Quote:
Originally Posted by loramin
[You must be logged in to view images. Log in or Register.]
You're massively misunderstanding randomness if you think that's true. By definition any RNG that can't (when given an infinite amount of time) generate lopsided results is a terrible RNG. A truly random RNG should generate a healthy series of "1 0 1 0 1 1 0 ..."-style sequences, but it absolutely should also generate "0 0 0 0 0 0 0 ..." from time to time, or else it's not truly random. And technically all computer-generated random numbers aren't actually random ... but the pseudo-random numbers that are generated are still far more random than that.
Also, to be clear every random number you will ever see in any computer game ever is timer based. That's just how computers generate random numbers, since they can't roll dice: they check the time, do some math on that time, and generate a (pseudo-)random number.
But, that in no way guarantees that "eventually you'll get hit because eventually you'll hit the right time". It's not like anything the computer is doing is based on time in that way: it's more like the computer looks at the last digit of the current millisecond (although it's not that simple), and obviously every time it checks that last digit could keep being 7 or whatever number it needs to be to fail.
Depending on how the EQ Emulator code works it's possible that their specific implementation ensures Trak never misses more than X number of times in a row, but there's nothing inherent in RNGs that would guarantee that, and unless someone did an experiment on their own private EQ Emu server to test it out I'd tend to believe the software really does have (pseudo-)random numbers.
|
Loramin, you always immediately want to argue with me, not sure why, but you are actually agreeing with some of my points as far as I can tell.
Nearly all software absolutely does use pseudo random numbers. These numbers are generated by a deterministic algorithm. They do not have an infinite range. They are indeed tested and designed to have a 'normal' distribution, meaning no huge lopsided runs as far as possible. As far as seeds, most are timer based but not all. Linux has it's own entropy source that is built up out of a bunch of different things, because just relying on an always increasing timer is actually *not* a very good source of randomness.
If you take all possible seeds, say a 32 or 64 bit integer range, and you feed them all one by one to a RNG, and you iterate each one fully across its range, you will get every possible number they can generate. That collection is not infinite. If you then do the exact same thing again, you'll get the same collection in the same order. Your point above about a 'truly random RNG' is the heart of the problem...P99 does not have, and there does not exist, a truly random algorithmic RNG. It is not possible. That's really my point - no computer generated series of RNG numbers is actually random. Can there be long sequences of 0's ? Sure, but if there are too many then that's not a very good RNG, the whole point is to have a normal distribution. I would expect there is no seed you can feed to a modern RNG that would wind up generating nothing but 0's across the whole range.
So whatever, put too much effort into this already. Rebut away.