I like to call this style of writing "advanced technical claudish"

I use claude and codex to write nearly all my code today but i can't accept using it to write prose. I still handwrite every single document I have to write and have them review and propose changes, i feel like if i can't put my own ideas into words for people (and claude) to work on i've completely lost it.

Just because scales are grouped by 128x128 tiles, does not mean you need a single compute tile that large. It works completely fine to process it with multiple smaller tiles that get given the same scales, like how this works on Hopper and Blackwell today

Wait, since when does it matter whether something being hyper-optimized is useful? The computer going brrrr on an interesting problem is in itself the goal!

That's fair, I just figure there are useful scenarios as well. Apologies if I came off as dismissive!

It didn’t come off as dismissive to me. I was curious as well as to where such optimizing helps and knew that the answers to your question would help me discover use cases I didn’t think of

It's not for slower devices, it's for lower data transfer bills for providers like YouTube.

It's a win for consumers as well if you can get better video quality or more reliable calls on a slower connection.

Sadly, Windows cannot do that. Every installed program has full disk access by default. It's very, very difficult to make it not so.

Windows has had that feature for 9 years. https://learn.microsoft.com/en-us/defender-endpoint/controll...

This is implemented the wrong way around. Each program should only have access to its own folders by default, with it being possible to grant additional access. Also, I don't believe Endpoint stuff is included in the normal Windows license.

Microsoft has refused to allow any kind of persistence of these sandboxes making them absolutely worthless. Such a waste of an otherwise good feature.

Windows Sandbox is not for persistent apps

AppContainer (e.g. used in uwp or msix)

Can you configure that as a user for an unsafe program you want to run such as an online game? I think not.

Maybe it isn't built-in, but most Windows user I've worked with, including myself, have been using Sandboxie for probably two decades at this point, probably hard to find any Windows software that is more ubiquitous than Sandboxie in developer circles.

Sandboxie is essentially a giant pile of fragile hacks on top of a Windows API that does not want to be used this way. Does it seem like it works most of the time? Sure. Has it had bypasses? Also yes. I've used it in the past but I don't truly trust it.

There need to be better instructions that you have to open the web page on your computer first, opening the link on the phone directly appears useless as it just displays a qr code on the phone.

My bad thank you for the useful feedback, I've added this :D

Now when you start a game in mobile, it'll explain clearly this is a two screen game and it should be opened on desktop/tablet

Btw, I am working on a single screen mobile only version too now thanks to yours and a few other comments

Bruh. It's getting hard to track down all these MAKE_IT_ACTUALLY_WORK settings that default to off for no reason.

For me it's gotten to the point where I have a wrapper script that applies like 5 environment variables and even patches the system prompt strings prior to every Claude Code invocation.

After the Claude Code source code leak someone discovered that some variables are read directly from the process environment. Can't even trust that setting them in ~/.claude/settings.json will work!

I've actually started asking Claude itself to dissect every Claude Code update in order figure out if it broke some part of the Rube Goldberg machine I was forced to set up.

That's the beginning of Googlification of feature evolution, via population statistics rather than quality.

If it increases a KPI by 5% for 95% of users but torpedos the experience for 5%? Ship it.

that sounds like a win-win to me.

on one hand 95% of users get an improved experience. While a competitor gets the chance to build a business for the remaining 5%.

You mean the 33k bots that created a nearly linear stars/day graph? There's a dip in the middle, but it was very blatant at the start (and now)

I didn't know we could change the base system prompt of Claude Code. Just tried, and indeed it works. This changes everything! Thank you for posting this!

But you can't. Many times I've seen claude write confusing off-track nonsense in the thinking and then do the correct action anyway as if that never happened. It doesn't work the way we want it to.

Maybe, but I’ve seen the opposite too.

In most cases, I don’t use the reasoning to proactively stop Claude from going off track. When Claude does go off track, the reasoning helps me understand what went wrong and how to correct it when I roll back and try again.