I don't know much about chemistry since I haven't done anything with it in the last 20ish years. What need is there to be able to accurately predict it before attempting to synthesize it?

That's also something of the paradox at work at something like this: you sometimes can't have models that strongly predict interesting or "good" outcomes (such as the "island of stability") without a lot more data from experiments and maybe you aren't running the right experiments because you don't have the right model, but you won't have the right model until you run more experiments.

You could say that about the constituent elements of pretty much any chemical necessary for life.

A boulder balanced 40m above your head is lethally dangerous; vs. if mostly-embedded in the dirt next to you it's perfectly safe.

But yes, in general it's interesting how little difference there is from "essential nutrient" and "human poison"

And why it is yummy, well they are pretty useful atoms in lot of chemistry and in general balancing things that happens in body. Thus it is nice to have sufficient amount around... Best way to encourage this is to make it taste good like sugar does too.

An example I like is with nitrogen. Nitrogen atoms really want to form nitrogen gas (N2), a form that is really stable and therefore unreactive and generally harmless. However, if the nitrogen atoms are not in this form, and they have the opportunity to turn into it, they will, and they want it so much that it can be violent. That's why a lot of explosives are nitrogen-based, they are made of nitrogen atoms that have been separated from their N2 form by giving them a lot of energy, and when they come back together as the explosive is detonated, all that stored energy is released.