He has responded to people who critiqued his incorrect predictions on Quora before. For instance, see here. His main reply is to say that his predictions were always conditioned on adequate funding (but that’s not my impression from seeing a few of his older predictions).
I’ve been tracking him for a while and it seems he rarely changes his methodology, and usually says things that makes me think he’s not doing proper Bayesianism (such as keeping a very specific prediction constant and not changing it despite years of evidence accumulating). The analogue in, for example in AI safety, would be someone who says for 5 years that there was a 50⁄50 chance of AGI coming by 2029 and they kept this specific prediction constant.
For these reasons, I am very skeptical of his reliability.
Do you have examples of such stagnant predictions? That would indeed be strong evidence against his reliability if the accumulated evidence clearly points in one direction relative to his expectations.
He made a similar prediction for 2020 here, and I think most would agree that he was incorrect under a reasonable interpretation.
Ah. Did he own up to being incorrect and has he changed his methodology as a result? If not, then I’d be very skeptical.
He has responded to people who critiqued his incorrect predictions on Quora before. For instance, see here. His main reply is to say that his predictions were always conditioned on adequate funding (but that’s not my impression from seeing a few of his older predictions).
I’ve been tracking him for a while and it seems he rarely changes his methodology, and usually says things that makes me think he’s not doing proper Bayesianism (such as keeping a very specific prediction constant and not changing it despite years of evidence accumulating). The analogue in, for example in AI safety, would be someone who says for 5 years that there was a 50⁄50 chance of AGI coming by 2029 and they kept this specific prediction constant.
For these reasons, I am very skeptical of his reliability.
Do you have examples of such stagnant predictions? That would indeed be strong evidence against his reliability if the accumulated evidence clearly points in one direction relative to his expectations.