It’s interesting that you filed this under “Media and Popular Culture” rather than “Science and Technology” — I was (and still am) planning to cover some Singularity-related ideas in a bunch of posts under Science and Technology, so it’s nice to see you broach the topic.

That said, there are some minor inaccuracies in your account, which may be because the documentary that you watched oversimplified the ideas.

(1) Kurzweil does not claim (at least, not as of now) that the Singularity will happen in 2029. In his book “The Singularity is Near” he sets the date for the Singularity at 2045. He does claim that by 2029, machines will be capable of emulating raw human brainpower. However, he claims that this in itself will not be a singularity, it’s only when machines get to the point of being thousands of times faster than humans in raw computational power than things will change in a powerful way. Just having the same computational capacity will not be good enough if the algorithms used by machines are different from those used by humans, but with thousands of times as much computational power, even somewhat inferior/different algorithms are likely to work.

(2) In his book, Kurzweil goes into several details regarding both the theoretical and practical considerations regarding the continuation of Moore’s law for the next few decades. I’m not saying I am convinced by his account, but I certainly did not get the impression that he was ducking critical issues. Again, the documentary may have presented a more simplified version of his ideas.

I think many people are agreed that Moore’s law will continue at least for the next 10-15 years, even without buying into the singularity hypothesis. Kurzweil’s book goes into detail into trying to identify approaches that are likely to mature by that time to allow for the continued use of Moore’s law.

(3) Kurzweil also points out that if the basic exponential premise holds, then even if the rates of progress are somewhat different from what he predicts, the dates for the singularity will likely shift by about two decades at most. This may make a mockery of the precise dates he gives in simplified versions, but the basic idea — that exponential growth could lead to a singularity — remains critical.

(4) Kurzweil’s version of the singularity isn’t the only one. Eliezer Yudkowsky has a somewhat different take on the subject. There are some excellent web write-ups on the subject, but I don’t want to bury links in the comments — I’ll probably take up the topic in a blog post of my own in the near future.