[holtjv]

Does anyone have any real experience with how this list is compiled and matched by the NSA?

Here's the context: my company develops very sophisticated matching software which matches on several different fields within the data record. This is nothing new as there are several companies doing this, albeit each in slightly different ways. But it is sort of new (last 4 years).

I've read, but don't know for sure, that the TSA/NSA/airlines matching system relies purely on Soundex software, which matches only on name. Soundex is a phonetic-type algorithm which applies a value to a letter, which can then be compared to the same value of another letter, or combination thereof.

For example, "Hakim" can sound like, "Gene"; and theoretically generate a false positive match. My selfish concern is that the TSA software is matching so many false positives (which isn't really a huge problem unless you keep getting nailed) that they're missing the legitimate matches.

We've done some testing on name data with Soundex and our software--multi-field teramatching produces about 40-50% more matches while lowering the false positives by 20-30%.

Obviously this is all about getting business for my company by getting attention in the media with these numbers--no obfuscation there--but nonetheless, it's just a bit frightening that the govt is using such poor software to ensure our national security. Not surprising though.

My suspicion is that, even if someone really knows how the matches are happening, they would be violating national security laws by talking about it. But there must be a bit more information available out there somewhere.

Anyone on the no fly watch list and have tried to get off of it? Any success?

Thanks for helping me support our host by increasing my income.

Jack