I am well aware of the information that is out there and was used to get the AFCI requirements into the code. I have been studying them since the first proposal for the 1996 code. I have also been studying what the AFCIs actually do. They stated that the device they proposed for the 1996 code would do what they now say the combination type AFCI will do. This whole process has been based on misleading information and the big play that if you oppose the AFCI you are opposed to safety. The misinformation that was put out by the manufacturer's of these devices has made me distrust most all information that is published by someone who has a profit motive in getting me to accept the information. If the code was a federal law a full cost benefit analysis would have been required and this rule would have never seen the light of day. Yes, there are those who say you can't put a value on human life, but manufacturer's do that everyday. Products can always be made safer, but at some point the cost factor overrules the safety factor.
As far as my numbers, you have to dig a lot deeper than that. The numbers I used started with 67,800 (over 50% more than you cite) fires of electrical origin in dwelling units in a year. The total number of dwelling units a bit more than 119 million units. This gave a fire rate of ~1.761 dwelling unit fires per 1000 units. I adjusted this to 0.264 per thousand based on the stats that show 85% of these electrical fires are in units over 20 years old. I took this rate and applied to the the ~1,650,000 new homes to be built in a year. (this was before the housing crash and the number would be much smaller) When I did this I found that there would be about 435 fires of electrical origin in the new dwelling units in the first year. Even the AFCI people don't say that these devices will prevent 100% of the electrical caused fires, but even if they did, and the cost of compliance was $400 per unit the cost per fire prevented would be over $1,500,000.
Note: The numbers are based only on the first year of compliance. When I run the numbers out for 20 years, I come up with about 128,000 fires prevented over the 20 years if the devices prevent 100% of the electrical fires in the new dwelling units constructed over these 20 years. If we assume the installed cost to comply with 210.12 stays at per dwelling unit stays at $400 for the full 20 years, we still have a cost of over $170,000 per fire prevented. As I said even the AFCI people do not claim 100% effectiveness and the number of fires prevented goes down and the cost of preventing each fire goes up as the effectiveness goes down. Some in the AFCI industry have stated that the effectiveness is 40% to 60%.