Real world usage will be probably different, and maybe even unexpected by the authors of this research.
Or from another angle the end-user is a game developer trying to actually work with this kind of technology, which is just a nightmarish prospect. Nobody in the industry is asking for a game engine that runs entirely on vibes and dream logic, gamedev is already chaotic enough when everything is laid out in explicit code and data.
I’m no fan of academia, but it undeniably produces useful and meaningful knowledge regularly.
Of course they don't. Stuff like this is a proof of concept.
If they had a product that worked, they wouldn't be in academia. Instead, they would leave the world of research and create a multi billion dollar company.
Almost by definition, anything in academia isn't going to be productized, because if it was, then the researchers would just stop researching and make a bunch of money selling the product to consumers.
Such research is still useful for society, though, as it means that someone else can spend the millions and millions of dollars making a better version and then selling that.
Also, games introduce randomness in a controlled way so users don't get frustrated by it appearing in unexpected places. I don't want characters to randomly appear and disappear. It's fine if bullet trajectory varies more randomly as they get further away.
People play and enjoy many games with varying levels of randomness as a fundamental component, some even professionally (poker, stock market). This could be made such a game.
Which also means an off-the-shelf deterministic engine.
Wanted to jump on the platform? Wow too bad, model had a little day-dream and now you’re floating somewhere. Wanted to peak a corner? Oops too bad, model had a moment and now the wall curves differently. Trying to pick a shot-well you’d definitely hit the other player if the model wasn’t making physics go out the window. Inconsistently.
Consistent bugs you can anticipate and play/work around, random ones you can’t. Just look at pretty much any speed running community for games before 1995.
Say goodbye to any real competitive scene with random unfixable potentially one off bugs.