It makes good sense that there is a clock that resets itself after a certain amount of time, and is used as the actual "luck" factor.

Though at the same time the Random function used in C and Java both use time as a seed, and when time changes the random sequence also changes. Therefore, I would question if the success would be below like 60/100 of the time and not like randomly somewhere in the middle?

Like say the random number selected is 70 and the success chance is 20% then the system would use 70-10 to 70+10 and the success range. and then you would get a success if you were between 60/100 to 80/100 of the "time".

Do we know if that system is a possibility? Because if it is then it becomes difficult to test the rythm.

---------- Post added at 05:52 PM ---------- Previous post was at 05:42 PM ----------

Let me also add something possibly useful to the mentioned theory, as stated by the OP, if his/her theory is correct then there is a certain time after which the "clock" will reset and allow for reinforcement success again.

Let us take some considerations here, the clock cannot reset after 1 hour, or even 10 minutes as then everyone refining in the first few minutes will get success, which is unheard of. At the same time let us assume that the clock time is not too shot like 1 or 1/2 sec, as this will cause reduce the chance of getting to +9/10/11 nearly impossible.

Therefore, a safe range of time to assume would be closer to around 2-10 seconds, and as already mentioned by a poster, 5 sec seems to be a good aim. Though I believe even 10 seconds or so is a possibility as the entire animation of refining takes around 5 seconds, as a result of which your next attempt could be too early or too late.

Therefore waiting around 2-5 seconds after a success might provide good results.