catch

@CheapPontoon @jalefkowit @machias 15 years ago Honda built a 'British road' in Hokkaido wth shitty road surface and loads of potholes so they could test cars against conditions they don't find in Japan (where the roads are surfaced to match the climate now instead of the climate 40 years ago and cars are generally still normal sized instead of domestic tanks). Sounds like Samsung needs to build an 'American grid' in Busan.

@akareilly@hachyderm.io I would love a situation where I collect nettles on my daily walk and drop them off somewhere (possibly involving some retting prep) and occasionally I get a shirt or pair of jeans out of it.

@ketan@climatejustice.social should've called him Burning Heath to match Sophie Money-Coutts

@vk6flab@mastodon.radio @akareilly@hachyderm.io it's OK my dog needs three walks per day.

@NotTheLBCGuy@mstdn.social It is 1945. I am being promised that a bold, dynamic new Labour leader will revitalise the country. Attlee sends troops to break the strikes by the London dockers then forcibly deports Chinese seaman from Liverpool separating them from their British wives and children.

@ai6yr@m.ai6yr.org @markc568@beige.party can vouch for m00de audio, I run three vintage hifis of that (two of them have turntables attached as well) although I bought a hifiberry HAT rather than a self build. Never seen this one before.

@ai6yr@m.ai6yr.org alphabetical but with classical LPs split out because I don't have many and some should be listed by artist (multiple composers) but others would make more sense listed by composer, so fitting them in with the rest would be painful.

@afewbugs@social.coop extra bonus. They're not retired. They run a YouTube channel and free (for your contact details which they are amassing) course, which is currently running at a small annual loss but which they clearly expect to gradually turn into money via YouTube earnings and a book deal. https://rebeldonegans.com/does-rebel-finance-school-make-money/

@akareilly@hachyderm.io was sure I'd seen this and it's not the specific thing I'd seen but that looks like the problem is already solved. https://www.riaburns.co.uk/blog/2018/2/9/knitting-machine-hacking

@jonny function speechBubble()

@farbel@mas.to @selzero@syzito.xyz if you're getting the bus in London there will often be multiple buses that stop at the same bus stop. And if one stops 20 minutes walk from your destination and the other 1 minute walk, you can be much better off waiting for the right bus. If you get on the first bus that comes, you may even get to watch the one you actually wanted drive past you at the next bus stop. Even more the case with trains where an express or semi express that sort of goes near your destination may either require changing to the train you needed anyway, or overshoot and require going backwards. So your analogy reads like it's from someone that either doesn't take public transport or lives somewhere where the service is so bad it might as well not exist. Also if you get married you can get separated or divorced any time. How exactly do you recall US politicians during their term and when's the last time that happened?

@swelljoe @glyph the very little I've used one I found the loop distasteful. It's like sending food back in a restaurant or going through a convoluted returns process. Either you're an arsehole or you're getting ripped off or both. And there is no manager. Just the automated customer service agent. I do not understand how this is fun and addicting for people. People appear to be trying to get around this 'manual' loop with the various Ralph-type workflows which just happens to be a way to set your wallet on fire at the same time. Like going to a restaurant that does randomly cooked steaks and buying five of them but only eating one.

@swelljoe @glyph before LLMs came along I had (and still do) questions about what makes some people predisposed to conspiracy theory rabbit holes - people can be fine for decades then fall into one. I don't think it's the same thing (except in the known cases where they literally are the same thing because people talked to LLMs about conspiracy theories) at all but they are not dissimilar. Potentially the feeling that there is some kind of discovery that is just out of reach?

@sluttymayo@jorts.horse this is the way

@sluttymayo@jorts.horse I see these pop up but thought it was a very short auction not a video. Is it like early 90s TV shopping then?

@burnitdown @davidgerard it's not bad faith reasoning it's literally the position I'm being put in as an open source maintainer who doesn't use LLMs. There's no consensus amongst the other maintainers whether to ban or discourage LLM usage or not, so while all of that discussion is happening there is a (fortunately still low, but increasing) number of LLM derived contributions I end up looking at one way or another. Some of them are complete bullshit and can be rejected out of hand without any policy, just bad issues/MRs. Some of them are not like that (usually from existing contributors trying things out, not random vibe coders).

@TheEntity @burnitdown @davidgerard you can ask the non profit that is responsible for the website which I am neither a member nor supporter of. Neither does Drupal core which I co-maintain have any LLM integration. It's in a contributed plugin. Why this justifies plastering false claims about AI all over the marketing pages on the website I have no fucking idea and I've pointed this out multiple times.

@vksxypants @jalefkowit the little indicator panel on the front and the light coming on when you open the door, so that a store can have it plugged in without keeping an empty fridge cold (x 50 other fridges in the store). Completely useless except to refrigerator retailers, I can kind of see why it would exist though. But this seems like something you would control with recessed switches inside a panel at the back or something if it exists at all.

@mattsheffield did he ask 'Claudia' about his little pot of honey?

@abolisyonista@ni.hil.ist it's like calling the Russian Revolution backed by German imperialism because they paid Lenin's train fare.

@agowa338 @resuna @davidgerard yes or found one with an LLM and wrote the report up by hand. Also sometimes I can spot LLM content from sentence one, sometimes it's four paragraphs in. The amount of time to determine if it's LLM or not is not necessarily less than the amount of time to determine if it's completely obviously invalid vs needs some kind of triaging either.

@resuna @agowa338 @davidgerard fuzzing tools were prior to LLMs a major source of bullshit reports to FOSS security teams because people love to run them, generate a report, then send it off without verifying any of the results. And they find a lot more false positives than real issues. Have to admit seeing those reports put me off trying them directly (much like LLM generated code does) so there could be good ones with a better rate out there that I don't know of. Someone mentioned coverity got better recently.

@mnl @glyph I have done this many times debugging but it's always hard to know how much headbanging and process of elimination was necessary before the fresh air eureka moment. Fortunately I have a dog that needs four walks per day and an 11 year old so there is a limit to uninterrupted computer time even if not necessarily going back to it later. Still managing to barely touch LLMs (not so much their output. Deal with that almost daily).

@agowa338 @davidgerard yeah it's resignation for a fair number of people, I'm sure it's also bad faith from some. Also a complete ban on LLM contributions means banning bugs or security issues found using LLMs even if the all the code to fix the issue is handwritten. While it's an inefficient way to find bugs and security issues, they do get found. Could make an exception for LLM 'analysis' and require code to be written without them but that comes under 'letting them through'. Or you could close the issues and re-open them as if you'd found them independently but that is essentially laundering - not even saying that would be a bad approach necessarily, but it's what it would be. Then you get down to 'you must understand what you're submitting as if you'd written it manually' which people also lie about and I increasingly have to point out reviewing MRs. And same with the fake bug reports and security issues that come in alongside the real ones.

@davidgerard@circumstances.run looked up the apparent lack of it on Linux and found this, which suggests it might be force installed in the coming months if not since February. https://support.google.com/gemini/thread/412505429/gemini-in-chrome-for-linux?hl=en

@davidgerard@circumstances.run my daughter got a used Samsung one off backmarket (the one that is flip phone sized not the massive one) and she has had to replace the built in screen protector with an after-market one. Was able to do it herself. But otherwise it is still working after two years or so. I don't think this means they are good or fine but it perhaps means they are not especially more broken than most other phones these days, except maybe a fairphone.

@burger_jaap@mastodon.social this reminded me of how underused home solar batteries are. Rather than a long series of replies, made a new thread. https://kolektiva.social/@catch56/116543191062783040

@adamgreenfield@social.coop @pvonhellermannn@mastodon.green @Christo_459@mastodon.me.uk isn't this pointing out the double standard of the press rather than saying there should be a double standard though? This right wing anti-Semitism against anti-Zionist Jews goes all the way back to Churchill. E.g. https://en.wikisource.org/wiki/Zionism_versus_Bolshevism

@celesteh was it effective altruism?

@resuna @agowa338 @davidgerard I may actually be doing this soon, or persuading someone else do to it, so I don't disagree. It's still a pain in the arse though, and I definitely wouldn't be thinking about it at all if the slop reports weren't likely to pick up. And it still doesn't guarantee a 1-1 match with what the LLMs find (real or not).

@TheEntity @burnitdown @davidgerard I have personally objected to that and find it offensive on multiple levels, so yes it's that project. Not least because it's not 'AI powered' in the slightest.

@jonny @david_chisnall @bms48 @chris_evelyn the anthropic glasswing thing is this. So far what I've seen and heard is it can find non-exploitable security-adjacent bugs or low severity security issues (at some point which one something is, is a matter of taste). Which feels like spending millions of dollars to be able to announce you emailed an un-triaged coverity PDF to thousands of OSS projects. Maybe high severity issues will come through in the next few weeks. Wouldn't rule it out but we'll know soon enough.

@zkat saw this vibe-ported python version that they are now vibe porting to rust apparently. https://github.com/instructkr/claw-code

@davidgerard @agowa338 But the bug exists, and if it's relatively serious you'd want to fix it. So what you're suggesting, unless the bug stays unfixed, is telling the person to fuck off, but accepting the bug report, albeit by possibly closing the original and opening a new one that 'rediscovers' it without the LLM. Which I already described in my reply. But that is not actually banning 'LLM contributions' because the report started with the LLM, or at least the training data regurgitated through the LLM. I mean I have done similar things with bug reports from people that were just rude and abusive, so I'm not saying this is even a bad thing to do, but it doesn't prevent LLM-derived information getting into the project as such. It's like laundering. Or decaffeination. Now if it's a real bug you can verify it yourself by looking at the code involved. And if it's really real you could almost certainly say you'd have found it without the LLM if you'd gone looking but this is an earnest version of 'yes I understand and reviewed every line of code'. And by doing this, you become the person that is relying on the LLM to submit the bug report. Unwillingly. And a step removed, but still fundamentally translating from possible slop into a real report. The fact it's a 1/500 hit when the rest is complete shit doesn't change this. You can't just unsee things, this is part of the problem with LLMs in the first place.

@yvan not exactly low carb but something with polenta?

@bms48 @jonny @david_chisnall @chris_evelyn yeah even if we allow that potentially they can find the same or more issues as e.g. coverity with the same or less false positives, if it's orders of magnitude more cost to generate those reports then who is going to pay for it. The answer to that might be 'gullible politicians in multiple countries' though.

@photovince @celesteh oh I was going to ask, was it 'neo-reactionism' / nrx ala moldbug? (Hoping you don't mind the guessing game)

@vees@jawns.club i got a pack of rubber/silicon port savers off AliExpress and they have extended the life of my phone usb-c ports from approx 1 year to longer than the rest of the phone.

@CppGuy@infosec.space @anon_opin@mastodon.social the UN said then and still says today that population growth is slowing. See this summary page. Slowing growth is not the same as dropping, so who exactly didn't read things? https://www.un.org/en/global-issues/population

@Rycochet @davidgerard not really how policing, laws, or crime works though is it? See enforcement of bans on protest, arrests for shoplifting and mild drug use vs flouting environmental protections, wage theft, tax evasion and fraud.

@shonalika@deadinsi.de haven't actually read the book, just some articles dealing with the concept, but homonationalism sounds close to what you're asking about. https://read.dukeupress.edu/books/book/1247/Terrorist-AssemblagesHomonationalism-in-Queer

@CppGuy@infosec.space @anon_opin@mastodon.social they have kept revising their PR as their original stances got exposed. See for example this article from 2011, but feel free to quote mine Wikipedia instead of reading up https://www.radstats.org.uk/moral-panic-about-overpopulation-the-distracting-campaign-of-population-matters/

@gkrnours @arclight I read it as staging, production, and backups were all on a single storage volume, which got deleted with a single API call.

@akareilly@hachyderm.io also cargo biking the nettles or later stages between places would be enjoyable. Assuming there are people who like extracting fibers/weaving/maintaining sewing machines etc. who don't like collecting stinging nettles or cycling who would also do it on the same basis.

@burnitdown @davidgerard I thought it was me that was making an argument that 'not using an LLM is actually using one' but now you're definitely making that argument yourself. It's either bad faith or not you can't have it both ways.

@CppGuy@infosec.space @anon_opin@mastodon.social he's a patron of Population Matters for a very long time

@davidgerard @burnitdown neither of you have addressed my point about what to do with valid security and bug reports (except ignore them or throw the entire code base away) so I don't really think either of you are in a position to tell me what my problems are either.

@david_chisnall @bms48 @jonny @chris_evelyn this is extremely similar to my experiences of LLM review (at me, not by me) of MRs or projects. Loads of false positives, sometimes finds a (usually minor) real bug amongst them, extreme nitpicky defensive things that could potentially result in a different error message, or swallowing them and failing silently. It's being set on things I have 20+ years experience on (Drupal) so it's easy for me to dismiss the bogus stuff but still a huge timewaste. Someone could argue that if it finds one bug out of 15 false positives that wouldn't have been found in that issue it's useful. But this is a project with plenty of open bug reports and many of them with fixes nearly ready to go. So if you applied that workflow to every issue it would grind to a halt/add additional noise that needs to be ignored all the time.