mden

It's hard to say exactly how much credit Astra gets if its training contained the research notes of Levent Alpöge and Tristan Buckmaster. Surely it's impressive to generate the result even if working from their notes but it muddies the waters on its capabilities quite a bit.

From https://openai.com/index/navier-stokes-solution/:

> we cannot rule out that de-identified data derived from their usage of our products helped improve our models

I tried out Opus 5 on a Bevy game app and was really surprised at how capable AI has become at testing visual applications without even being prompted to. It wrote itself a mini testing harness in the form of various startup flags. Then it would use them to setup game scenarios and play through them using mouse and keyboard. It would do this while implementing or debugging features. With gameplay time acceleration as one of the flags, it became quite fast at testing and debugging.

Not to say it was perfect, e.g. sometimes it would get temporarily stuck in a testing loop or it would test scenarios that didn't necessarily seem reasonable. But overall rather effective and capable. This was for a city building game so pre-scripted builds, even if by AI, are likely much easier to create and execute than say playing an ARPG.

This seems less about frontier models and restriction and more just lack of compute capacity to meet demand. This has always been an issue for large clients running on cloud, though not to this extent.

Can you explain what your point really was then? Belittling the idea of hotlines seemed central in your messaging and the possible exploitation of data more of a secondary thing.

It's fine to be cynical but it's also good to remember that there are real people that do care and try to improve the world as well.

> React is the literal worst of all the modern JavaScript frameworks and yet that's what everybody insists on using.

That's rather amusing. "Am I wrong? No it's the rest of the world!"

How are these other frameworks so far ahead? I've used a couple of frameworks and React has been the easiest from the big ones to wrap my head around.

My (personally) humorous interpretation was that they were extending your joke by implying the only users using FF are devs and they are the ones that care.

Anyone have examples where fold leads to easier to read code than a for loop in Rust?

I think people are missing that you're being satirical (maybe there is a better word for it?). Unless I'm the one being wooshed.

"most fraudsters are not evil people, they just follow the incentives and almost non-existent disincentives"

Maybe I'm too idealistic but why does following incentives with no regard for secondary consequences not evil?

> Why invent "safety" practices and ignore every documented software engineering best practice?

That seems unnecessarily brutal (and untrue).

> 2,000 line long modules and 200-line methods with 3-4 if-levels are considered harmful

Sometimes, not always. Limiting file size arbitrarily is not "best practice". There are times where keeping the context in one place lowers the cognitive complexity in understanding the logic. If these functions are logically tightly related splitting them out into multiple files will likely make things worse. 2000 lines (a lot of white space and comments) isn't crazy at all for a complicated piece of business logic.

> Comments that say what the code does instead of specifying why are similarly not useful and likely to go out of date with the actual code.

I don't think this is a clear cut best practice either. A comment that explains that you set var a to parameter b is useless, but it can have utility if the "what" adds more context, which seems to be the case in this file from skimming it. There's code and there's business logic and comments can act as translation between the two without necessarily being the why.

> Gratuitous use of `nil`

Welcome to golang. `nil` for error values is standard.

I'm not saying you're entitled to a paycheck - a company has the right to fire you for your differences in politics. But if you frame it as a relationship, communication can help both sides as long as it's done with some mutual understanding. There are times companies listen to their employees and change their stance. And if they don't then you still have the ability to leave. That said it also matters how differences in views are expressed, I doubt there are many cases where yelling is effective..

A lot of work can be inherently political. If your company pivots into mass surveillance is it unreasonable to speak up if it's against your ethics framework?

Isn't having a private entity be able to do what previously required a nation a good sign for that nation? The idea that the USA cannot re-obtain that capability again seems silly to me and there are a lot of pros for having private companies pushing the state of art forward.

Money and tech. Their dev experience is imo better than Google. I wouldn't be surprised if it's the best out of all large tech companies.

I mean, it's not really wrong, is it? :D Have you tried the same with gpt4?

4) Throughput is probably a better metric than density and that's hard to guess from static images.

Is the gas in the car you parked publicly really yours ("private") if there is a way to siphon it out?

Just because there is a way to obtain a resource doesn't make it yours automatically..

Yeah.. but do you think most ultra-rich elite are pushing against entropy or contributing more to it or at best staying neutral while further growing their wealth?

For certain things I think so. For me the fascinating and really useful utility of LLMs is their ability to synthesize answers to questions. This is something search generally speaking can't do. I find myself using ChatGPT to answer question like:

> Explain tar -xzvf

> Answer: The command tar -xzvf is used to extract a tar archive. Here is what each option does:

-x: This option tells tar to extract the contents of the archive.

-z: This option tells tar to decompress the archive using gzip compression.

-v: This option tells tar to run in verbose mode, which means it will display the names of the files being extracted as they are extracted.

-f: This option specifies the file name of the archive that you want to extract.

So, the full command tar -xzvf is used to extract a tar archive that has been compressed with gzip and display the names of the files being extracted as they are extracted.

However for other types of information I much more care about being taken to a trustworthy source of information rather seeing a summarized view or worse, a synthesized view of various sources that might not agree with each other. References are useful, but it's risky to rely on what a bot says without validating the references at which point the utility of the bot for that type of query is questionable.

> You have no idea the wave of disruption coming.

Cool, but can you give some concrete examples so it doesn't sound like you're being extremely hyperbolic?

> Even adapting to the new world order puts their primary revenue stream at risk.

Even if Bing and Google end up having similar LLM chat capabilities with Bing being first to deliver, would it be enough to get Google users to move to Bing?

The current business models need to be more than being able to converse with a chatbot. Ads will still need to be integrated and they might come off a lot worse when integrated into a conversation. It's definitely interesting times but I don't see how it's a new world order yet.

Then you would encourage people to have behaviors that trend towards criminal so they would get paid.

But I agree with the thrust of the idea - investing that money in reducing factors that would cause people to turn to crime is a good idea. I assume it's already being done but maybe not as effectively as we would hope?

I can't speak for large games but the Bevy build-run loop is quite fast for smallish games (say 10k loc). Certainly more pleasant than say Python or JS where it starts fast and then you spent five iterations fixing errors Rust's type & ownership systems would have caught.

Agreed. Most code at the top tech companies isn't interesting or even necessarily good. The hardest part of jumping into a new code base is almost always understanding the problem it's solving rather than the technology used to solve it.

That's too simplistic of a statement. Sometimes you have to drive some segments out to keep others or to keep advertisers happy. If HN didn't put effort in curating the community would a lot of us still be here?

True, but achieving that sustainably is a problem at least 2 orders of magnitude more difficult. Let's take the wins where we can.

There are good use cases for WASM in the browser where the js shim is transparently provided for you, e.g. for game engines that export to web. As for non-web use, I believe the wasm runtimes that exist generally don't require js shims.

Also even if you don't care about performance, WASM can arguably provide a sizable security benefit for a number of use cases such as not-fully trusted plugins.

You need employees to care about their employer. This won't happen if the employer doesn't care about the employees first. I am not sure how true this is but I have a feeling a lot of the issues come from already successful businesses hiring MBAs who find and push for these worker optimizations that drive short term profits at the cost of eroding employee loyalty and company values. So maybe at the company level execs need to push back on viewing their employees as soulless automatons that need to be optimized. And at the nation level it would help if people didn't view continuous employment as vital to survival and separating health care from the employer (specific to the US) is a necessary step imo.

Or more explicitly the people who would see the message would have no incentive to do anything about it as it would likely add more work for them with no gain. I think this is in part a result of all work "efficiency" optimizations and the exact performance metrics that come with them that have been applied to many workers in the past decade or so.

The next NASA space telescope is The Nancy Grace Roman Space Telescope - https://www.jpl.nasa.gov/missions/the-nancy-grace-roman-spac....

Can you share a source? Here's a quick one I found that shows the opposite: https://covid.cdc.gov/covid-data-tracker/#rates-by-vaccine-s...