The EFF’s representation is also misleading. For example, they talk about “the more than 17,000 numbers that the NSA was querying everyday, [of which] the agency only had “reasonable articulable suspicion” for approximately 1,800 of them.” This claim was surprising to me, because my understanding had been that there were only about ~17,000 targeted phone numbers in total (and only about 10% of them supported by RAS, which is problematic in itself, especially since it appears that only 4 of these have resulted in actionable leads). Now 17,000 might be a fairly high number, but it’s very different than the EFF’s 17,000 per day claim, which suggests an absurdly high amount of daily surveillance on a vast number of subjects. And it turns out that if you follow the links, the original EFF source indicates that there are only 17,000 numbers whose metadata is being queried, in total:
The court had told the NSA they were only allowed to query numbers that had "reasonable articulable suspicion (RAS)" of being involved in terrorism. Apparently, out of the more than 17,000 numbers on this list in 2009, the NSA only had RAS for 1,800 of them.
Let’s do some math on this. Apparently there were some 672 million active phone numbers in the US in 2009. The reason NSA surveillance is so troubling is that they are collecting metadata on everyone, so lets assume they are all in the database. Of those 672 million, the database can be queried on only 17,000 of them, which represents 0.0025%. Assuming that the NSA is now properly applying the RAS standard after the FISC decision, there should now be a 0.00025% chance of having your metadata examined by a living person; your chances of having your metadata read are just 1 in 400,000, not 1 in 4 million.
