HN rank: 1

HN score: 91

comments: 41

original url: https://www.motorsport.com/f1/news/horrible-totally-unacceptable-powerless-f1-drivers-frustrated-by-bahrain-f1-software-glitch/10861968/

评论信号：Seems wild that they can code and deploy an update of a critical-seeming component in the field like that. I thought such a firmware update would need extensive QA in the lab, along with root causing why this bug slipped through their processes anyway.But maybe that's the difference between shipping a controller for F1 versus shipping one for an industrial or consumer product? shrug. reply: I have no idea if modern-day SV’s “tests pass… idfk, it’s a web app… LGTM I guess” coding ethos was the cause of this particular bug. However, it’s a great, relatively benign demonstration of why it has absolutely zero fucking place controlling extremely dangerous hardware shooting a human around a race track surrounded by people.

评论信号：A question for someone who knows more about F1: Is this firmware shared by all cars? Or some local server at the track the cars connect to? Why is F1 able to control individual cars like this? The article is very light on details as to what exactly is going on... reply: The firmware situation is mixed. The ECU hardware is a standard TAG700 from McLaren. Basically these are software-defined ECUs where you can do whatever. However, FIA has some rules on what you're allowed to modify and has some "compulsory" software in there for monitoring and competition regulatory reasons. It appears there was very limited testing of this module in wet conditions which seems to have affected a lot of vehicles. (I work in Motorsports)

评论信号：Auto software engineer here.It’s frankly because Motorsport software guys are pretty arrogant. They look down on road car and truck engineers, but reality is they develop stuff with virtually no processes and they usually have only very basic engineering skills, yet every day they are told Motorsport is at the forefront of technology.“Motorsport grade” auto components usually means crappy, and everything requires a software patch or constant maintenance to work correctly.Not surprised there was a bug, and not surprised they were able to develop and deploy a patch in 1 hour without V&V testing it or anything. reply: > everything requires a software patch to work correctlyIsn't that literally the point? Most auto components aren't designed to be patched by the end user.

HN rank: 2

HN score: 647

comments: 303

original url: https://github.com/Niko1221/Strata

评论信号：I'm a little skeptical of going below 4-bit quants due to the potential for significant degradation in quality. I'm running 4-bit quants on an RTX Pro 6000 rented for approximately $1/hour and getting about 1.2 million tokens out and 40 million tokens in per hour with caching. The quality of 4-bit quant is good enough for difficult but well-scoped coding tasks. Here is the inference stack I am using: https://www.reddit.com/r/BlackwellPerformance/s/FrKwk3GoDK reply: I just ran a set of benchmarks, ninfer-3090-qwen3.8-27b (so mix of Q4 and Q5) vs strata-qwen3.8-flash-next-iq3_xxs (so Q3):│------------------- │ Ninfer-3090 │ Strata│ Code generation │ 52/78 (66.7%) │ 70/78 (89.7%)│ Code completion │ 40/50 (80.0%) │ 44/50 (88.0%)│ Total------------- │ 92/128 (71.9%) │ 114/128 (89.1%)│ API failures------ │ 10 │ 5- Ninfer generation: ~122 min total.- Strata generation: ~142 min total.So strata is a little slower, but keep in mind that ninfer-3090 is very optimized for a Qwen 3.8....

评论信号：I've just tested Strata on a simple 50 image vision benchmark. The task is to output the exact coordinates of a requested object. The result via Strata had a median error distance of 154.8 pixels, avg of 168.8. Running the exact same GGUF and vision adapter weights on llama.cpp gives me a median error of 46.5, avg 81.4.To put that into perspective, here are some more numbers from other models via llama.cpp:Median/AverageQwen 3.5 9B BF16: 46.5 / 193.3Qwen 3.6 35B Q4 K XL: 38.4 / 76.4Qwen 3.5 122B Q3 K M: 32.9 / 68.6The difference in vision performance is as large as the jump from a 9B model to a 35B model. All tests were performed at temp=0.I have done no further testing, as these results line up perfectly with my expectations. reply: How big are the images? Is 160px 1%, 10%…?

评论信号：For what's worth, on high, 27b is better, but both models on high are too slow to be used (way too many output tokens).On low, 3.8 Flash Next is better AND considerably faster.My results[0][1], on a RTX 3090 + 128GB DDR4 RAM. I still have to see if I can optimize any more settings for either, but I think I will switch from 3.8 27b to 3.8 Flash Next for my local LLM uses.[0]: https://aibenchy.com/?q=qwen+3.8+27b%2C+qwen+3.8+next[1]: https://aibenchy.com/compare/qwen-qwen3-8-27b-low/qwen-qwen3... reply: These benchmarks seems to suggest that Flash Next performs best on Low thinking compared to the same model on higher thinking levels?And that Qwen 27b outperforms both when set to high?What quants are being compared here?

HN rank: 3

HN score: 54

comments: 4

original url: https://filmstories.co.uk/news/tippett-studios-in-the-wake-of-its-closure-a-digital-archive-of-animated-materials-appears-online/

评论信号：Tippett Studios or Phil’s name should probably be in the title here to give this the audience it deserves. He’s a legend and many many nerds will know his name.

评论信号：Direct Link to the digital archive at Internet Archive -https://archive.org/details/tippett-archive/

评论信号：Huge thanks to him. It’s a real shame that so much material in this world, for one reason or another, will never see the light of day.

HN rank: 4

HN score: 49

comments: 18

original url: https://www.themoscowtimes.com/2026/10/02/nearly-200-people-under-observation-after-irkutsk-lab-worker-dies-from-plague-a93857

评论信号：I literally just finished Annie Jacobsen's book 'Biological War: A Scenario' last week where the central plot point is an accidental release of modified form of pneumonic plague from a lab in Siberia. Very coincidental timing, especially since it made me wonder how often labs like these have accidents and we never hear about them because it is cleaned up without issue. reply: > made me wonder how often labs like these have accidents and we never hear about them because it is cleaned up without issue.Did anyone ever confirm if the lab in Wuhan China was the originator of COVID? I remember that was a rumor but never heard if there was anything to it.

评论信号：So breaks a test tube of live bacteria at an “anti-plague institute”, possible infects 100 colleagues, then when she developed plague like symptoms is taken to hospital and possibly infects 100 patients.I think they need to investigate their safety protocols> She reportedly told medical staff that she had accidentally broken a test tube containing live bacteria while collecting samples for testing.> At least 197 people who may have had contact with her were reportedly placed under medical isolation, including more than 100 in hospital wards. reply: Was this a glass test tube? I understand why glass is probably the best material for the inner lining of containers that hold biological material, but I'm surprised that there aren't better solutions for hazardous material (e.g. an unbreakable, transparent polymer outer coating).

评论信号：How often does this sort of thing happen? I'm guessing more than we think, either because we forget the news stories we read or because it isn't reported. reply: There was a COVID 19 lab leak in Taiwan about a year into the pandemic.https://www.taiwannews.com.tw/news/4374289Six errors contributed to this leak.

HN rank: 5

HN score: 13

comments: 4

original url: https://plicerin.github.io/riverraid-rom-port/

评论信号：I think that most people who once loved this game moved on from videogames a long time ago. Like almost all games on the Atari, it won't interest the younger generations for more than a few minutes. But I still remember it fondly. reply: What is the appropriate age to move on from gaming? Asking for a friend.

评论信号：The cartridge rebuilt from the original ROM, one instruction at a time