHY it is being done is what matters for being accepted by Linus and the other maintainers. Most software projects don't care so much about the why - if works to spec they don't care why it is doing things in an obtuse way. If it is faster than the previous code they don't care what specifically resulted i | Oct 24 18:05 |
| schestowitz[TR2] | n that improvement and why. Linus and the other top maintainers care very much about that sort of understanding - they would view that as more important than comments. | Oct 24 18:05 |
| schestowitz[TR2] | Comments can be added after the fact - i.e. "that code looks good, but beef up the comments for what is happening with all those variables and indirection in the inner loops and then its accepted". | Oct 24 18:05 |
| schestowitz[TR2] | 4 | Oct 24 18:05 |
| schestowitz[TR2] | 3 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 23 hrs | Oct 24 18:05 |
| schestowitz[TR2] | VoiceOfTruth | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Even if you like the taste, it's still slop | Oct 24 18:05 |
| schestowitz[TR2] | A lot of very pertinent points. | Oct 24 18:05 |
| schestowitz[TR2] | I have never accepted the "it's inevitable" argument. | Oct 24 18:05 |
| schestowitz[TR2] | 8 | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 18 hrs | Oct 24 18:05 |
| schestowitz[TR2] | Benegesserict Cumbersomberbatch | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Even if you like the taste, it's still slop | Oct 24 18:05 |
| schestowitz[TR2] | Workers get deskilled, quality goes down, and more of the value is captured by the ownership class. | Oct 24 18:05 |
| schestowitz[TR2] | While I agree with much of what you say, I'll take issue with this. | Oct 24 18:05 |
| schestowitz[TR2] | Torvalds' makes the point that quality isn't necessarily going down, quite the opposite. Grudgingly he's admitting that he has to take note of the effects of this technology because it's obliging him to make better quality code. | Oct 24 18:05 |
| schestowitz[TR2] | As for the worker deskilling argument, there are people of Torvalds' calibre who can tell whether what is being produced is quality code or not, and he seems to be advocating the use of this technology to inspire individuals to think the same way as he does. If you find you can't, you probably shouldn't be in the game in any case, you won't be for very long. | Oct 24 18:05 |
| schestowitz[TR2] | The analogy from the history of mechanisation of manual labour is that mechanisation actually increased the quality of the end product, because the product was more consistent. Analogies with the typo/thinko/brain fart that slips through even the most diligent human quality control. | Oct 24 18:05 |
| schestowitz[TR2] | The value of the tool is the use to which it's put in the hands of a skilled worker. Micros~1 might be at risk of having its product line clean-room reverse engineered en bloc by a bunch of open source agents, but Linux and the FOSS industry in general are good, AFAICT. | Oct 24 18:05 |
| schestowitz[TR2] | 2 | Oct 24 18:05 |
| schestowitz[TR2] | 2 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 6 hrs | Oct 24 18:05 |
| schestowitz[TR2] | Anonymous Coward | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Anonymous Coward | Oct 24 18:05 |
| schestowitz[TR2] | Re: Even if you like the taste, it's still slop | Oct 24 18:05 |
| schestowitz[TR2] | Using AI to code effectively requires a set of skills many programmer's do not have or have little experience of. | Oct 24 18:05 |
| schestowitz[TR2] | You need to be effective at | Oct 24 18:05 |
| schestowitz[TR2] | - defining a set of detailed requirements | Oct 24 18:05 |
| schestowitz[TR2] | - defining the acceptance criteria for each requirement | Oct 24 18:05 |
| schestowitz[TR2] | - breaking the requirements into a series of tasks that can be validated | Oct 24 18:05 |
| schestowitz[TR2] | - be able to pay attention to detail when reviewing | Oct 24 18:05 |
| schestowitz[TR2] | The LLM is only part of the toolset that you use. | Oct 24 18:05 |
| schestowitz[TR2] | You need to look at the rest of the tools around the LLM. | Oct 24 18:05 |
| schestowitz[TR2] | Some people lack curiosity and thus fall at the first hurdle. | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 1 hr | Oct 24 18:05 |
| schestowitz[TR2] | Anonymous Coward | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Anonymous Coward | Oct 24 18:05 |
| schestowitz[TR2] | Re: Even if you like the taste, it's still slop | Oct 24 18:05 |
| schestowitz[TR2] | This guy definitely invested heavily in Betamax and HD-DVD back in the day. | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 1 hr | Oct 24 18:05 |
| schestowitz[TR2] | Anonymous Coward | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Anonymous Coward | Oct 24 18:05 |
| schestowitz[TR2] | Re: Even if you like the taste, it's still slop | Oct 24 18:05 |
| schestowitz[TR2] | "Workers get deskilled, quality goes down" | Oct 24 18:05 |
| schestowitz[TR2] | Only if you buy the industrially cranked out crap. | Oct 24 18:05 |
| schestowitz[TR2] | What it really means is that more people get access to stuff that was previously out of reach because of price...that's what has people like you rattled. | Oct 24 18:05 |
| schestowitz[TR2] | Being able to industrially crank out flat pack furniture did not kill off the handmade furniture market...you can still buy furniture made by hand...similarly, if I want a handmade sausage roll, I can go to my local baker...or if I haven't got the money, I can go to Greggs and get a mechanically cranked out one. | Oct 24 18:05 |
| schestowitz[TR2] | Thing is, if people can afford something, they keep buying what they can afford...you're not going to stop someone who can easily afford Dom Perignon to stop drinking Dom Perignon because a 24 pack of Fosters is on offer for £6.99. | Oct 24 18:05 |
| schestowitz[TR2] | What will actually happen is the cost of bootstrapping a small business will go through the floor, anyone will be able to spin up a simple small business in an afternoon, test it for a while, then wrap it up if it doesn't work...previously, it would take weeks or even months and considerable cost...if you could even find a way to do it in the first place. | Oct 24 18:05 |
| schestowitz[TR2] | Competition across many industries will increase several times over...which is why the rich are brainwashing you to hate AI...they don't like competition. | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 1 day | Oct 24 18:05 |
| schestowitz[TR2] | Mostly Irrelevant | Oct 24 18:05 |
| schestowitz[TR2] | Bronze badge | Oct 24 18:05 |
| schestowitz[TR2] | You can just see Linus' temperature riding, how long will it be before another major blow up? | Oct 24 18:05 |
| schestowitz[TR2] | 7 | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 1 day | Oct 24 18:05 |
| schestowitz[TR2] | Blazde | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Linux | Oct 24 18:05 |
| schestowitz[TR2] | Linus | Oct 24 18:05 |
| schestowitz[TR2] | For all his widely-acknowledged abrasiveness, and somewhat in defiance of age-cliches, Linus has really impressed me with the nuance of his recent embrace of both Rust and AI/vibe-coding. I don't think you have to necessarily 'like' either to accept that he recognises the inevitability of both, and that fighting them would only harm Linux. Read one of his rants and it's easy to think he's lost it and gone too far in any one directi | Oct 24 18:05 |
| schestowitz[TR2] | on, he's always way too black and white (and uptight?), but collectively there's a kind of mature herding in what he says, strategically focused on the future, surprisingly positive and agile (in the non-loaded dictionary sense). | Oct 24 18:05 |
| schestowitz[TR2] | I still fondly remember my first encounter with Linux, a 'shareware' footnote on an extremely packed "Windows & DOS" '96 cover disk: https://dn721604.ca.archive.org/0/items/pcw-may-1996-linux-ukuug/pcwmay96%20%20cd.jpg | Oct 24 18:05 |
| schestowitz[TR2] | Even Descent 2 (one of my favourite ever games) took a backseat when I realised what Linux was. Bit wild how far it's come. | Oct 24 18:05 |
| schestowitz[TR2] | (Bit wild how far Duke Nukem has come too, now I think about it) | Oct 24 18:05 |
| schestowitz[TR2] | 17 | Oct 24 18:05 |
| schestowitz[TR2] | 6 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 9 hrs | Oct 24 18:05 |
| schestowitz[TR2] | Dan 55 | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Linus | Oct 24 18:05 |
| schestowitz[TR2] | The "inevitability" of AI will last until the providers have to start charging customers the real cost + a margin. Then suddenly everyone will wonder why they thought it was inevitable. | Oct 24 18:05 |
| schestowitz[TR2] | 2 | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 1 day | Oct 24 18:05 |
| schestowitz[TR2] | Throg | Oct 24 18:05 |
| schestowitz[TR2] | Bronze badge | Oct 24 18:05 |
| schestowitz[TR2] | Every fad… | Oct 24 18:05 |
| schestowitz[TR2] | …is driven by the idea that it’s “inevitable”. | Oct 24 18:05 |
| schestowitz[TR2] | I seem to remember El Reg posting a report about a study saying that developers are fooling themselves if they think that using “AI” is making them more productive. Oh. Wait. Here it is… | Oct 24 18:05 |
| schestowitz[TR2] | https://www.theregister.com/software/2025/07/11/ai-coding-tools-make-developers-slower-study-finds/1143832 | Oct 24 18:05 |
| schestowitz[TR2] | It’s worth remembering as ever that the goal of professional software engineering isn’t “code complete”. It’s “shipped and being used reliably by customers”. | Oct 24 18:05 |
| schestowitz[TR2] | 15 | Oct 24 18:05 |
| schestowitz[TR2] | 2 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 20 hrs | Oct 24 18:05 |
| schestowitz[TR2] | Richard 12 | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Terminator | Oct 24 18:05 |
| schestowitz[TR2] | Re: Every fad… | Oct 24 18:05 |
| schestowitz[TR2] | The bottleneck hasn't been "writing the code" since the very first paper tape or punched card, if not earlier. | Oct 24 18:05 |
| schestowitz[TR2] | It's always been defining the problem, and verifying that the proposed solution actually solves it. | Oct 24 18:05 |
| schestowitz[TR2] | LLMs spewing code actually makes the job of test and verification more difficult. Everyone knows that "LOC" is a stupid thing to measure, yet it's literally how LLMs are paid for. | Oct 24 18:05 |
| schestowitz[TR2] | On the other claw, LLMs may have value in checking and testing - fuzzing, spotting possible typos and similar. | Oct 24 18:05 |
| schestowitz[TR2] | The frontier models are quite simply utterly useless though, because they cost way too much. No real company would spend $15 million renting an LLM to partially test one product (or to hack Hugging Face), and that's the kind of money it costs for the frontier models to perform that task - then add finance, staffing, and training costs. | Oct 24 18:05 |
| schestowitz[TR2] | Far smaller sums hire a lot of skilled humans, and runs a smaller model almost indefinitely. | Oct 24 18:05 |
| schestowitz[TR2] | 8 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 7 hrs | Oct 24 18:05 |
| schestowitz[TR2] | MrXonTR | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Every fad… | Oct 24 18:05 |
| schestowitz[TR2] | I have upvoted your post for largely making sense. However I must also criticise it because I too thought chatbots could only output blindly. | Oct 24 18:05 |
| schestowitz[TR2] | The modern agentic workflow is improving rapidly, and the junior devs along with it. They are very much benefiting from analysis being done behind the scenes. The bots aren't simply screaming code into the void, rather they are reading existing code and making tests within existing systems. It's this integration which produces higher quality code. The devs are then spared the uphill struggle of learning how everything works. Their | Oct 24 18:05 |
| schestowitz[TR2] | incompetency period is lesser now IMO. | Oct 24 18:05 |
| schestowitz[TR2] | There is always a bottleneck somewhere and I wonder what it will be next? | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 1 hr | Oct 24 18:05 |
| schestowitz[TR2] | Filippo | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Every fad… | Oct 24 18:05 |
| schestowitz[TR2] | >The modern agentic workflow is improving rapidly, and the junior devs along with it. | Oct 24 18:05 |
| schestowitz[TR2] | But are they really? I've been trying Copilot in agent mode for a while, and I'm not at all sure it's saving me any time. The last thing I asked of it was to build a full suite of unit tests for a class, aiming for 100% coverage. This is a class with a very simple interface, and a few private methods which handle some branching logic. Nothing too fancy, and writing unit tests is boring. | Oct 24 18:05 |
| schestowitz[TR2] | Copilot spent a significant amount of time fooling around, and eventually produced a chunk of unit tests that made just a few calls to the interfaces - and used reflection to get at the private methods and run them directly. I'm not going into why this is bad right now, but it's pretty bad, even while technically producing pretty green lights. | Oct 24 18:05 |
| schestowitz[TR2] | I'm pretty sure that your average junior dev would either not know why you don't do unit tests that way, or not care because they haven't yet spent enough time cleaning out the unholy messes that eventually result from working that way. Tests are green, branches are covered, and who cares if it's all nearly meaningless. | Oct 24 18:05 |
| schestowitz[TR2] | I've redone the tests, and now they're meaningful, they are half as long, far more readable, and they let me figure out which branches I can safely remove. It took longer, but if it saves me just one time from a screwup that Copilot's tests would never have detected, or if it lets me tinker freely with the private logic without having to faff around with the unit tests, I'll be far ahead. | Oct 24 18:05 |
| schestowitz[TR2] | Another time, I asked it to refactor a particular pattern across a whole solution. It did it. Eventually. I would've searched for the pattern and then fixed it by hand, but Copilot decided to try its hand at writing a script to fix it massively; it eventually succeded, but it took like half an hour. I had like 10 instances of the pattern. I let it run to see what happened, but I could've done it by hand in half the time. | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 1 day | Oct 24 18:05 |
| schestowitz[TR2] | elsergiovolador | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Frontier | Oct 24 18:05 |
| schestowitz[TR2] | dev> assume you are me. Lazy, but somewhat bright developer. Look at this next ticket. | Oct 24 18:05 |
| schestowitz[TR2] | ai> fuck me, I am going for a walk. | Oct 24 18:05 |
| schestowitz[TR2] | 4 | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 1 day | Oct 24 18:05 |
| schestowitz[TR2] | Dan 55 | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Not convinced | Oct 24 18:05 |
| schestowitz[TR2] | Linus' change which furtled around with Read-Copy-Update had a race condition in it and it turned out introduced a performance hit. Sounds like he was fooled by the chatbot which will write up impressive-looking docs and tests but the code quality is just not that good. And then he left it to someone else to fix and went on holiday. | Oct 24 18:05 |
| schestowitz[TR2] | Then there's the attitude towards new developers. Seems it's basically git gud or get a good LLM. Then also they say they're snowed under with work and there's there's insufficient feedback on patches. You wonder why. | Oct 24 18:05 |
| schestowitz[TR2] | I think like many companies they realised they had a problem - technical debt and not enough documentation or training or new people coming in. Then they suddenly they think the chatbot's going to magically fix the problems which have been building up over years but it does the opposite - the tech debt increases, the documentation is unreadable, instead of training you hand your thinking over to the chatbot so you're not even learn | Oct 24 18:05 |
| schestowitz[TR2] | ing on the job, and people are put off contributing to the kernel. | Oct 24 18:05 |
| schestowitz[TR2] | 12 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 1 day | Oct 24 18:05 |
| schestowitz[TR2] | TonyHoyle | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Not convinced | Oct 24 18:05 |
| schestowitz[TR2] | Yes he vibe coded a fix then left others to deal with.. which created more work for other people to fix and would have been quicker to do properly in the first place. | Oct 24 18:05 |
| schestowitz[TR2] | And based on this goes all in on support for LLMs. He's lost it. | Oct 24 18:05 |
| schestowitz[TR2] | 8 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 1 day | Oct 24 18:05 |
| schestowitz[TR2] | Blazde | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Not convinced | Oct 24 18:05 |
| schestowitz[TR2] | Totally taking a complex issue out of context. Several others had tried to manually code a solution before without success, he was upfront about using an LLM to try it, described limitations and caveats, and asked people to critique it. No big deal, and no different to any non-LLM solution. | Oct 24 18:05 |
| schestowitz[TR2] | 6 | Oct 24 18:05 |
| schestowitz[TR2] | 4 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 23 hrs | Oct 24 18:05 |
| schestowitz[TR2] | Dan 55 | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Not convinced | Oct 24 18:05 |
| schestowitz[TR2] | Perhaps what should be done given RCU and its history is formally specify it using a language like TLA+ then code from the formal specification. If people have tried and failed over years then an LLM won't have any successful training data it can use. | Oct 24 18:05 |
| schestowitz[TR2] | Asking people to critique something generated by LLM while on holiday is a novel development method I haven't seen until now. | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | 2 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 22 hrs | Oct 24 18:05 |
| schestowitz[TR2] | Blazde | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Not convinced | Oct 24 18:05 |
| schestowitz[TR2] | an LLM won't have any successful training data it can use | Oct 24 18:05 |
| schestowitz[TR2] | (Depending which one he actually used) this isn't a limitation any more. It's a good 18+ months since they moved beyond merely cobbling together previous patterns from Stack Overflow. | Oct 24 18:05 |
| schestowitz[TR2] | It would probably help if he pointed the agent at the previous failures, and described potential pitfalls, but overall they're plenty capable of reasoning their way to creative solutions. | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | 4 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 9 hrs | Oct 24 18:05 |
| schestowitz[TR2] | drankinatty | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Not convinced | Oct 24 18:05 |
| schestowitz[TR2] | You've hit upon a good point, and it's worth fleshing out a bit. When you turn your coding over to a model, coding skill deterioration is real, and it happens quickly. | Oct 24 18:05 |
| schestowitz[TR2] | When I say "coding skill" I mean the lump-sum total of all the finely tuned and usable information you hold in your mind, the ability to use the language itself, across the breadth of the features you use on a regular basis, your ability and understanding of the language standard (if applicable), what it requires, what it permits and what it leaves to the compiler writers to implement, and the wisdom to know the difference. | Oct 24 18:05 |
| schestowitz[TR2] | If you play golf, your tennis game suffers. | Oct 24 18:05 |
| schestowitz[TR2] | For those that truly know and have mastered their chosen language, while deterioration still applies, it's more like riding a bike. Hopping back on brings the information back to the forefront in quick order. It's not instant and still takes work, but you know exactly where to look to find whatever detail you are in need of. | Oct 24 18:05 |
| schestowitz[TR2] | The deterioration is much more acute for those that switch to model coding before mastering the language. The model becomes an alternative to putting in the "butt in chair time" with a book open in your lap and the keyboard in front of you slowly appreciating the nuances of whatever language you've chosen and how to get the most out of whatever hardware you are working with. There is no hopping back on at that point and simply | Oct 24 18:05 |
| schestowitz[TR2] | riding away. For those beginning to code, the prospect of ever having the opportunity to reach that level in programming is all but foreclosed. | Oct 24 18:05 |
| schestowitz[TR2] | LLMs don't address the technical debt at all, at most they may appear to the obligor to be a bandaid that stops the bleeding, but only for a while. At some point the debt comes due, and when it does, no amount of agents will spare the need for payment with experience and currency. | Oct 24 18:05 |
| schestowitz[TR2] | Hell of an experiment the world finds itself as a lab rat in, that in never signed up for. | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 1 day | Oct 24 18:05 |
| schestowitz[TR2] | RedGreen925 | Oct 24 18:05 |
| schestowitz[TR2] | I am old enough to remember when free/open source software people respected copyright and not stealing other people work. Apparently that has all gone into the distant past with their support of the outright theft and abuse of copyright ever single AI does training their models and their using it. Oh well cannot wait for the inevitable lawsuits coming once some stolen code is discovered in a commit. The schadenfreude will be so del | Oct 24 18:05 |
| schestowitz[TR2] | icious when them hypocrites get their oh so hard done by whine on because of it. | Oct 24 18:05 |
| schestowitz[TR2] | 8 | Oct 24 18:05 |
| schestowitz[TR2] | 2 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 1 day | Oct 24 18:05 |
| schestowitz[TR2] | Anonymous Coward | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Anonymous Coward | Oct 24 18:05 |
| schestowitz[TR2] | As I've stated many times previous this won't work because you have the excuse AI did it. You may have to remove it but how would a lawsuit work? You didn't steal it. AI just gave it to you. | Oct 24 18:05 |
| schestowitz[TR2] | This goes back to something a lot of people struggle with. A company is not a person. This has been something done on purpose. I can litigate against a company but rarely to a person (unless it's caused physical harm to people). This means I'm limited if the company ceases to exist or liquidates or becomes bankrupt or even gets bought by another company in certain circumstances. | Oct 24 18:05 |
| schestowitz[TR2] | Who are you going to go after? The AI companies themselves have committed the abuse of copyright but no government is going to regulate or stop it because they are all invested in AI. The UK passed regulation that data centres are essential infrastructure and as such can negate planning law. That tells us everything about the way this is going to go and there is nothing we can do about it. It's only when some rogue government uses | Oct 24 18:05 |
| schestowitz[TR2] | it create a virus and cure (that's the key) that we will all be thinking, you know what, we should have done something about this earlier. | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | 6 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 1 day | Oct 24 18:05 |
| schestowitz[TR2] | alkasetzer | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | "the AI did it": you are still responsible by what you produce (with or without AI help). In doubt check the T&C of all cloud based code assistants for where the liability lies. | Oct 24 18:05 |
| schestowitz[TR2] | Also you forget that different countries have different laws, and that IP is a thing and applicable to persons not just companies. Just ask all the people Nintendo sued over the years. | Oct 24 18:05 |
| schestowitz[TR2] | 7 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 19 hrs | Oct 24 18:05 |
| schestowitz[TR2] | Anonymous Coward | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Anonymous Coward | Oct 24 18:05 |
| schestowitz[TR2] | No. | Oct 24 18:05 |
| schestowitz[TR2] | The issue with Nintendo is you are taking the IP of a specific thing. Code doesn’t work like that. You are getting the code to do a thing. Can you get it to do a thing without using someone else's IP? Obviously you can. | Oct 24 18:05 |
| schestowitz[TR2] | You're still not getting this. I am not responsible for what AI creates for me unless I specifically go after a specific IP. If I say to AI write me a mail client and server then it rips off everyone else I didn't ask for that so how can I be responsible for what it gives me? Are there laws that force me to check it? No. | Oct 24 18:05 |
| schestowitz[TR2] | As for laws and countries. They don't matter any more. These AI companies are all based in America. It's capitalism. They will win whether we like it or not. You think we are going to start reverse engineering applications and taking them to court in Finland? | Oct 24 18:05 |
| schestowitz[TR2] | Downvote me all you want. I'm right. Whether people like it or not. I don't make the rules on how the world works. If you want to live in a fluffy blinkers on paradise be my guest but this is going to be the end result. Licences mean nothing now. AI is consuming it all. | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | 6 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 9 hrs | Oct 24 18:05 |
| schestowitz[TR2] | alkasetzer | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Not being confrontational so take the next comments at value. | Oct 24 18:05 |
| schestowitz[TR2] | In law you have the concept of due diligence that can be applied to the situation you describe. And is appliable to companies or individuals. Obviously it depends on who is accusing whether or not to use it, demonstrate (or not, depends on the country) and on judges to accept it. | Oct 24 18:05 |
| schestowitz[TR2] | Never the less I do agree with you that people are being hypocritical, expecially linus et all, as you said they are vibe coding functionality without checking if there are issues with licences or IP. So, it follows that they don't care about those questions because they are developers and following laws is for lawyers and if that is the case they should also drop the GPL requirement of the linux lernel. Every one w | Oct 24 18:05 |
| schestowitz[TR2] | ould benefit. | Oct 24 18:05 |
| schestowitz[TR2] | the same could be said to Disney, epic, ubisoft and any AI users, if you don't actually care about following the law with respect to licences of stuff that you made that may be from third parties you should lose the right to sue when third parties infringe on your stuff. | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 1 day | Oct 24 18:05 |
| schestowitz[TR2] | Stevie | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Bah! | Oct 24 18:05 |
| schestowitz[TR2] | Nearly passed out laughing from the "quality" argument after the industrial weaving example. | Oct 24 18:05 |
| schestowitz[TR2] | If recent technological history teaches us anything it is that once scaled up to industrial production CHEAP trumps QUALITY every time. | Oct 24 18:05 |
| schestowitz[TR2] | You disagree? | Oct 24 18:05 |
| schestowitz[TR2] | Then I guess we all used Betamax for our video recording until our universally-adopted Apple computers took over that niche. | Oct 24 18:05 |
| schestowitz[TR2] | Software on the industrial scale has NEVER been "quality" focused. Pick any off-the shelf product in iteration 1.0 and tell me otherwise. | Oct 24 18:05 |
| schestowitz[TR2] | So the software we're talking about will be vigorously monitored you say? People are lazy. When Ai use becomes automatic there will be damn little oversight, and the sheer volumes of stuff made will defeat even a well-intentioned auditor unless they use ... AI to scan it. | Oct 24 18:05 |
| schestowitz[TR2] | 4 | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 22 hrs | Oct 24 18:05 |
| schestowitz[TR2] | ecofeco | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Bah! | Oct 24 18:05 |
| schestowitz[TR2] | Bit of trivia about Betamax: it WAS what the entire TV production industry used. Not the end customer. The production side. | Oct 24 18:05 |
| schestowitz[TR2] | 4 | Oct 24 18:05 |
| schestowitz[TR2] | 2 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 20 hrs | Oct 24 18:05 |
| schestowitz[TR2] | Richard 12 | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Um, actually | Oct 24 18:05 |
| schestowitz[TR2] | That was Betacam, not Betamax. | Oct 24 18:05 |
| schestowitz[TR2] | Very similar technology, but notably higher spec and cost - hence why a lot of tapes got wiped and reused. | Oct 24 18:05 |
| schestowitz[TR2] | 2 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 1 day | Oct 24 18:05 |
| schestowitz[TR2] | Groo The Wanderer - A Canuck | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Linus is 100% correct in this - if you are not a competent developer who knows how you expect the code to look and work, you have no business using Artificial Ignorance to speed up the destruction that "programmers" like you cause. It isn't the tools that really do the damage, it's your level of incompetence at the job basics. | Oct 24 18:05 |
| schestowitz[TR2] | 5 | Oct 24 18:05 |
| schestowitz[TR2] | 2 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 1 day | Oct 24 18:05 |
| schestowitz[TR2] | Dan 55 | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | But then he vibed up changes to RCU which included two bugs and left it to someone else to fix, and now that someone else has to sort out the mess to please the boss. Probably the very definition of a 'drive-by' patch. He wouldn't tolerate it from someone else. | Oct 24 18:05 |
| schestowitz[TR2] | 5 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 24 hrs | Oct 24 18:05 |
| schestowitz[TR2] | Anonymous Coward | Oct 24 18:05 |
| schestowitz[TR2] | Anonymous Coward | Oct 24 18:05 |
| schestowitz[TR2] | interesting | Oct 24 18:05 |
| schestowitz[TR2] | I'm pretty much an AI Luddite here. One of the specialized software packages we use at work just announced some shiny new AI features. My first question was "thay AI feature, how do I shut it off". | Oct 24 18:05 |
| schestowitz[TR2] | I do see where Linus is coming from, though. | Oct 24 18:05 |
| schestowitz[TR2] | I think back to the early days of web development. At first, everything was hand-crafted HTML. Then web development tools came along to streamline the process. | Oct 24 18:05 |
| schestowitz[TR2] | I have a friend who built websites back then. He proudly let us know that he hand wrote every line of HTML, because he didn't trust the tools (or because that's what "real" web developers did). | Oct 24 18:05 |
| schestowitz[TR2] | Web development tools advanced so that "anyone can build a web page" without knowing HTML. That's true, but it doesn't mean you'll be any good at it, or that your site will perform well (think bloated frontpage site with loads of plugins). | Oct 24 18:05 |
| schestowitz[TR2] | Also true that in 2026 you're not going to hand craft anything meaningful out of raw HTML. | Oct 24 18:05 |
| schestowitz[TR2] | AI feels like it's on a similar path. Marketing hype is that anyone can be a developer (with the implication that the CEO can improve profits by firing all the geeky programmers, engineers, developers, QA, etc.). Reality is that you still need smart people who know how to use the tools. | Oct 24 18:05 |
| schestowitz[TR2] | 5 | Oct 24 18:05 |
| schestowitz[TR2] | 3 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 22 hrs | Oct 24 18:05 |
| schestowitz[TR2] | ecofeco | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | FAIL | Oct 24 18:05 |
| schestowitz[TR2] | Re: interesting | Oct 24 18:05 |
| schestowitz[TR2] | Oh yes. Websites are the perfect example: The everyday bloat of websites IS THE PERFECT EXAMPLE. | Oct 24 18:05 |
| schestowitz[TR2] | I too, used to create websites by hand. Then I started using Netscape Composer. But I still had to troubleshoot/debug the website and if I did not know how to manually write HTML, the website was buggy. | Oct 24 18:05 |
| schestowitz[TR2] | I then moved to Dreamweaver. But still had to do the same thing. And Dreamweaver was much more forgiving. But building for different display aspect ratios was the obstacle back then. As was including more CGI Bin. I could fit large websites, with pictures, and roll-over menus on an entire 3.5" floppy thanks to my ability to optimize like hell. I even built webservers. (god bless Apache) | Oct 24 18:05 |
| schestowitz[TR2] | Then came Frontpage's crap.Then came CSS to counter Frontpage's crap. Then came, JSON, XML, advanced JAVAScript. Then template abuse. Then dynamic content. Then Wordpress. | Oct 24 18:05 |
| schestowitz[TR2] | And then it got worse. And then more worse. Until we are where we are today. With every past "helper" system being abused beyond sanity. And the backend cluttered worse than a cat lady's hoarder's house. | Oct 24 18:05 |
| schestowitz[TR2] | And that's JUST the website ecosystem. The same thing happened to ALL other systems as well. | Oct 24 18:05 |
| schestowitz[TR2] | I saw the prefect description the other day: digital friction. AI/LLMs are not a shortcut, they are just more shite on the pile. | Oct 24 18:05 |
| schestowitz[TR2] | 2 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 21 hrs | Oct 24 18:05 |
| schestowitz[TR2] | ecofeco | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: interesting | Oct 24 18:05 |
| schestowitz[TR2] | > Also true that in 2026 you're not going to hand craft anything meaningful out of raw HTML. | Oct 24 18:05 |
| schestowitz[TR2] | Not true at all. | Oct 24 18:05 |
| schestowitz[TR2] | Giant shopping websites? Message boards? Yes, True. | Oct 24 18:05 |
| schestowitz[TR2] | General information websites? Easy peasy. | Oct 24 18:05 |
| schestowitz[TR2] | 3 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 18 hrs | Oct 24 18:05 |
| schestowitz[TR2] | doublelayer | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: interesting | Oct 24 18:05 |
| schestowitz[TR2] | I'd take that in the other direction. It's very common for the basic website to be built around some CMS or framework because people don't want to reinvent several boring wheels or to have to teach HTML to everyone who is allowed to post something to it. But more complex web frontends do require frontend developers who are writing HTML from scratch. They're likely using JS frameworks, either client or server side, but those jus | Oct 24 18:05 |
| schestowitz[TR2] | t do some standard tasks for you. They organize the HTML that frontend devs write. | Oct 24 18:05 |
| schestowitz[TR2] | So depending on what "raw HTML" means, it is either incorrect or much less important than stated. If that means that every page is a single static file, then it's true that there are less of those. But HTML written in a template system is still HTML and there's a lot of that which is written by hand. | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 15 hrs | Oct 24 18:05 |
| schestowitz[TR2] | Groo The Wanderer - A Canuck | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: interesting | Oct 24 18:05 |
| schestowitz[TR2] | Whether you can "craft anything meaningful out of hand-crafted HTML" depends on which tools you use to replicate your pieces of hand-crafted data where they are needed. Even a simple website include directive does this - it brings in a whole pile of HTML-compatible text based on a simple directive. That is all that most "HTML editors" do - auto-insert begin/end tags around what you're typing. | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 24 hrs | Oct 24 18:05 |
| schestowitz[TR2] | RM Myers | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Coat | Oct 24 18:05 |
| schestowitz[TR2] | ...think bloated frontpage site with loads of plugins | Oct 24 18:05 |
| schestowitz[TR2] | Was that meant to be a polite way to criticize El Reg? Not that their new design has a bloated front page, of course. | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 22 hrs | Oct 24 18:05 |
| schestowitz[TR2] | ecofeco | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: ...think bloated frontpage site with loads of plugins | Oct 24 18:05 |
| schestowitz[TR2] | MS Frontpage. M$ attempt to monopolize the early Internet. | Oct 24 18:05 |
| schestowitz[TR2] | 2 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 23 hrs | Oct 24 18:05 |
| schestowitz[TR2] | VoiceOfTruth | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | You want it both ways | Oct 24 18:05 |
| schestowitz[TR2] | You complain that 'AI' is built on sucking in lots of code and ignoring the licences. True. | Oct 24 18:05 |
| schestowitz[TR2] | But then... | Oct 24 18:05 |
| schestowitz[TR2] | >> Another AI win is that you can use AI to reverse-engineer Windows drivers for Linux more easily | Oct 24 18:05 |
| schestowitz[TR2] | What if the licences for those drivers do not permit reverse engineering? Ignore them for the good of Linux? | Oct 24 18:05 |
| schestowitz[TR2] | 5 | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 12 hrs | Oct 24 18:05 |
| schestowitz[TR2] | Benegesserict Cumbersomberbatch | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: You want it both ways | Oct 24 18:05 |
| schestowitz[TR2] | The PC clone industry came into existence without violating any reverse engineering clause in licensing. It isn't reverse engineering if, in analysing a piece of code, you generate a very concise specification of what it does. Someone else who takes that spec and produces their own code that faithfully complies with it is equally innocent. | Oct 24 18:05 |
| schestowitz[TR2] | Hell, you could probably argue that as long as you use different machines, or even different software on the same machine, to perform each task, that one person could achieve the whole task of reimplementation and still not have reverse engineered the software. | Oct 24 18:05 |
| schestowitz[TR2] | This is the irony I'm waiting to see in real life: corporate behemoths build open weights models on wholesale violation of copyright. Random coders use said models to deconstruct and rebuild their own compatible instances of software, OS and web services. The company is probably guilty as hell, but the random coder comes out producing free software without having breached copyright at all. | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 23 hrs | Oct 24 18:05 |
| schestowitz[TR2] | BinkyTheMagicPaperclip | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Gateway drug to programming | Oct 24 18:05 |
| schestowitz[TR2] | There's so much wrong here it's unreal. Whilst I personally think the LLM argument is a foregone conclusion, and the genie isn't going back in the bottle, the attitude of using an LLM as a force multiplier for a beginner panders to the worst excesses of everyone's early programming journey. | Oct 24 18:05 |
| schestowitz[TR2] | You're going to set the world alight with your new thing! For most of us this is unachievable and we calm down a little, but now through the joys of an LLM it can slop out something that vaguely works, where you've no understanding of what it did, and then get upset when experienced programmers refuse to fix the errors in your slop baby. | Oct 24 18:05 |
| schestowitz[TR2] | There are so many areas which needs help. Not all of them require experienced programmers, just enthusiasm, spare time, and the willingness to repeatedly bang your head against a wall until you learn. | Oct 24 18:05 |
| schestowitz[TR2] | The only issue is most of these areas aren't 'create the next MMO' or 'I'm going to code an office suite from scratch', and most non nerds are not going to be impressed by your efforts. | Oct 24 18:05 |
| schestowitz[TR2] | Try some software. Find something that's annoying. Start searching about how to fix it. You don't need an LLM for that. | Oct 24 18:05 |
| schestowitz[TR2] | Although, poorly created code is not a new thing. Pre LLM programmers have been dealing with abstractions from the hardware for years, and it's an increasing problem when developers think they have unlimited resource and code to that attitude. LLMs are just going to make this worse unless there's enforcement of programming standards. | Oct 24 18:05 |
| schestowitz[TR2] | 4 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 22 hrs | Oct 24 18:05 |
| schestowitz[TR2] | ecofeco | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Reply Icon | Oct 24 18:05 |
| schestowitz[TR2] | Re: Gateway drug to programming | Oct 24 18:05 |
| schestowitz[TR2] | Idiot savants don't know they are idiot savants. | Oct 24 18:05 |
| schestowitz[TR2] | The GIGO will be glorious! | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 22 hrs | Oct 24 18:05 |
| schestowitz[TR2] | DrXym | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Too true | Oct 24 18:05 |
| schestowitz[TR2] | AI can work with supervision but you need to have the knowledge for when it outputs good code or slop. It's best to give it small, achievable tasks on existing code and use it more for boiler plate and donkey work than anything that needs thought. | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 20 hrs | Oct 24 18:05 |
| schestowitz[TR2] | tekHedd | Oct 24 18:05 |
| schestowitz[TR2] | I'm getting a mixed message here, hmm... | Oct 24 18:05 |
| schestowitz[TR2] | "We all start clueless, but some people spend the time and effort to become something more." | Oct 24 18:05 |
| schestowitz[TR2] | Translation: I know I said AI is OK but what is with these contributions?! Learn to code! (But please don't cut my funding, I love AI really I do no really.) | Oct 24 18:05 |
| schestowitz[TR2] | 3 | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 20 hrs | Oct 24 18:05 |
| schestowitz[TR2] | Anonymous Coward | Oct 24 18:05 |
| schestowitz[TR2] | Anonymous Coward | Oct 24 18:05 |
| schestowitz[TR2] | If is that great then automate it | Oct 24 18:05 |
| schestowitz[TR2] | If really AI is great for producing code then corps and orgs should use AI to generate, test and deploy AI code and not ask humans in the middle to press button and take liability for it. That is the bottomline. | Oct 24 18:05 |
| schestowitz[TR2] | 3 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 14 hrs | Oct 24 18:05 |
| schestowitz[TR2] | Sitaram Chamarty | Oct 24 18:05 |
| schestowitz[TR2] | you're being too optimistic | Oct 24 18:05 |
| schestowitz[TR2] | "Competent" only goes so far down the chain starting from Linus Torvalds. At some point there will be slop turned in. | Oct 24 18:05 |
| schestowitz[TR2] | It's all very well to say "discourage the drive-by patches" but when the *rate of incoming patches* is 10X what it used to be, good luck. The first competent guy it hits will either drown in workload or reject more than he would have if he had the time. Worst case, something nasty will slip by because he was overloaded. | Oct 24 18:05 |
| schestowitz[TR2] | Me, I'm playing with NetBSD on a VM as and when I find time. Linux will be safe for a few more years, but I'm hedging my bets. Worst case, I learn a new OS and have some fun. | Oct 24 18:05 |
| schestowitz[TR2] | PS: What happened to "biting the hand that feeds IT"? Is that not El Reg's tagline anymore? | Oct 24 18:05 |
| schestowitz[TR2] | 2 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 7 hrs | Oct 24 18:05 |
| schestowitz[TR2] | dsmithhfx | Oct 24 18:05 |
| schestowitz[TR2] | This will end in tears, and rollbacks. For now, I have paused updating. | Oct 24 18:05 |
| schestowitz[TR2] | 1 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 7 hrs | Oct 24 18:05 |
| schestowitz[TR2] | blu3b3rry | Oct 24 18:05 |
| schestowitz[TR2] | Silver badge | Oct 24 18:05 |
| schestowitz[TR2] | Meh | Oct 24 18:05 |
| schestowitz[TR2] | Can't say I'm convinced | Oct 24 18:05 |
| schestowitz[TR2] | I happily admit I'm a skeptic when it comes to anything to do with AI, especially as a lot of thinking in the corporate world seems to be "more AI, more better" with regards to a lot of products (Microsoft I'm looking at you) along with cramming it into places it really isn't needed or wanted. Just look at all those spectacularly shite and useless customer support chatbots. | Oct 24 18:05 |
| schestowitz[TR2] | Reading through the article made me feel like this was more of a shill than anything else. I'm not going to dispute that its possible to create "good" code with an LLM given due diligence, but equally having a tool do the thinking for you reduces your ability to do it yourself. | Oct 24 18:05 |
| schestowitz[TR2] | As the cost of these tools seems to be unsustainable in the long term, is a reduction in skill and ability really something you want? Even more so for those who have never learned to do anything without the LLM holding their hand. | Oct 24 18:05 |
| schestowitz[TR2] | 2 | Oct 24 18:05 |
| schestowitz[TR2] | 0 | Oct 24 18:05 |
| schestowitz[TR2] | Reply | Oct 24 18:05 |
| schestowitz[TR2] | 44 mins | Oct 24 18:05 |
| schestowitz[TR2] | HMcG | Oct 24 18:05 |
| schestowitz[TR2] | Production ready code | Oct 24 18:05 |
| schestowitz[TR2] | “ Because while you still can't vibe code your way into delivering production-quality software” | Oct 24 18:05 |
| schestowitz[TR2] | You’re having a laugh. As the last couple of years have proven, with security breach after security breach, most if not all “production ready” code is already a sloppy bag of crap. | Oct 24 18:05 |
| schestowitz[TR2] | The fact that software companies have never been held responsible for shipping shoddy, bug ridden products, or even being fit for purpose, means the vast majority of shipped code is already slop. AI isn’t going to make it any worse. | Oct 24 18:05 |
| schestowitz[TR2] | " | Oct 24 18:05 |
| -TechBytesBot/#techbytes-www.theregister.com | AI coding tools make developers slower, study finds | Oct 24 18:14 | |
| schestowitz[TR2] | <li> | Oct 24 18:16 |
| schestowitz[TR2] | <h5><a href="https://www.bsdnow.tv/686">BSD Now 686: Yell in your Shell</a></h5> | Oct 24 18:16 |
| schestowitz[TR2] | <blockquote> | Oct 24 18:16 |
| schestowitz[TR2] | <p>! marks in shell, learning at scale, Thinkcenters and BSD, and more... </p> | Oct 24 18:16 |
| schestowitz[TR2] | </blockquote> | Oct 24 18:16 |
| schestowitz[TR2] | </li> | Oct 24 18:16 |
| schestowitz[TR2] | Oct 24 18:16 | |
| -TechBytesBot/#techbytes-www.bsdnow.tv | BSD Now 686: Yell in your Shell | Oct 24 18:30 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has joined #techbytes | Oct 24 18:37 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has left #techbytes (Error from remote client) | Oct 24 20:30 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has joined #techbytes | Oct 24 20:46 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has left #techbytes (Error from remote client) | Oct 24 21:06 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has joined #techbytes | Oct 24 21:40 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has left #techbytes (Error from remote client) | Oct 24 21:45 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has joined #techbytes | Oct 24 21:52 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has left #techbytes | Oct 24 22:02 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has joined #techbytes | Oct 24 22:09 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has left #techbytes | Oct 24 22:52 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has joined #techbytes | Oct 24 22:54 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has left #techbytes | Oct 24 22:59 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has joined #techbytes | Oct 24 23:01 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has left #techbytes | Oct 24 23:06 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has joined #techbytes | Oct 24 23:09 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has left #techbytes | Oct 24 23:16 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has joined #techbytes | Oct 24 23:19 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has left #techbytes | Oct 24 23:24 | |
| *psydruid (~psydruid@jevhxkzmtrbww.irc) has joined #techbytes | Oct 24 23:26 | |
Generated by irclog2html.py 2.6 | ䷉ find the plain text version at this address (HTTP) or in Gemini (how to use Gemini) with a full GemText version.
