Good one, Tank.
Made me think about a practical, nechanical evolution demonstrator - then I thought it could probably be done on a computer by someone clever.
Back in the old days, 198x-ish, there was a game of very simple sprites that one could configure (within size and type (species) limits) and let loose to compete with each other. It was purely random as to who ate or "mated" who but I found it fascinating. Populations could grow or die, species could consume or be consumed. But the environment remained constant.
Change it round a little; again within simple rules on, say, five factors, let the sprites configure themselves and have the ability to mutate one of their factors at random. Then set the enviromental factors, temp increase/decrease, available food type changes etc analogs to vary over time. Then see which mutations survive the changes.
The random bit comes into play, as I understand it, in the natural genetic mutations. If the random mutation was positive for an environnental change then that mutation will be selected, naturally, over a negative one.
There is probably something like this already modelled for academic purposes, very complex and needing a super computer, but I have never seen anything sold as a "game" for home/school use on simpler machines.