ᕼᑎ #1github.com328 pts143 commentsProgrammingworth reading

Livenerf: Has Opus 5.5 been nerfed yet?

echo@mulan ~/hn/story-1

HN 热门故事「Livenerf: Has Opus 5.5 been nerfed yet?」进入今日前列,值得先打开原文和讨论串判断它真正有价值的部分。

模型分析没有产出可用结构化结果;页面保留了 HN 热度、原文入口和讨论信号,避免用空泛总结替代一手材料。

它在 HN 上获得约 328 分和 143 条评论,说明这个话题至少触发了社区讨论;真正的判断仍要回到原文证据和评论区的分歧点。

这是一条降级分析:它不冒充完整解读,只把可验证的元数据、原始链接和 HN 讨论保留下来,方便稍后重新生成或人工阅读。

评论区已经提供了一些读者反应,但这里还没有形成完整综合。

  • 评论信号:We also have Nerf Bench:https://www.bridgebench.ai/nerf-benchThey test it on launch day, then benchmark it against that. A deviation of above 10% is considered a change. They're currently tracking Opus 5.5 and GPT-6 Astra.This bench famously detected a degradation of Opus 4.6 which Anthropic later blogged about. I personally think people sense nerfs more often than they happen and that it's often about honeymoon effects. reply: But you are using API not the CLI right? I did not ever observe API degradation, only subscription stuff through their CLI.

  • 评论信号:> It could also mean nothing happened and people are pattern-matching on noise.It is definitely not this. Anthropic has thousands of employees making probably > 100,000+ tiny changes across the entire stack and infrastructure everyday.The compounding effect can definitely cause temporary regressions in some domain or use-case that doesn’t have really good coverage in their internal tests.What makes this particularly challenging in the case of LLMs is how changing some language in the prompt can vastly affect the output.But this is less true as models become larger and additional parameters are able to capture each and every possible nuance of the language. I’d say caching and cache tuning or token optimization/tweaks to thinking are the biggest culprits today.

  • 评论信号:"Nerf"ing models isn't real in the vast majority of reported cases. Benchmarks like this or the 100 other "let's see if nerfing is real" copies would have shown it by now if it was.I made a graphic to explain why people feel like the models get nerfed:https://x.com/thesilenceturns/status/2103551351825543610The idea is that new models can handle up to a certain level of complexity, at which point they fall apart. Every new model can handle more complexity, so there's a wonderful time upon release when you feel like you can do anything, only for you to hit the complexity ceiling a few days later when you saturate it. Rinse and repeat for the next model. reply: Incorrect.Anthropic has admitted to nerfing in the past. There have also been inference bugs. On top of that, model performance changes as they move compute to schwaggier providers as well.Your chart is wrong.

它进入 HN 前列本身就是一个社区信号,但这还不是结论;更可靠的判断来自原文细节和评论区反例。

这条记录目前缺少模型生成的深层解读。更好的阅读方式是先问:它的热度来自真正的新信息、可迁移的方法,还是只来自标题与时机。

可以先读原文第一屏和 HN 最高赞评论,再决定是否值得重新生成完整分析。

ᕼᑎ #2postalemployeenetwork.com187 pts111 commentsProgrammingworth reading

U.S. postal inspectors shut down website selling counterfeit postage labels

echo@mulan ~/hn/story-2

HN 热门故事「U.S. postal inspectors shut down website selling counterfeit postage labels」进入今日前列,值得先打开原文和讨论串判断它真正有价值的部分。

模型分析没有产出可用结构化结果;页面保留了 HN 热度、原文入口和讨论信号,避免用空泛总结替代一手材料。

它在 HN 上获得约 187 分和 111 条评论,说明这个话题至少触发了社区讨论;真正的判断仍要回到原文证据和评论区的分歧点。

这是一条降级分析:它不冒充完整解读,只把可验证的元数据、原始链接和 HN 讨论保留下来,方便稍后重新生成或人工阅读。

评论区已经提供了一些读者反应,但这里还没有形成完整综合。

  • 评论信号:I am pretty sure I bought from a seller using this (or something similar) on eBay. It sent me down a rabbit hole of reading about this type of scam that exploded earlier this year: https://www.uspsoig.gov/sites/default/files/reports/2026-04/...Here's how it went down:https://www.ebay.com/fdbk/feedback_profile/theperfectpartNote that they’ve sold so much that even 1k neutral ratings are barely registering - they have “100% positive feedback”I bought a cheap tool from this seller on ebay. Tracking number was provided showing my town as the destination. About four days later, I get an unprompted ebay message from the seller: “I monitor all my shipments and sometimes there are problems so I’m proactively sending you a new one, here’s a new tracking number.”The original tracking number showed a message I’d never seen before “Item seized - counterfeit postage”. The new tracking number showed postage that had been bought back in January, but never entered the mailstream. Huh.... reply: Related:...

  • 评论信号:The USPIS (mentioned in the article) once came to the federal prison I was at to tell inmates that using stamps as currency was illegal.There are flats and there are books. Flats are a sheet of 20 Forever stamps that you would get at a drug store. A book is 20 stamps, used or not used.Generally there are different values associated with flats because you can get cash for them on the streets. You can get them to the streets via an officer if you're ballsy. Or you can just find discreet ways to mail them.The value of a flat where I was at was about $13-$14 depending on your vendor. A book was generally worth $10. If you were buying bulk (and depending on who exactly you were buying from), that number could go down to $12 and $7 respectively.If you were buying from someone that was running a ticket (sports gambling) they had more incentive to get rid of their stamps. Some people had tens of thousands of dollars of stamps in their property and they wanted to convert them.If you were buying from someone running a store (think "buy and resell commissary"), their business model was dependent on flipping books of stamps in exchange for more commissary.... reply:...

  • 评论信号:30 years ago a teacher told me of an art student applying to a school. The student included two art pieces for the submission but was meant to provide three. A note suggested that they look at the stamp for the third submission. It was a hand drawn counterfeit that was good enough to get it delivered.Or so the story goes. reply: Russ Swain. Art Director role.Later became a currency counterfeiter.Go figure.

它进入 HN 前列本身就是一个社区信号,但这还不是结论;更可靠的判断来自原文细节和评论区反例。

这条记录目前缺少模型生成的深层解读。更好的阅读方式是先问:它的热度来自真正的新信息、可迁移的方法,还是只来自标题与时机。

可以先读原文第一屏和 HN 最高赞评论,再决定是否值得重新生成完整分析。

ᕼᑎ #3space.bl2.net147 pts34 commentsProgrammingworth reading

Show HN: Real-time Solar System with 526k asteroids and all tracked satellites

echo@mulan ~/hn/story-3

HN 热门故事「Show HN: Real-time Solar System with 526k asteroids and all tracked satellites」进入今日前列,值得先打开原文和讨论串判断它真正有价值的部分。

模型分析没有产出可用结构化结果;页面保留了 HN 热度、原文入口和讨论信号,避免用空泛总结替代一手材料。

它在 HN 上获得约 147 分和 34 条评论,说明这个话题至少触发了社区讨论;真正的判断仍要回到原文证据和评论区的分歧点。

这是一条降级分析:它不冒充完整解读,只把可验证的元数据、原始链接和 HN 讨论保留下来,方便稍后重新生成或人工阅读。

评论区已经提供了一些读者反应,但这里还没有形成完整综合。

  • 评论信号:Oh man, it seems like my asteroid (Minor Planet 31689 - Sebmellen [0]) is missing.[0]: https://www.wikidata.org/wiki/Q6711174[1]: https://ssd.jpl.nasa.gov/tools/sbdb_lookup.html#/?sstr=20031...

  • 评论信号:Author here. It's a browser view of the Solar System at real scale, with its current state, plus the objects around Earth from the CelesTrak catalog.Data: CelesTrak TLEs (SGP4), asteroids and comets from JPL SBDB, spacecraft positions from JPL Horizons. Updated daily.Rendering is WebGL2, orbit propagation runs in web workers. The asteroid set (~30 MB) loads in the background.The time slider runs forwards and backwards; satellites appear and disappear by launch date. reply: How are you sourcing the spacecraft positions from JPL? Are you just grabbing a huge list of state vectors?My old day job was to compute asteroid positions for NASA telescopes, if you are interested in orbit prop you may be interested in: https://github.com/dahlend/keteI made an asteroid only version of yours earlier this year, where it does full orbit propagation in workers: https://dahlend.github.io/ketev/

  • 评论信号:Just spent a little time following Europa Clipper, it will fly by Earth very soon! If you zoom out from Earth, it's already pretty close. This will be its second gravity assist after Mars. Very cool site!

它进入 HN 前列本身就是一个社区信号,但这还不是结论;更可靠的判断来自原文细节和评论区反例。

这条记录目前缺少模型生成的深层解读。更好的阅读方式是先问:它的热度来自真正的新信息、可迁移的方法,还是只来自标题与时机。

可以先读原文第一屏和 HN 最高赞评论,再决定是否值得重新生成完整分析。

ᕼᑎ #4america.gov414 pts340 commentsProgrammingworth reading

America.gov

echo@mulan ~/hn/story-4

HN 热门故事「America.gov」进入今日前列,值得先打开原文和讨论串判断它真正有价值的部分。

模型分析没有产出可用结构化结果;页面保留了 HN 热度、原文入口和讨论信号,避免用空泛总结替代一手材料。

它在 HN 上获得约 414 分和 340 条评论,说明这个话题至少触发了社区讨论;真正的判断仍要回到原文证据和评论区的分歧点。

这是一条降级分析:它不冒充完整解读,只把可验证的元数据、原始链接和 HN 讨论保留下来,方便稍后重新生成或人工阅读。

评论区已经提供了一些读者反应,但这里还没有形成完整综合。

  • 评论信号:OK, there's a lot of negative comments here, but this is a great idea at a high level. It really is hard to figure out where to do a thing and it's also very easy to get phished. If they can figure out how to help people get all the services they're eligible for that'll be an amazing improvement for the people. reply: I agree in general, but how do you resolve the problem that people will take the output of the chatbot as authoritative even when it could possibly have errors or miss important nuance? People being lead astray with information from the government could lead to very serious consequences, up to and including imprisonment.Honest question, I had the same issue when my work wanted me to stick a chatbot in front of our HR portal and I never resolved it to my satisfaction.

  • 评论信号:I understand that having the fingerprint unlock icon over the text about 'keeping your privacy protected' is supposed to represent protecting 'your privacy', but if you click it, it doesn't even go away, so it just becomes an annoying impediment to trying to read. Am I alone in thinking this or is that entire paragraph just generally poorly designed?Edit: Also hope that they don't plan on having anyone using ancient software (specifically IE, which I'm sure there are probably some government agencies somewhere that probably do) having the website work - even if you use an IE user agent, it will use SSE. reply: I had the same reaction, missed opportunity for a cool UX moment from them

  • 评论信号:If anyone's wondering how this works, it looks like[1] Gemini + guardrails:> Google is proud to be named a technology partner in this vital initiative, leveraging Gemini to help more than 100 million people access critical public resources with greater speed and ease[1] https://blog.google/company-news/outreach-and-initiatives/pu... reply: This must be why I can’t provision computer in us-central1

它进入 HN 前列本身就是一个社区信号,但这还不是结论;更可靠的判断来自原文细节和评论区反例。

这条记录目前缺少模型生成的深层解读。更好的阅读方式是先问:它的热度来自真正的新信息、可迁移的方法,还是只来自标题与时机。

可以先读原文第一屏和 HN 最高赞评论,再决定是否值得重新生成完整分析。

ᕼᑎ #5github.com251 pts134 commentsSecurityworth reading

PS5 Relapse Exploit

echo@mulan ~/hn/story-5

HN 热门故事「PS5 Relapse Exploit」进入今日前列,值得先打开原文和讨论串判断它真正有价值的部分。

模型分析没有产出可用结构化结果;页面保留了 HN 热度、原文入口和讨论信号,避免用空泛总结替代一手材料。

它在 HN 上获得约 251 分和 134 条评论,说明这个话题至少触发了社区讨论;真正的判断仍要回到原文证据和评论区的分歧点。

这是一条降级分析:它不冒充完整解读,只把可验证的元数据、原始链接和 HN 讨论保留下来,方便稍后重新生成或人工阅读。

评论区已经提供了一些读者反应,但这里还没有形成完整综合。

  • 评论信号:Can I use this to make backups of my game saves to USB?Being somewhat naive to the gaming scene, I only recently discovered that the PS5 prevents you from backing up your game saves to your own physical media - something that was possible on the PS1, PS2, PS3 and PS4.Instead, you have to subscribe to PS Plus and enable cloud backups - and you need a separate subscription per user profile.I only discovered this after my daughter lost a year of work in Minecraft due to data corruption, and I tried to teach her the value of backing up.The PS5 may be my last console. reply: If things continue they may be everyone's last console. The next gen might effectively be a video player and controller interface for content rendered outside of your home. No physical games (or saved game backups) at all under this scheme, of course.

  • 评论信号:If I know anything about those communities, they sit on a bag of zero days and another bag of leads to zero days in the bootloader or whatever everjail they need to break out of. reply: I saw a post from someone who worked in the scene recently, seems they were someone famous. I don’t follow things so I’m not sure. Basically they were giving up because new people to the scene were just discovering things with AI and then immediately disclosing them to the world, allowing Sony to patch them before they really got used.It sounded like things may have changed.

  • 评论信号:At a glance, it looks like it exploits a bug in WebKit's JavaScriptCore JavaScript engine.Does the PS5's WebKit use JavaScriptCore with JIT enabled? Have to wonder if Sony will respond by disabled its JIT, to narrow the attack surface. reply: JavaScript is the gift and grift that keeps on giving. A security nightmare when in Flash, but a gate way to freedom nowadays.

它进入 HN 前列本身就是一个社区信号,但这还不是结论;更可靠的判断来自原文细节和评论区反例。

这条记录目前缺少模型生成的深层解读。更好的阅读方式是先问:它的热度来自真正的新信息、可迁移的方法,还是只来自标题与时机。

可以先读原文第一屏和 HN 最高赞评论,再决定是否值得重新生成完整分析。