manmal

I like sprinting a couple times up our hill, ca 10s each, 4-6 times. It’s full out, and when I‘m untrained (like now), I‘m quite exhausted afterwards.

But the dopamine boost I get from this is immense. It lasts 24-48h. I‘m in a general good mood for this period. Sometimes, a little more competitive than normal - that’s a downside.

Normal jogging or long walks don’t have this effect, at all.

When it takes longer than 36h, just revoke and submit again. It's probably just stuck in some queue.

Exactly my thoughts today. They have to make it cheaper after demoing what’s possible initially.

Depends what actors you trust. They are equally bad but personally I‘m way more comfortable with the Discord story.

If you came at it from the perspective of imitating what the human brain does, we now have a very very powerful speech center and short term memory, and vision catching up. The other parts are missing. I‘m sure that’s being heavily researched.

I‘m using mostly the browser automations for things like that now. Same for research - ChatGPT is banned from reading many pages, but Codex can read anything I can.

But isn’t it funny that Cloudflare is blocking AI on their pages, but on the other hand is researching and marketing things like „you can put a browser in a CF worker“

Have you? It’s a dumb model that gets a lot wrong. Codex voice is IMO only good for when I cannot dictate into the good models. The delay introduced by letting it work with a good model kills it for me.

I guess `caffeinate` doesn't go that far because it can easily lead to overheating.

I think I'd use Discord instead. Clankers are happy to set it all up for you.

If author runs the local model for privacy reasons, then I don't understand why they give Telegram access to all their conversations. It's well known that Telegram doesn't end-to-end encrypt bot accounts.

The Amphetamine app from the App Store keeps a MacBook awake even with the lid closed.

They are better at many things I‘m a novice at. At least at the operational level. I can’t rely on the semantics being fully correct because they tend to get subtle things wrong or just don’t ask. Like tax forms - not a good idea to put them on full auto. You‘ll leave money on the table.

Codex with computer use, yes. GP mentioned it.

> Codex requires you run the desktop app and additional authentication requirements. The result of this has been codex is almost always relegated to fleet worker rather than orchestrator.

I’m often using my MBP but also the Mac Studio remotely via the mobile app. What’s missing in your mind?

One of the Codex employees on Twitter promised (threatened?) that the next gen of Codex will be cloud focused, and less local. I wonder now if that means it will be a more powerful GPT Work.

I got a 404 when heading to the examples.

I haven’t used Haskell much. I thought to try it with agentic help, but then read this post that its compilation times might make it unviable. But what’s your experience?

They have exactly one guy as the expert in this video and he contributes all the facts. That’s bias.

That video is quite biased. Just tell your clanker to critique it. It will point out that those numbers are quite cherry picked, and leaving out important details. E.g. the „GHG would reduce by only 2.6%“ study assumes that all crop production stays the same, which doesn’t make any sense.

My problem with memory is that it goes stale, and updates to facts are often not changing all locations of that fact. Such a system should make it easier to maintain a single source of truth, and versioning, no?

I guess you also get very high bandwidth that way? I‘m not sure that would come for free though.

Isn’t it just for development purposes? I see it as a web sandbox for iOS development.

They aren’t reasoning though, are they?

How do they like it, compared with 3.8 and DS4?

I didn’t expect an answer I can actually agree with - keep up the good work ;)

Sorry for the snark, but are you trying to cure cancer? What could possibly need 24/7 research in our domain, that doesn’t need your input every 30 minutes?

A Mac Mini is easy to match by a PC in terms of inference, and the PC will win without getting noisy.

1-2 RTX5090 will be better value than Macs because they have the memory bandwidth for somewhat fast local inference.

The upgrade to 6E was a big jump in my network (Wifi7 Unifi Wall AP). On the same spot, the M3 MacBook (6E) has basically zero lag over VNC, while the M1 MacBook (6) has a noticeable lag. It’s a bit mind blowing how big the difference is.

I‘m outsourcing all tedious searches to ChatGPT now. Anytime I try searching anything non-obvious, it seems to be buried in the Nth page. Both Google and DDG have this problem. Chat spares me the brain power of sifting through the endless slop and grift, and I can easily click through and read the actual content.

> There is no framework here. The entire frozen layer is seed.py

Slop language in the second paragraph. I refuse to read on.