With Boost 1.33 the release notes say that there is improved initialisation for mersenne_twister. Does this mean that the pseudo-random sequence from the same seed will be different with Boost 1.33 to Boost 1.32? The reason I ask is that I use the random sequence to add noise to a signal and check how my algorithm handles it. As it's pseudo-random noise I can expect the same result each time and test for that expected result. It appears that with Boost 1.33 the results are very subtly different. That's fine, but I want to be clear that it's due to a change in the pseudo-random sequence (if it is) rather than some other regression. Richard