KirtCurtKert

[KirtCurtKert @KirtCurtKert]

@patrick_kuhnke_ More than 10x once you get the patterns down. I agree.

And 10x products? Shallow thinking. Is slashing the number of products. For example. Half the Mac store is useless now.

Window manager? Space optimizers? Inbox cleaners? Claude code takes care of it.

[KirtCurtKert @KirtCurtKert]

@icanvardar Harnesses are like iOS apps.

[KirtCurtKert @KirtCurtKert]

@weswinder We’re all learning. It’s a great thing. I learn a lot by token maxxing, then fold those back into more concentrated work flows.

[KirtCurtKert @KirtCurtKert]

@hendmere @RoundtableSpace Yup.

It’s going to be great. We’ll see better stories, art, mechanics, etc.

Pushing graphics is a “capture the masses” play. But graphics are good enough, we don’t need triple A graphics, we want great games.

[KirtCurtKert @KirtCurtKert]

@warraich_Ai2 @grok Is this a legit google video?

[Tim Sweeney @TimSweeneyEpic]

The games we've long wanted, which didn't exist because they'd cost several times more to make than they'd earn back, will soon become possible. As will the games developers wanted to build, but couldn't with the resources available. It'll be a new golden age of gaming.

[KirtCurtKert @KirtCurtKert]

There's going to be a 5 year period starting in 2027 where the technically inclined are making their own polished games to play fully from scratch..

Putting themselves, their own inside jokes, friends and family into the games too.

[Hendmere @hendmere]

@RoundtableSpace At some point, you can make great games with only 10 devs in a studio. AI will help them up so much. Maybe the game prices will go down, too cause of low production cost

[KirtCurtKert @KirtCurtKert]

@theo good to know.

[KirtCurtKert @KirtCurtKert]

@ForrestTheWoods Probably a use case thing.

If you’re working with 10k engineers in a mono repo you have different problems than smaller outfits.

[KirtCurtKert @KirtCurtKert]

Sapling is an iteration on git.

And when we move from fully distributed to a centralized server… we’re talking about something very different, that I would not call better.

Whatever this Scion thing turns out to be, it’s likely good for agents no doubt but git was pretty good for humans.

[KirtCurtKert @KirtCurtKert]

Good, I haven’t burnt through yesterday’s reset yet.

Working on it.

[Tibo @thsottiaux]

Day 9/

We've rolled out optimizations to how we store state on disk for the chatgpt & codex desktop app that should reduce the amount stored on disk by 60%. Small ship but we've heard enough feedback about power users running out of disk space.

[KirtCurtKert @KirtCurtKert]

@roblabonne No.

A very small percentage of people could prompt even a small todo list app into existence.

Even when it comes to product, good software engineers are lost on how to manage ui/ux in user friendly ways.

Focus Groups and hands on marketing research are still very relevant.

[KirtCurtKert @KirtCurtKert]

@theo Unfortunately I have a real use case for 3d so it matters a lot.

[KirtCurtKert @KirtCurtKert]

@madebygps You must not have ADHD.

I’ve been in permanent flow state since Claude code came out.

[KirtCurtKert @KirtCurtKert]

@nickreames @yacineMTB Yes. Keep it on a short leash.

[KirtCurtKert @KirtCurtKert]

@pcshipp Fable Ultracode. Gone on in 60 seconds.

[KirtCurtKert @KirtCurtKert]

@varun_mara @davidfowl Yes.

Agents understand and build quickly because it’s a new protocol and purpose built.

Just building a CLI you can get some haphazard legacy BS that wastes tokens.

[KirtCurtKert @KirtCurtKert]

@constexprvoid @zencephalon @eriklangille Yup

[KirtCurtKert @KirtCurtKert]

Yeah.

I have a feeling of “fake” productivity with OpenAI models. They work for hours on a goal but never seem to get the thing all the way there.

ADHD type models distracted by every little thing.

Great for short attention span stuff and research. I’d say world class on things 30 minutes or less with no steering.

Once you go longer and start steering you’re fucked.