Furthermore, writing down the contract calmly, as part of a plan, can avoid having to bang it out in frustration and leaving a bad taste.

Way too much work :( At least my video collection is mostly miniDV.

Unfortunately, the author spent so much time just ingesting and managing the video that he did not get to the fun stuff you can do in 2026: index, query, and restore it.

Capturing off Video8 is no different to capturing off MiniDV, if you use a Digital8 camera.

I shoot on full-size DVCAM as well as HDV occasionally. Yes, it's 576i or 1080i depending on which you use, but you'd be surprised how good that looks with a decent lens and a bit of care in shooting.

The bad old days of de-interlacing and post-production with avisynth. I've shot and watched more videos since cell phones, thanks to the immensely improved UX.

How is spatial reasoning useless??

I'd rather say it has a mind of its own; it does things its way. But I have not tested this model, so they might have improved its instruction following.

Well, one thing i know for sure: it reliably misplaces parentheses in lisps.

Clearly, the AI is trying to steer you towards the ML family of languages for its better type system, performance, and concurrency ;)

And how does the book suggest countering the problem?

To address the issues of an automated entity functioning as a detractor? I don't think I can answer that specifically. I can brainstorm on the some of the dimensions the book talks about:

- societal norm/moral pressure shouldn't apply (adversarial actor)

- reputational pressure has an interesting angle to it if you think of it as trust scoring in descentralized or centralised networks.

- institutional pressure can't work if you can't tie back to the root (it may be unfeasible to do so or the costs may outweight the benefits)

- Security doesn't quite work the way we think about it nowadays because this is not an "undesired access of a computer system" but a subjectively bad use of rapid opinion generation.

Current agents "live" in discretized time. They sporadically get inputs, process it, and update their state. The only thing they don't currently do is learn (update their models). What's your argument?

Brought to you by the same AI that fixes tests by removing them.

Can you show one?

The end result would be a normal PPT presentation, check https://sli.dev as an easy start, ask Codex/Claude/... to generate the slides using that framework with data from something.md. The interesting part here is generating these otherwise boring slide decks not with PowerPoint itself but with AI coding agents and a master slides, AGENTS.md context. I’ll be showing this to a small group (normally members only) at IPAI in Heilbronn, Germany on 03/03. If you’re in the area and would like to join, feel free to send me a message I will squeeze you in.

It really is infuriatingly dumb; like a junior who does not know English. Indeed, it often transitions into Chinese.

Just now it added some stuff to a file starting at L30 and I said "that one line L30 will do remove the rest", it interpreted 'the rest' as the file, and not what it added.

The purpose of communication is to reduce the cost of obtaining information; I tell you what I have already figured out and vice versa. If we're both querying the same oracle, there is nothing gained beyond the prompt itself (which can be valuable).