Here’s an idea. What if we can make a game engine that beats Unity, Unreal, Godot, Bevy, etc, and also an asset editor that beats Blender, Maya, Houdini, etc?
Not that I can do it right now, but perhaps in theory.
It’s said that the Julia language is a language where you could write scripts in the same high-level language as the engine and thus have behaviors be performant without having to hard-code things into the engine. That’s certainly a theoretical advantage. Here’s another one.
Abstractly, asset editors are things that turn user high-level specifications into executable data (interpreted or executed by the engine). That is, it is a compiler. Asset creators want to be expressive and capable of supporting pretty much arbitrary behavior, so, the asset creators want to become a compiler. Why not build it out of a compiler? MLIR is the professional infrastructure for making a compiler, so maybe it could make an excellent game engine.
Again, this is just a purely theoretical guess.
What do I do with the idea? I don’t know. The responses I get range from saying this could work, to those who don’t know, to those who think it might be over-engineered.
I’m not sure what could be done with the idea. I’m just passing along the idea. Asking for whether or not it would be a good idea seems redundant at this point.
I think I’ve been obvious enough that I did not promise it would happen. They’re floating ideas at this point.
As for what smaller stuff I did, a few gimmicks with Julia, a few Stockfish patches, and so on. But yeah… this is just like… one of my occasional ideas.
To be fair, I was painfully ignorant of the difficulty of many stuff. I tried to cheat things to become easier, like saying maybe we should design a physics system that’s easy to simulate for a game instead of sticking to real-world physics, but then the reality hit me back because designing a physics system is a hell of a task of its own.
Such is my life. I see something. It looks extremely hard. I brought in some tricks to bypass the normal challenge, and then new challenges hit me.
To be fair… this is a theoretical projection on what Julia could achieve.
Though to be fair, Julia is theoretically good at most things and desperately lacking in geniuses bringing out those theoretical advantages in many things.
It doesn’t take a “genius” to make something useful in Julia. Such statements negate the opportunity for everyone to contribute, and they might actually discourage people who don’t think of themselves as “geniuses” (aka the majority of people).
By the same token, enormously ambitious and probably impossible ideas are not the only way to move forward. Start building something, then see where it goes, and whether other people get interested. If you keep asking for feedback on vague moonshots, you will keep getting the same answers.
If I am being honest, I would say that my ideas are often more like experimental ideas that might produce something interesting even if it probably wouldn’t work, and the goalposts are like the north stars not meant to be reached but to give a sense of direction. Maybe I would attract the right kind of people if I framed it that way. However, with my limited knowledge, I am unable to assess the potential of some ideas. They’re like wildcards, perhaps worth an enormous amount of value, perhaps completely pointless.
One thing is that my ideas have evolved past the stage where I could rule out the ideas using theoretical analysis, like perpetual motion machine. I’ve reached the stage where I need to test. However, I’m not strong enough yet to test large ideas. I’ve had a few passed Stockfish patches (and a lot of failed ones) for example. However, that was possible because Stockfish patches are typically only a few lines of code. If I want to grind through more challenging ideas, I would need to be capable of consistently churning out code. I’ve had days where I could write like 500 lines of code in a day, but it’s still extremely rare. There are also probably lots of bad habits I need to get rid of, like the need to conceptualize the entire code in my mind before I start writing code. I used to conceptualize the code in my mind, then theoretically evaluate the plausibility of the code, then write code, then test, which is a very efficient optimization if you could code in your mind then theoretically evaluate, except that larger pieces of code are typically too large to conceptualize in your mind and checking the plausibility in real-life problem is sometimes very challenging. It’s not like a programming problem where you could analyze the big O of a program.
I don’t think making these exact same types of threads over and over, like this, is going to be productive for what you want. You have been given the same advice in almost all these threads as well. You are trying to be too grand with your ambitions. I advise you discontinue with the threads and try and reflect on the advice others have already spent the time giving to you.
What people are trying very hard to find a polite way of saying to you is that these threads are not useful, and it just results in noise on the forum.
People aren’t lacking in vague, unfinished ideas, those are everywhere and pop into everyone’s heads all the time. The hard part is actually doing the work.
Maybe starting your own personal blog website would be a better use of this energy if you feel its helpful for you to post it somewhere. Then people who don’t want to read it can just not follow your blog. Unfortunately Discourse has no way to mute threads created by certain users. If it did, this wouldn’t be a problem.