Someone on a forum suggested that the reason why d20s are preferable to d10s for core resolution mechanics in tabletop games is because they're inherently more fair dice. I've read two arguments for this 1) Because d10s are not platonic solids, they are less fair (what??) and

2) The shape of d10s makes them inherently harder to manufacture in a way that is balanced (compared to d20s). Neither of these make sense to me because 1) in practice, a platonic solid shouldn't make an object any more "fair" to roll than a non-platonic isohedron (e.g., a d10).

2) IME, d20s have the highest incidences of casting errors. That's not surprising to me. My father is a sculptor and mold builder. For a given die size, an icosahedron has the smallest face size and, obviously, the highest number of faces and edges in a standard die set.

Setting aside that the physical properties of dice are irrelevant if people are playing CRPGs or on a virtual tabletop (VTT), I wanted to test this, because here's the thing: I don't really care what the theory of something is if it isn't observable in practice.

Dice are unfair, but most players will not perceive that lack of fairness on an individual die due to the relative small margins of difference.

I rolled 4 d20s and 4 d10s 100 times each. They are opaque "jumbo" Koplow dice. I charted the results.

https://koplow.ecadv.com/bk/ 

Every time I rolled the dice, I put a tally (a period) in the corresponding cell. Excel formulae helped track everything from that.

*Burn After Reading voice* What did we learn?

Well, based on 4 d20s and d10s of the same type from the same manufacturer, the avg roll of the d20s across all dice was closer to the ideal average than on the d10s, BUT one of the d20s had a much larger margin than any of the d10s.

Two of the d20s were quite close to the ideal average, while the d10s were closer to each other, overall.

The distribution deviation for all d20s was 2.10 (from a mean of 5, or 42 RSD), for the d10s it was 2.89 (from a mean of 10, or 28.9 RSD).

All of this falls into the "who gives a shit" zone for me. While using the black d20 might cause some pause because of the hellacious number of 1s it rolled, I don't think any of this falls outside of what is to be expected, nor would I base a game's die mechanics on this data.

If d20s produce the range that you want, use them with confidence. If d10s produce the range that you want, use them with confidence. Even if individual dice show bias, it's almost certain the players will never perceive any of it.

And yes, if you're using dice pools, the biases of individual dice are less likely to be impactful on a given roll.