The so called random function in any computer is more properly called pseudorandom. Output depends on what you seed the generator with, so there is one opportunity to insert some mischief. The other problem with DKO's approach is; random results have no memory of what was last to come up. In a selection of 100 items; and using Excel's random generator, I'd be very surprised if the split was not closer to 60/40 than 50/50. Whether it would favor the 1s or 2s is anyone's guess.