On: What's Next for WebGPU

[stackghost]

>To provide users a way to instantly play a game without having to download all assets at once

There's a reason QuakeLive didn't catch on and it's because streaming resources to the player makes for awful UX.

>What about users who don't know how to download a model and run it locally?

Those users also don't know how to compile MS Word from source but they have been getting along just fine with installers.

[dartos → stackghost]

The ability to paste a URL into a browser and have that be everything you need to do to play a game is pretty compelling for many kinds of games.

That’s what made flash games so big back in the day.

[pjmlp → dartos]

URLs can also be mapped into native applications, it is a matter of the right OS.

https://developer.android.com/training/app-links

[dartos → pjmlp]

That’s a different flow though.

Pasting URL into browser, then having access to game

Vs

Pasting url in browser Get link to install app Install app Navigate to the original url again.

Technical capabilities don’t always change user experience

[pjmlp → dartos]

Application streaming, no need for installation before using.

[dartos → pjmlp]

Now go ahead and let anyone embed that application in another party’s application.

A big part of what makes web successful is the ease of syndication.

[int0x29 → pjmlp]

That needs a hell of a lot of sandboxing before I get anywhere near it. Which sounds like a good use for WASM and WebGPU.