As i see it, there are pretty much 2 cases for GMOs right now, and a third that worries me.

1, editing down a sequence to remove unwanted genes. Now, corn without gene X may be toxic. But, that seems unlikely. Billions of people have eaten corn one way or another, and there's bound to be a mutant corn without gene X that people have eaten, just due to random mutation. For this case, GMO's seem very safe.

2, combining genes from 2 things. Now, perhaps feature A weakens us in some way that feature B then exploits and becomes dangerous. Even then, as you say, our biology is fabulously complicated, and has worked solutions for a whole ton of problems. A potato with a cow gene doesn't see obviously more risky than eating potatoes and cows. For this case, GMOs seem a bit more risky, but not substantially more risky than the editing case.

3, Genes created from the imagination. Our biology, specifically digestive system, has had to deal with the prior cases as long as there have been things eating each other. Creating entirely new sequences that our biology has never seen, well, that's scary. we don't have 4 billion years of evolution to rely on.

So, to sum up, case 1 and 2 seem reasonably safe. At least the kind of thing we can try out, if there are bad effects, we can outlaw those specific bad sequences. case 3 is really scary, and those types of GMOs ought to go through something like clinical drug trials for widespread adoption.

edit

by safe, i mean safer than driving. likely far far safer than driving.

because biology is fabulously complicated, heuristics are the best we can do right now. The above, or more sophisticated variations seem reasonable. we don't know for sure, furthermore we can't know for sure. Perhaps a limited release/clinical trial style is better for case 2. But, there's a ton of those kinds of foods already, they've appeared safe so far. Today, that seems like a lot of effort for not much value.

1) "Ye cannae change the laws of physics". We're pretty sure that there isn't any molecule that you can flip around to have unicorns start popping out of thin air[0]. We have some bounds on the amount of damage one can do.

2) Yes, biology is messy and complicated. It's also extremely important that we master it. So there's little choice but to push forward.

[0] - well, unless it's a molecule your VCs know how to monetize.

Sure, but every interaction is a hard fought lesson at the moment. If you naively chart it out we're going to be spending literally eons figuring this stuff out.

Now I realize that advances in a bunch of technologies means that we'll know way, way more than we do now in probably 20-30 years. Not just 50% more, not 500% more but probably a thousand times more.

But if there are a million molecules, there are on the order of n! permutations which give you a rough idea of the number of possible molecular interactions. Maybe instead of P(n,n) (which means that all molecules could react with all other molecules all at the same time for one reaction) you'd be talking about P(n,n-5) or so, but 10^6^5 is still 10^30 or a really, really big number. It'll overflow a 32 bit int, a 64 bit int and it uses up a substantial fraction of a 128 bit int.

People tend not to understand that biology isn't just a little squishy or weird, or "hard", the math makes the fact that we understand anything a tribute to human intelligence. It's not finding a needle in a haystack, it's finding a needle in a galaxy's worth of planets that are all covered in nothing but haystacks.