We won’t know for sure what’s actually going on under the hood until the console is cracked wide open or there’s a devkit leak, but my speculative guess is that some details of the GPU are ‘emulated’/recompiled. PC AAA games tend to include lengthy shader pre-compilation wait times, console games don’t have that wait time because the shaders are pre-compiled by the developers when building the game, specifically for one piece of hardware. The games themselves then fully rely on those pre-compiled shaders. They’re going to need shaders that work with the Switch 2’s GPU, which is going to involve some kind of imperfect translation process.

AMD was able to design better hardware that works with older compiled shaders, as done in the PS5/Xbox Series (and Pro consoles). That’s not a super common feature, but I imagine that AMD is more motivated to keep Microsoft/Sony happy than Nvidia is to keep Nintendo happy. AMD’s graphics division might as well shut their doors if it wasn’t for the consoles, meanwhile Nvidia is raking in trillions from the AI boom and would rather forget about gaming.

It was supposed to be private between Mitchell and someone else in the scene. I don’t mean to say that it makes Jobst’s big claim true, just that Mitchell isn’t entirely absolved and Jobst wasn’t completely wrong on everything either.

The 118 page court judgment is online (I can’t expect anyone to read it) and in general, the court sides with Jobst on most things, it’s just that none of it really applies to the big “caused Apollo’s suicide” claim Mitchell actually sued over. That’s on a different level.

As it turned out though, Mitchell very much did joke about Apollo Legend committing suicide years before it actually happened. https://www.brisbanetimes.com.au/national/queensland/video-game-champion-regrets-jokes-about-death-rumour-20240917-p5kbd9.html ( https://archive.is/sYEwg )

For what it’s worth, this game was formerly “Monolith”. Fantastic twinstick bullet hell shmup roguelite. Difficulty is somewhat on the hard side but it’s learnable.

For current exports, it’s some custom .csm/.csd file combo. Not sure if there’s any tools for working with it, seems like it’d be more annoying than just using a normal archive format either way.

It technically still exists in the game properties -> installed files tab, but it doesn’t really work. The backup files you get require you to be online to meaningfully restore and will trigger a patch to the latest game version.

Practically speaking it’s better to just make a copy of the game install directory manually, gives you a better chance of things working (even though most games require some kind of external tooling for that).

Aside from all of the problems with the game itself, I think they must’ve had one of the most unfortunate launch moments. Hero shooters had been pretty much on the downturn and then just before they launched, Deadlock went public and suckered quite a lot of the hero shooter audience into playing a full-on MOBA/FPS hybrid. And Deadlock is very quietly breaking all kinds of silly records for what’s technically an invite-only alpha (currently #8 on Steam’s most played with 137k concurrent players).

Looking at the slides in the original Japanese source, this tooling also has a whole lot of analysis options and can pull/push game data/positioning both to and from a real Switch or something along those lines. Integrating that much custom features into an off-the-shelf tool would probably take just as long.

Probably an anti-piracy thing. It’s pretty common in the console hacking scene for only specific versions to be vulnerable, or only have exploits released for a specific set of versions. People can get around it by looking for games released with specific updates on the disc/cart but it’s a pain.

Browsing through the PDF, I’m getting the vibe that their way of measuring “skill” is weird. They claim to use multiple methods of measuring, they list a few obvious ones that they’ve found to be bad, but they don’t say which ones they are using because “we are constantly iterating on our performance metrics to optimize the player experience per game-mode”.

Elo-like systems tend to adjust skill based on the chance of winning current match X win/loss, but they’re not (just) doing that. I wonder if they have a few weird metrics that look good on paper/in the lab but don’t feel good in play.