Yea, when using a timer as a seed, which in turn is accessed in software at a more or less predictable time or sequence seemed to be my problem. The light values seemed to take on a repeating color pattern, even tho that pattern may be developed over tens of cycles. No means scientific, but more like a perception thing. Keeping a log of the values, or reading them out on hyperterminal might have proven otherwise?
The human intervention aspect has to be the ticket I would think.