Thanks, that confirms my original suspicion that there must be more than two possible values for an allele or marker.
I do wish I knew if I was conversing with one or two ACs.
If we make a stack of simplifying assumptions, first that each of 20 allele (marker? what is the best term?) occurs from 1 to 20 times, with each count equally probable then the probability of an individual match would be 1/20 or 5%, and the first AC's 0.05^20 probability for a perfect match becomes a reasonable calculation.
The real world isn't that neat and I wouldn't expect any of those assumptions to hold in real life. Each marker will have its own range of possible values with its own frequency distribution. The ideal of non-correlation between marker-values won't be perfectly met etc.
I need to go educate myself.