criemen

There is a long interview with Anders Hejlsberg here https://www.youtube.com/watch?v=cywK3XYYJ2o&pp=ygUbYW5kZXJzI... where he talks about that, in short from what he said 1-2 months ago, I doubt that they will change directions.

On the other hand, Francis Halzen reportedly never went down to the South Pole to the facility at all, as probably the only person working on IceCube :D

Let's see: 1. paper 2. cardboard 3. organic 4. metal 5. residual waste 6. batteries 7. hazardous waste 8. garden waste 9. plastic 10. indoor wood 11. large items (i.e. furniture) 12. glass 13. electronics

All of which is picked up from my apartment buildings backyard.

> all models have search capacities these days

one would hope that they disable websearch and internet access (maybe all tools?) when doing generalization testing?

> In addition to things like battery-backed smoke detectors in every bedroom, etc, which need periodic 9V replacements on a schedule.

Just as aside, lithium-battery powered smoke detectors are a godsend. The battery lifespan lasts the recommended life span of the detector, i.e. around 10 years. No battery replacements needed, although you still ought to test them once in a while.

Just 5? Lucky you, here (depending on how you count) we're doing 13 fractions.

It doesn't even make sense.

> On Apple's unified memory that transfer compiles to almost nothing. It is still written down, because data locality should be provable by reading the source rather than by profiling the binary.

How does profiling the binary play into this? Elsewhere the hypothesis is "it crashes", whereas profiling is about performance.

> Most languages treat the accelerator as infrastructure: you write math, and a large opaque runtime decides how to ship it. Vx treats it as semantics.

What does that mean? Vx doesn't treat the accelerator as infrastructure? Vx doesn't have a large opaque runtime? "You write math, Vx treats it as semantics". I don't understand that.

If launch costs are assumed to be $0/kg, how does that solve the waste heat problem? Add more mass? What about the maintenance of failing hardware? I'm genuinely curious, I thought basically heat radiation was a hard physical boundary, regardless of launch cost.

That's a bit out of context, no? The preceding sentence is > Even DeepMind, with its exceptional budget, couldn’t build a machine that reliably beat the best human players.

so the contrast here is the budget available, not the quality of the talent, if we accept the premise that DeepMind and the universities have approximately similar level of talent.

Given how flawed the benchmarks are (see for example https://epoch.ai/benchmarks/deepswe/review), not even the mythical 10x developer is gonna get 100% on these benchmarks.

They also have a distinct style/naming convention, see for example `a_list_entry` as type name.

Wikipedia claims they have 6 employees, so they didn't exactly enjoy exponential growth over the years ;)

My kettle (Fellow EKG) has an annoying display bug. With a simple prompt, Astra went and downloaded the firmware, disassembled it, interpreted it and confirmed the bug. It also told me that there's a firmware signature check, so I have unfortunately no way of flashing a fixed firmware myself. I'm not sure I'd be adventurous enough to do that anyways, as I don't know enough about ESP32 microcontrollers to verify what it's doing vs. the chance of bricking my device. Yet,

> but openai went kamikaze

I don't quite understand your point here. OpenAI has a consistent history of releasing cheap/small models - first nano/mini, then luna/terra. Of course, those are now more capable than half a year ago, but I don't see a behavior change from OpenAI here.

The latest Haiku release is almost a year old. Clearly they don't care about the small-but-capable part of the market at all.

> to what end?

I'd expect that their business strategy is to compete in more markets, and if successful, they can capture more value. This is the "easiest" for them as they already have GPUs, a training environment etc. For that platform it's not the worst if there's an internal customer team that can help shape the future and provide immediate feedback, and if it results in a good model, even better.

Other things I'd not be surprised they offer in the future in the same vein: A multi-model harness, coding agent (cloud and local), and maybe at a later point in time even a CPU-only cloud compute product.

My apologies, and it's too late to edit the parent now.

To be fair, though, I mainly ran this experiment to distinguish models, not to see if they were better than humans or not, so for that purpose it didn't really matter if the basic conditions were the same or not.

Thanks! I went back, and had a look at both the contest rules and the highest scoring submission again. Turns out I benchmarked against infinite games, whereas the contest measured a 5min time limit. So, Astra solved a different problem.

Comparing strategies, though, Astra does something pretty different from the highest scoring submission. It doesn't predict the RNG, instead it reacts to the visuals on-screen, planning ahead by estimating velocity of all objects on screen. Unlike the winning solution, it actively flies the ship, whereas the winning solution basically only rotates and teleports. So Astra behaves more like a regular "perfect" player would, rather than one that breaks the PRNG.

A German magazine (c't) ran a Asteroids programming contest in 2008 - create a client with access to the emulator output of the game that provides keyboard inputs to control the game.

The highest scoring submission that won the contest had a high score of around 137k. Last week, I had GPT-6 Astra, Sol and Luna implement and hill-climb on this task, as I wanted to see how big the difference in smartness is. Luna implemented something, but never exceeded ca. 20k points, with a large variance. Sol got something in the area of the humans implementation.

Astra, which finished fastest, had a highscore of around 1.7Mm when the game seemed to fairly reliably crash. On the way, it disassembled parts of the ROM to extract information about the game.

I didn't do a ton work to document and measure the specifics, but it was very impressive.

My colleague Steve Sanderson showed off VibeOS already back in June https://www.youtube.com/watch?v=7NfyZhV1dKM

Sounds like this is a more serious and phone-centered take on what Steve demoed. We sure live in interesting times.

I hope that labs are actively working on the voice and writing styles of their models. I don't want to read slop in large part because their prose is overly verbose and complicated. If the models were writing simple sentences in mainstream English, mirroring the vocabulary used by my colleagues, I'd mind less than the current status.

Surely that's something that can be trained for.

To paraphrase Churchill, LaTeX is the worst form of typesetting experience except for all those other technologies that we've tried.

Glad that that's no longer the case!

They're targeting non-fiction books, so romance novels would be out.

If you don't believe AWS follows their ToS, you shouldn't use any cloud provider for CPU/data storage/anything else either. That's not a mainstream position in the industry.

well for test runs you kinda need the binary to run, but you're correct, if the job is just "does this build" no download necessary.

If you don't need to rebuild anything, bazel can fetch only the final artifact (not the intermediates) from the remote cache.

Also, if you have persistent CI workers with a persistent bazel instance, you save on some network roundtrips, but that's obviously harder to set up and make bulletproof.

Nothing substantial, no. I've never personally experienced builds to be so slow to begin investigating bazel as an option for JS/TS.

Tsgo, oxlint, caching dependencies etc. what linear outlined in their blog post would be more impactful for the average TS project I've worked on.

I lead a bazel conversion for a pretty complex piece of software written in 5+ programming languages and shipping native binaries to all 3 major OSes a few years ago, and it took multiple years to get it done.

For a less complex project (1 programming language, still shipping to all 3 major OSes), with my knowledge and agents I got the bazel conversion done in 2 weeks.

The setup cost for bazel just went down by a lot, and I don't think the industry as a whole is aware of that yet.

Do cloud providers even share data center locations so you can assess the "far enough" bit yourself?

While I love the books as much as everyone else, the pedant in me has to point out that the Trisolareans have a four body problem at hand (3 suns and their planet), not a three body problem.