Gpuccio had this weird argument that if the FI for some function(performed by some class of proteins, say) was 500 bits or more, then it couldn’t evolve, because then sequences capable of performing the function would be incredibly rare in sequence space and thus evolution would not be expected to be able to find them. And then supposedly he could infer design because evolution is the only other game in town. Something to that effect.
The whole argument hinges on this idea that evolution must happen on one of these extremely rare sequences by a lucky guess. It is essentially built into the argument as an assumption that these rare sequences can’t be incrementally evolved towards through some alternative route where sequences have other, or simpler/more likely functions.
Gilbert is trying to get out of this by putting the burden of proof on us to show, for any given example of a protein function, that there are such indirect routes to the protein in question. There are some few cases where scientists have done that, but he can just play the “but what about this example?” game (with millions of uncharacterized protein sequences in the databases there will always be proteins who’s entire, or plausible histories and origins, haven’t been experimentally characterized.) Ultimately the argument comes down to not being able to prove evolution can do it to his satisfaction and if we can’t, “intelligent design” gets an automatic pass (nothing needs to be proven about this, to design proponents, it’s just assumed designers can do basically anything.)
Edit: Here is where Gpuccio states the if 500 bits or more → design argument:
So @Giltil is just wrong. This is what Gpuccio’s “FI method” is supposed to be doing. Finding examples of “big information jumps” of “500 bits or more” and through that ruling out evolution so you can infer design (by magically just knowing).