[rayiner]

I think its an issue of discounting the return by probability, and being able to internalize positive externalities. You can spend $10 billion trying to cure cancer, but how much do you have to get in return for the risk that you won't? And if you need to make a 50x return in the successful case to justify the failure risk, how will you do it while society is demonizing you for withholding cancer cures from dying people?

[strlen]

I think part of it also diminishing returns: life of cancer patients has been greatly improved, with early detection it's no longer a death sentence and at times not even a major handicap. See, e.g., quality of life after treatment of prostate cancer now vs. quality of life after treatment of prostate cancer treatment just 20-30 years ago.

I imagine power law holds in most disciplines: going from 99.99% uptime to 99.999% is far harder than going from 99.9 to 99.99% which in turn is harder then going from 99% to 99.9%. Likewise, I'd imagine same holds true with death tolls over N years from various illnesses (but obviously with different constants involved).

Can you explain to me what you meant by externalities here? Do you mean medical industry benefits from cures without contributing to funding the NIH grants?

[rayiner]

No I mean that when you discover or invent something that "saves the world" it might have a large net positive benefit, but it might be difficult for the inventor to capture enough of that benefit to justify the risk of investing in the discovery. Say you need a $500bn payout to justify the risk of investing $10bn in curing cancer, as opposed to spending that money in something with surer returns. Curing cancer might generate $2 trillion dollar in net social benefit, but it might be very difficult for practical and political reasons for the company making the invention to extract 25% of that to justify their investment. A cure for cancer would quickly be subject to generic drug licensing by the government, under the premise that it would be wrong to profit so much from withholding cancer treatment from sick people. So why deal with that when you can get 100x return on an internet startup?

[strlen]

Ah, got it. It also seems the more useful the drug, the higher the pressure for low cost generic licensing (though, then again, the larger the circulation).

I wonder if an X-Prize like model might work better here, with, e.g., individuals with genetic risk for a specific cancer "crowd funding" various grades of prizes, with the actual awarding of prizes, setting and advertising the prizes, being done by an organization/consortium that knows what they're doing.