If an expectued utility maximizer is willing to pay a cost C to get a benefit with probability ~1, it should be willing to pay p*C to get the same benefit with probability p. If C is unbounded, then so is p*C even for very small p.
it should be willing to pay pC to get the same benefit with probability p.
This was about real people, not ideal utility maximizers. Even if one agrees with "it should be willing to pay pC to get the same benefit with probability p", which most risk-averse people won't, "Any and all cost" does not mean infinite cost to most people (sacrificing their firstborn is probably not on the list, neither is killing the rest of the humanity).
http://www.cbc.ca/news/yourcommunity/2012/07/human-immortality-could-be-possible-by-2045-say-russian-scientists.html
The nice thing about Russians (I'm from that neighborhood originally) is that they are absolutely crazy and will try just about anything. They also probably have/had second-best science culture behind US (though they suffered significant brain drain as huge numbers of educated Jews left in the last 25 years). They have less regulation and quite a few rich people with ideas. Seems like a worthwhile group to keep in touch with.