(I think it was discussed on LW and written up on gwern.net before)

No, just LW: http://lesswrong.com/lw/5qm/living_forever_is_hard_or_the_gompertz_curve/

As for the amyloidosis thing - it's interesting but the sample is very small and I don't understand why Coles thinks there may be causality rather than him just finding a random biomarker correlate (none of his GRG emails help here either).

I'm also a bit skeptical about claims of the centenarian death rates flattening out and ceasing to follow a Gompertz curve since this sort of dieback may be fitting the population better to the curve...