How does Warp/Openwarp relate to Opencode or Claude code?

They all use LLMs as a backend for certain features?

The problem is "just". "Just" getting pytorch to work and to work well is a huge undertaking.

Just, in this case means “at minimum” or “first and foremost, no excuses”. I obviously understand this is a huge undertaking. Nobody said attempting to be competitive with NVIDIA in AI would be a walk in the park.

for a trillion dollars, they should be able to figure it out.

In my opinion there is also a deficiency of the models who should be able to say "I don't know" when asked for something unreasonable.

Can't wait for people to migrate to open tools (opencode/openrouter). This will unlock a lot of innovation.

(I know openrouter is not open, but it allows competition and should be easily replaceable if needed)

And risk of loss of control on the software ecosystem.

This is the big one IMHO. Apple is all about control of the stack, top to bottom. Any sort of "help" with linux on macos would be threatening to that control. Apple "helped" even more than I would have expected by not locking alt OSes out of the bootloader. Probably for less than altruistic reasons, but they did do it.

loss of control, fewer services sales w/o their built-in upsells etc… tragedies

Go team Asahi!

> Is the wearable accurate enough to be sure that 3bpm is not a measurement fluke

If the statistical tests show significance (and are valid), the answer to this question is yes. If you have enough data you can make strong conclusions even witwith imperfect hardware.

Not at all, if you have a lot of data coming from imperfect hardware (which can mean both a fixed bias and unknown variance), and you don't know the variance for plenty of practical reasons, you are left with a result that is statistically significant, but wrong

Unless the effect they're measuring is that the wearable measures differently in sauna days.

Strong conclusion that the hardware is precisely imperfect?

There's a difference between trusting Anthropic and trusting random mac owners.

I know where my answer lies in that; but i don't claim to be an objective truth.

For example OpenAI has been caught sharing data with the gov. agencies.

Note that you don't need YouTube's permission to remove shorts from the homepage; there's a Firefox extension for that.

The good news is that, apart from the models themselves, we don't need much from these companies:

- Use Opencode and other similar open-source solutions in place of their proprietary harnesses. This isn't very practical right now because of the heavily subsidized subscriptions that are hard to compete with. But subsidies will end soon, and with progress in inference, it should be very doable to work with open-source clients in the near future.

- Use Openrouter and similar to abstract the LLM itself. That makes AI companies interchangeable and removes a lot of any moat they might have.

> You are responsible for security. I saw good devs skipping basic SSH hardening and get infected by bots in <1hr. My go-to move when I spin up servers is a two-stage Terraform setup: first, I set up SSH with only my IP allowed, set up Tailscale and then shutdown the public SSH IP entrypoint completely.

Note that you don't need all of that to keep your SSH server secure. Just having a good password (ideally on a non-root account) is more than enough.

Disable password auth and go with key based, it's easier and more secure.

I'd call it unnecessary exposure. Under both modern threat models and classic cybernetic models (check out law of requisite variety) removing as much surface attack area as possible is optimal. Especially disabling passwords in SSH is infosec 1o1 these days. No need to worry about brute force attacks, credential stuffing, or simple human error, which was the cause of all attacks I've seen directly.

It's easier to add a small config to Terraform to make your config at least key-based.