The gist of Taleb's argument against GMO as I understand it is that a small change in a number of small variables (i.e., genes) could have emergent multiplicative risks out of all proportion to their understood individual risks down the line, and that these could end up being catastrophically bad because ecology is all interconnected.

The problems with the argument imho is that a) it isn't at all obvious that these variables will necessarily interact, and b) if they do, it isn't obvious that such an interaction will be any more likely to be catastrophic than would be leaving things as they are. The variables as they exist now could still have a catastrophic effect at some time in the future (we'll never know until it happens because it's impossible to predict all the interactions, which is kind of Taleb's starting point for saying we shouldn't mess with the unknown).

It's like saying you have ten jars full of 100 marbles each, and each jar contains one or more marbles that fit into a piece that if combined becomes the 'giant doomsday marble'. You pull one marble from each jar simultaneously and if you pick 10 doomsday pieces at the same time you create the doomsday marble and the world implodes. His argument is analogous to saying that we COULD be increasing the number of doomsday pieces in each jar by GMO, thus increasing the chance of the world imploding. But with all the maths in the world it can't be measured objectively because it's impossible to identify the doomsday pieces' interactions by definition. Which he then uses as an argument that we basically shouldn't accept any risks with any GMO because it is akin to creating these potential doomsday marbles in different jars (I'm paraphrasing).

tl;dr Taleb seems to be talking out his ass a bit on this one.