HN rank: 1

HN score: 603

comments: 384

original url: https://www.discoveryloop.com/

评论信号：From Jeff's twitter post:> Our general approach is to automate the experimental loop. We think this approach is broadly applicable across many different fields of science and engineering. We’ll initially focus on ML research and engineering, but believe the approach can help with important subproblems in nearly every one of the fourteen NAE Grand Challenge problems. We think doing this well requires strong expertise in machine learning as well as large-scale systems.See also: https://www.nae.edu/20782/grand-challenges-projectThose 14 are:NAE Grand Challenges for Engineering1. Make Solar Energy Economical2. Provide Energy from Fusion3. Develop Carbon Sequestration Methods4. Manage the Nitrogen Cycle5. Provide Access to Clean Water6. Restore and Improve Urban Infrastructure7. Advance Health Informatics8. Engineer Better Medicines9. Reverse Engineer the Brain10. Prevent Nuclear Terror11. Secure Cyberspace12. Enhance Virtual Reality13. Advance Personalized Learning14. Engineer the Tools of Scientific Discovery reply:...

评论信号：I think people are missing what this really is: Google giving some of its most senior engineers the best retirement home to keep them away from competitors. This isn’t in jest; I wish i could make enough money to not care for more from my job and then do research after i get old. Its honestly a brilliant move.

评论信号：I am siding with the "intelligence is not the bottleneck" crowd. Science takes more than reading literature and making a hypothesis. You have to run the experiment. And that is where messy reality will crush the naive, and resist any attempt to package it up into a factory-like innovation engine. But they will take your money, should you have some to invest. reply: I'm a scientist. On the one hand I take some comfort in thinking that I will always have an advantage in the lab. On the other hand I'm not taking anything for granted. And my advantage in the lab has to translate into an employer being smart enough to keep me around until if and when the AI takes over, which kind of translates into their investors wanting to keep me around.We know what happened to manufacturing when investors were no longer interested in it.

HN rank: 2

HN score: 308

comments: 156

original url: https://zed.dev/deltadb

评论信号：Zed should focus on basics. When it focuses on basics Zed is good.https://github.com/zed-industries/zed/discussions/54150 failing to show newly created files and declining to provide a refresh button, instead adding a polling backend, breaks Zed on WSL.Why a new version control system? Why not git, jj, or another existing system? reply: There should be a concept called "engineering privilege".I see some companies do things that are widely out of their core product and engineers there just having fun with a bunch of different stuff with seemingly no commitment to bring in revenue. They'll re-invent/re-write things constantly, write a new compiler/package manager for their chosen language, create new frameworks, new open source projects unrelated to their companies, etc.It must be awesome to be able to work in these conditions and still get paid. I'm not hating on it, it's how we get a lot of the good stuff we use.

评论信号：Hmm, yeah before "DeltaDB" I have a big list of things I would rather they develop/fix:- Vertical activity bar (https://github.com/zed-industries/zed/pull/58868#issuecommen... - rejected because it "doesn’t match our design vision for Zed")This is my #1 annoyance: clicking microscopic icons at the very bottom of the screen to switch views.- Don't crash my entire PC when there's a large (100+ MB) JSON file in the working directory- Don't freeze when resuming from suspension (I suspect this is GPU-driver related on linux) forcing me to killall zed-editor reply: >This is my #1 annoyance: clicking microscopic icons at the very bottom of the screen to switch views.It’s crazy that is still an issue. I really tried to use Zed because I was trying to move away from VSCode and VSCodium did not have a featureful python lsp. I just couldn’t stand little things like that that. Also, the font was really blurry. VSCodium now has Pyfly which is decent, so I just switched back.

评论信号：Why develop this when the core editor is almost unusable? On Linux wayland, copypaste is broken, the file manager is laggy and snippets do not trigger before the heavy hint logic is done. This is just on top of my head, the list is long reply: I had to stop using Zed because of this issue (which persists despite being closed): https://github.com/zed-industries/zed/issues/38109I can't use an editor where the file contents I'm being shown don't match the reality of the file. I guess this issue doesn't affect most people somehow, but in the age of CLI agents I'm not sure how more people aren't running into it.I was really enjoying Zed for a short while before I ran into this fundamental issue.

HN rank: 3

HN score: 147

comments: 63

original url: https://randsinrepose.com/archives/blade-runner-title-cards/

评论信号：Well, besides the language and the typeface, the other element of this sequence at the beginning of Blade Runner that makes it so special is the accompanying score by Vangelis, and the sound effects that start with twinkling computer notes and gradually build in volume to explosions and the whoosh of futuristic vehicles. It effectively immerses viewers into this incredible world and story. reply: It was only a few months back that I found out that Vangelis couldn't read or write music notation. He would just improv play and record on a tape, recording additional parts and mixing as needed. If he needed to have other musicians involved, then he would hire someone to listen back and write down the notation.Considering he recorded something like 35 albums, he was doing something right.

评论信号：Also touched on at https://typesetinthefuture.com/2016/06/19/bladerunner/ (and the book) reply: Came here to say the same thing.I think the Alien one is even better: https://typesetinthefuture.com/2014/12/01/alien/

评论信号：With the em dashes and the conclusion "This was not called execution. It was called retirement." I expect a lot of people would believe today that these title cards were written by an LLM. reply: You're not wrong, but part of the core problem with LLMs is that they write every blog post and README like they're writing the opening titles for Blade Runner.

HN rank: 4

HN score: 481

comments: 595

original url: https://blog.google/company-news/inside-google/message-ceo/next-chapter-ai-momentum/

评论信号：So, in last several months, all the prominent names Google lost: Demis Hassabis (technically still with google but these things are usually presented with a spin), Jeff Dean, Sanjay Ghemawat, Oriol Vinyals, Quoc Le, Noam Shazeer, John Jumper, Jonas Adler, Alexander Pritzel, David Silver, Denny Zhou, Fernando Pereira, Alex TurnerAnd all the prominent names Google gained: NULLCombined with no gemini frontier GA release in about 14 months. You have to have created an environment pretty hostile to innovation for this to happen reply: This funny post reads surreal, but it may carry some truth: https://x.com/signulll/status/2067446889956430273?lang=en

评论信号：It seems like the real news is Jeff and Sanjay are leaving Google, and Demis is effectively replacing Jeff as Chief Scientist for all of Alphabet.The bigger deal is the departure of Jeff and Sanjay, rather than Demis moving into a different role. reply: It seems there's a big shake up on the underperforming Gemini side. Before there was Shazeer (already gone) and Vinyals as co-leads, reporting to Hassabis, now Gemini comes under Kavukcuoglu reporting direct to Pichai as SVP of DeepMind.Hassabis seems to have been pushed aside. He had been CEO of DeepMind, but that position no longer exists and it seems Kavukcuoglu is now leading DeepMind with a title of SVP. Hassabis is now just "Chair" of DeepMind, and has been given the newly created title of Alphabet Chief Scientist. Are these just face-saving titles, or does he still have any real influence over Google/DeepMind's pursuit of AGI?Shane Legg remains as DeepMind "Chief AGI Scientist", but I wonder if the DeepMind founding mission of creating AGI is really intact, or if he will be next to go....

评论信号：> Lastly, after an incredible 27-year run, Jeff Dean is at a moment where he wants to try something new, and we’re excited to support him in that. Jeff and Google Senior Fellow Sanjay Ghemawat are launching an independent public benefit corporation to accelerate discoveries in ML, science, and engineering.Oof. Good for Jeff and Sanjay (who just joined Twitter), bu that is a big loss for Google. Google stock is down 5%. It might not be much of an exaggeration to say these two are worth ~$200 billion. reply: > Sanjay (who just joined Twitter)Clarification: this comment is saying Sanjay Ghemawat joined Twitter as a user recently (new account @Sanjay_Ghemawat as of July 2026), as opposed to Sanjay working for Twitter the company.

HN rank: 5

HN score: 178

comments: 106

original url: https://research.meta.ai/blog/introducing-muse-code-and-muse-spark-1-2

评论信号：Meta is offering a 10x discount on input ($0.10 vs. $1.25/Mtok) and 20x discount on output ($0.20 vs. $4.25/Mtok) if you opt in to let them train on your data.https://developer.meta.com/ai/models/muse-spark/ reply: I actually really like that pricing strategy. It's very transparent

评论信号：They chose to compare against Open AI’s mid tier model Terra instead of Sol and still lost some benchmark against it.They left Opus in and got beat in all but one benchmark.Nothing wrong with trying to improve, but why the marketing games?Instead of trying to say in the post you’re “closer” to frontier, first set a clear goal to beat the Chinese labs on price or performance and demonstrate it convincingly.Then when your ready, come back and talk frontier without playing hide the model. reply: Given the current throughput figures on OpenRouter (~180 tk/s), its likely a much smaller param count on the order of something like Luna. I think the better, more timely comparison (re: your point on Chinese labs) would be to DeepSeek-V4-Flash-0731.It's definitely confusing from a presentation perspective, but they are somewhat coherent comparisons if you account for the inference heuristics involved.(They could in theory be gaming the decode speeds with much larger than normal batch sizes given the TTFT is pretty high at around 8s)

评论信号：If you got the $20 in free credits from Meta for signing up when muse-spark-1.1 was release, please note that there's now small print stating "While using free credits your content may be used for product improvement" which was not present at muse-spark-1.1 launch when the credits were given out.If you don't mind Meta retaining your data, the "Contributor" pricing is deepseek-v4-flash-level of low, roughly 1/10th normal muse-spark API pricing currently. Attractive if you're OK with them retaining and using your data. reply: The API costs for the version of their model that feeds things back to meta is also drastically lower.