Another problem is that any errors in assumptions or omissions made when calculating the odds will be enormously magnified. I wouldn't trust any of these figur

Another problem is that any errors in assumptions or omissions made when calculating the odds will be enormously magnified. I wouldn't trust any of these figures unless they have ongoing efforts to test them empirically. E.g. create distributed databases of 100 trillion objects, mess with them in various ways, and perform correctness checks on them.