What would happen if you went back in time to get a random number? For example, you go to a website that generates a random number. You get the number 8. What would happen if you went back in time a few minutes, and repeated the same actions. Would you get the same number, or would it change?
 A: There are solutions to the field equations of General Relativity which support a form of time travel, but it is restricted to closed time-like loops.
In this case you would obtain the same random number each time.
The conditions required for these solutions are not physically achievable, as currently understood, and at this point quantum gravity is still in the future. If it ever is achieved, all of the random numbers will come up '42'.
A: One characteristic of the universe, so far as we know, is that it is self consistent. If one observer measures the interval between two events, all other observers will get the same value. 
A quantum mechanical system may not be deterministic, and outcomes must be described in terms of probabilities. But if one observer measures a value, all other observers will measure values consistent with it. E.G. an entangled pair of electrons may have total spin 0. If one is measured as spin up, the other will be measured as spin down. 
"Going back in time" is usually science fiction. There are all kinds of paradoxes that matter in science fiction, but haven't been observed. 
There are theoretical paths around spinning black holes that lead back in time. But nobody has tried to follow one. So we don't know what would happen. One can speculate, but usually this leads to science fiction. At best, it is opinion. 
A: Your question brings up the ages old debate of the deterministic vs non deterministic nature of the universe.
Computers, by themselves, are incapable of coming up with a random number
because there is no mathematical function or algorithm that can produce a
random number. So, we programmers were stuck using what we call pseudorandom numbers, which are numbers generated by applying an algorithm to a seed number to turn it into a value that is as random as possible. 
Random number generators (or RNG for short) are procedures a program can call to obtain a pseudorandom number. RNGs are generally built into BIOSes, Operating systems, or can be specifically implemented by a programmer in his program if he has specific needs. The RNG uses a seed number to produce a random number that has more digits than the seed it was fed. however the same number being fed into the RNG will output the same value again. This is why it is called pseudorandom.
Now, for the longest time, RNGs  were using the computer's internal clock, which is very precise,so it returns a number that has many decimals, as a seed for its random algorithm. This worked pretty well on old, slow computers, But computers evolve very quickly, and old RNGs based on the clock speed of old PCs weren't updated, and this quickly became a problem, because if your program needed to come up with random numbers very quickly, let's say a loop that gets a random number than prints it on screen than loops again and again you'd end up with series of 4 or 5 or even more times the same   supposedly random number, following each other, before changing to another value which was repeated the same number of times and so on. (ex:22227777333399992222444433337777)  This is because the program called an old random number generator, which was built for older computers that had less precise clocks, so the RNG was only looking for the first few decimals (lets say 4) of the number returned by the clock, but on today's fast computers that have far more precise clock (that have let's say eight digits instead of 4) and run many more cycles per second, when the program gets to the next iteration after printing a number and calls the RNG again only the last of the 8 digits of the clock has been incremented, but the first 4 that the RNG is looking for have remained the same, hence the same value is fed to the RNG as a seed, so it outputs the same value as before.
So, now that you understand how classic random generators are closely tied to TIME itself you'll easily understand why if you went back in time let's say before the year 2000 you'd end up in a period where the random number was directly tied to the date and time, and so using the same device at exactly the same date and time (to Planck time scale) would end up with exactly the same number.
Now, this was a MAJOR problem for us programmers. RNGs are everywhere, from   Determinating who will win at a Casino's slot machine to who will win that ultra rare and expansive piece of gear in your favorite MMORPG those pieces of software need to be reliable, because an exploit could be costly for the operators of the hacked device. This is why in the recent years vendors have begun selling web based random number generation services which are based on
what science considers being true random events. For example, a service called HotBits generates random numbers by timing successive pairs of radioactive decays detected by a Geiger-Müller tube interfaced to a computer.
In my sense it is difficult to imagine a more random value, and programmers can call that service within their program and get real random numbers through internet.
So if you traveled back to more recent years, and requested a random number to a computer using such a service at exactly the same point in time, I'm not certain if you'd end up with a different value, IT comes to ask if the nature of the universe is deterministic or not, or if Laplace's Demon would work which is an age old question that unfortunately does not yet have an answer... Einstein and Bohr argued a lot about this....
