I have to disagree with Dave a little on the technology part because of the original claim of being able to detect a series arc, the problem with this as back when they were first introduced is how do you differentiate between a good arc (switch closing or opening) and a bad arc, they claim there is a difference but I can not see how other then a good arc will only last a few cycles, this is why I offered to UL to use a gated input that would shut off detection for a few cycles so a good arc will not trip the electronics, they said thank's but that was about as far as it went.
Of course equipment that an arc is a normal part of its operation such as a welder and electronic igniter's and such would have to be dealt with separately, the electronic igniter's are of such a low current that the AFCI could be designed to ignore them, but a welder would be a problem.
I have had some dealings with UL on the matter and a few here remember the e-mails I posted that went nowhere with them as they just brushed off my concerns even after they acted like there was a problem, but these problems were on long circuit runs when AFCI's required a 75 amp arc signatures to trip, well it must have had some effect because it wasn't long after that they lowered it to 35 amps or something around that.
But still since I was involved with our state building commission I worked hard to get them removed just because they could not prove to us they worked as advertised, and the short time from the 1999 requirement of having to use them at the start of 2002 till we adopted the 2002 (which we removed 210.12) they proved to be very costly to the problems they caused because of too many false positives and or the fact they had so many recalls along with there were no options for multi-wired circuits which caused even more cost to be burden by the home owners, for the very little to none in safety they added.
We must remember cost imposed upon the public is a big factor when adopting laws, like some said on here in the past "show me the body count" as to get a red light at an intersection their must be a reason and a design or product that is proved to reduce the deaths before you can add the cost of this design or product for the citizens of a state to bare the cost, it's called the "Need for a fix" the "options for the fix" and the "results of the fix" that a law makers must consider before being able for a change to be adopted into law, this breaks down to the body count or injuries, the redesign or product that will work to fix it, and the cost to the public if it out weighs the expense and need to the public, of course this is not always followed unless some watch dog group catches it.