The OP has already been answered in earlier posts and a discussion about RNGs is completely related to the topic. Look bud, I said "Torque does not do this" and then you posted an incorrect snippet to which I explained thoroughly and politely why it was incorrect. I don't understand how you can call Torque a hybrid language, it sticks to its data orientated OO roots pretty thoroughly. I am absolutely here to tell people how RNGs work, because a) I explained why Torque's builtin RNG would not regress and hence the OP's assumption that the RNG was to blame was wrong b) you posted a statement to which I kindly directed an explanation of why it was incorrect and again why the RNG was irrelevant to the OP's bug.
1) The "more local something the better" is a bad attitude in _most_ non-concurrent use cases, as that takes up more memory and cycles. Like I mentioned before, using it in the update function of the game is bad because it will (depending on the language) need to reallocate the random number generator each time the function is called, reseed it (which can be expensive depending on the PRNG type) and it won't be random because the same numbers will be generated each time.
2) TorqueScript does not "modify the seed to keep fully random, instead of psuedo random". Every random number generator is psuedo-random, calculating actual entropy is impossible and seeds are only used once. You keep complaining about me explaining how PRNGs work when you apparently already know -- even though you say ridiculous things like the above and are only now listening to why instantiating a RNG in a game update function is bad even though I've detailed this 2 times before.
3) hhehea u so funy