I think a version of the second and third possibilities is probably true.
For DEC2 and GRM1, the mutation decreases the ability of the gene to block wakefulness (DEC2) or promote tiredness (GRM1), and loss-of-function mutations seem easier to create. And indeed, I think one of the papers mentioned that there are other mutations in the DEC2 gene that correlate with reduced sleep, and GRM1 already has two identified short-sleeper mutations. Therefore, I expect that similar mutations have appeared in the past, but didn't catch on.
Why would evolution have rejected them? I suspect that, as you say, 10,000+ years ago (perhaps even 1000 years ago), being awake an extra two hours before dawn brought little benefit and basically just meant that you wasted calories; the tradeoff only became worthwhile very recently, in evolutionary terms. Perhaps there were also effects like "people had a higher parasite load, and/or were constantly scraping their skin or stepping on sharp objects, and thus had a higher constant need for some type of repair that the mutations don't accelerate".
Presumably, it depends a lot on what the gene does. If it affects how your bones grow during childhood, and you're now an adult, then you're out of luck. If it produces a protein that has effects in the present, is easy to synthesize and deliver to the body where it matters, then it's straightforward. (For example, if you want to digest lactose as an adult and lack the right genes, you can buy lactase pills apparently for 10¢.)
I'd also say that the complexity of the phenomenon may be very different from the complexity and difficulty of the cure: e.g. if you had some broken enzyme that caused your body to waste half of some essential vitamin, the consequences to your body might be very complex, but the solution might be to just take a vitamin supplement. At any rate, Wiki says that there are over 6000 genetic disorders, and that over 600 are treatable. The citation on the latter describes a database of known treatments for genetic disorders, and has some interesting numbers:
Enzyme replacement therapy (ERT) is available for 28 disorders. [...] Supplements are important in the treatment of 175 disorders [...] Medications, vaccinations, and blood products play a role in the management of 328 disorders. This includes 154 medications, 3 vaccines, and 9 different blood products.
Regarding the FNSS genes. For DEC2 in particular, it's the interaction with orexin that makes the difference (the second DEC2 paper showed that an orexin receptor antagonist turned off the short-sleeper effect), and the orexin producers and receptors are all neurons in the brain. And ADRB1, NPSR1, and GRM1 all affect some kind of wakefulness-related receptor (it's the R in all their names), which I expect are also in the brain. Delivering to the brain is more delicate than to the stomach, but it's certainly doable.
My impression is that, by trying out different chemicals, you can hope to find one that binds to the receptor and either increases or decreases the activity of it (an "agonist" or "antagonist") and ideally doesn't bind to anything else; Ying-Hui said something along those lines. That's roughly the extent of my knowledge.
Thanks! It's great to see that other people have discovered FNSS research and had similar thoughts about the implications. I have indeed submitted the post for the Cause Exploration Prize. At this point I feel reluctant to make significant changes to the post, but if there are major/glaring issues, those could be worth pointing out.
I think both Ying-Hui and I had the impression that the research had to be somewhat further along before any profit-minded people would fund it. But someone recently explained to me that Silicon Valley companies are often funded with much less scientific backing than this, so this week I've written to one venture capitalist I know, and will probably contact others. Regarding getting funding from an existing company, I don't know much about that option.
Advice is appreciated.
Thank you very much for the high-effort post and the significant findings. I agree this seems to be strong evidence that the DEC2, ADRB1, and GRM1 mutations don't work as expected in the general population. That is disappointing. A couple of questions:
How do we explain this? My brain generates "ADRB1 (or something correlated with it) works, but it requires extra mutations to work (or, similarly, there are some mutations common in the population that stop it from working); Paul married someone with concordant mutations, Brad and Janice married people with discordant mutations". Does that seem likely, or do you have other explanations in mind?