Coded Artist @coded_artist@gameliberty.club

@wolf480pl Yeah, all of this is a "we don't know what we don't know" sort of situation.

And for the same reason, we always search for something better.

@wolf480pl Yeah, that's all of technology basically.
Even asbestos and leaded fuel were good enough, until they weren't.

Even deeper than that, it's on the level of the abstract.
What is a good idea?
One that works.
But what if it stops working some time in the future?
Well, we don't know that, so a good idea is one that works... for now.

That's why prototyping is all about "fail faster".

@wolf480pl Entirely depends on your use-case, and I don't have enough knowledge in this field to make the call.

In the end of the day, due to the limitations of digital computations, it's all approximations, and you have to make the call if the approximation will suffice.
If you assume perfection, you're skipping this step, and invite disaster.
Like you said: "no one will be able to differentiate from actual random bits", is a decision to accept the approximation as good enough, and that may be a good call to make (depends on use-case of course).

An example from my field of simulations:
Is "float" a good enough data type to differentiate two points in space?
Sure... as long as you don't position anything 10k units away from the origin, where floating point precision issues arise.

@wolf480pl
True, but some assumptions are reasonable to make, such as "reality is not a simulation", while others are not, such as "my system is perfect, actually ☝️🤓".