What about maintenance? I’d naively assume that’s the killer.

> What about maintenance?

Simply put, you don’t. Your DC is launched into its graveyard. If a chip burns out it burns out—maybe rack design is a bit more redundant to keep failures as independent as possible.

Maybe at some point repair is a valid optimization. But it’s not necessary for an MVP, namely, one that is competitive against 3 to 5-year terrestrial delays and sub-10% costs of capital for such projects. That’s what has surprised me.

It seems like that could change the math quite a bit, since you’d presumably be losing a lot of capacity to failures. I’d assume you would have a much higher failure rate in space, and component failure is already pretty common on earth.

> I thought we might finally have a high profile prompt injection attack against a name-brand company we could point people to.

These folks have found a bunch: https://www.promptarmor.com/resources

But I guess you mean one that has been exploited in the wild?

Yeah I'm still optimistic that people will start taking this threat seriously once there's been a high profile exploit against a real target.

Gemini 1.5 Pro actually has 2M!

No other model from a major lab has matched it since afaik.

Edit: err, I see in the comment below mine that Grok has 2M as well. Had no idea!

I built a tool at work that allows claude code and codex to communicate with each other through tmux, using skills. It works quite well.

Why through tmux?

The correct answer is “Postgres would handle it, but if it needed to scale even higher, I’d…”

The point of a system design interview is to have a discussion that examines possibilities and tradeoffs.

And people weeded out by this kind of questions are probably rightfully so. I for one could not ever work with someone that says "my answer is correct, period.". Part of the answer and the discussions made by mature individuals must ask for feedback, incorporate it in your design, be open to compromises sometimes but also to die on a hill when it makes sense. And in an interview context, you ought to show the hiring manager all these faces.

Then, there are hiring managers that suck and you might be discarded because you didn't follow the script. Sure, but that's a bullet you dodged.

I have a similar process and have thought about committing all the planning files, but I've found that they tend to end up in an outdated state by the time the implementation is done.

Better imo is to produce a README or dev-facing doc at the end that distills all the planning and implementation into a final authoritative overview. This is easier for both humans and agents to digest than bunch of meandering planning files.

Very cool! A noob question about how models handle video: do you do everything via sending frames as images to the model at some framerate? Are there tricks to avoid what it seems like would be massive token use from this approach?

I’m very pro AI coding and use it all day long, but I also wouldn’t say “the code it writes is correct”. It will produce all kinds of bugs, vulnerabilities, performance problems, memory leaks, etc unless carefully guided.

So it's even more human than we thought

There are different kinds of tribal knowledge. Some is company-specific, some is role-specific or domain-specific.

The date is just a useful fiction to:

- Create urgency

- Keep scope creep under control

- Prioritize whatever is most valuable and/or can stand on its own

If you just say “I don’t know” and have no target, even if that’s more honest, the project is less likely to ever be shipped at all in any useful form.