When trying to create something representative, it often gives a better result to over-represent than exclude, because the latter may imply there were none.

I'd argue a large reason we're having this discussion at all is exactly because the idea of black people in Britain at that time is surprising to a lot of people, exactly because a lot of people have an image of the Romans as being mostly white Italians to the point where seeing a black character at all is a surprise, rather than out of concern that they're not accurately representing the precise proportions.

dragonwriter was talking about the limitations of a numerically "representative picture" to tell a history, not about "absolutes," whatever that means. I thought it was pretty clear.

This shouldn't even be an argument, since history over-represents minorities all the time. It's not a mystery as to why. Trade and war have a couple of things in common. They feature prominently in histories because they are a catalyst for change, and they are practiced as often as not by "minorities" - ethnic, religious, racial, etc.

Tell a history well (or poorly) or try to make some kind of historical point and it is likely you are going to be talking about minorities. Over-representing them, to be sure. Taleb knows all this, and his recommendations regarding written histories have been pretty good. Regarding this thing, he was just being a combative douche. (or he has some other goofy racial agenda, as some have speculated... who knows)

And you'd have a far less than a 1/1000 chance that they were a Roman soldier. So what?