4 ms·

Just to try and think through the situation: I don't use the phone particularly a lot, and I probably make 1 call per day on average. So I make 365 calls a yea

Just to try and think through the situation: I don't use the phone particularly a lot, and I probably make 1 call per day on average. So I make 365 calls a year. Let's say 350 to round. There are about 200,000,000 US adults. So if everyone makes as many calls as me, that's 70,000,000,000 calls a year. Each call should generate two book-keeping entries at the carriers -- so ~150 billion records a year. (Note: pure supposition this is what they mean by 'record', but phone carriers giving them a metadata firehose seems consistent with what we know.) So, we're talking 0.1% (and quite possibly less) of the phone calls? Ya know, I'm probably okay with that level of snooping on metadata from a spy agency (especially if they're not sharing the data with domestic agencies for domestic cases). Ed: I'm likely off by being at least an order of magnitude too low -- divide all my estimates of people targeted by 10 or so.

brewdad 9y ago
One would presume that SMS and MMS messages get caught up in this too. I send an order of magnitude more texts than phone calls. The % of users caught up in the dragnet must certainly be smaller than your estimate.
comex 9y ago
It's not a random sample of 0.1% of calls, though. Rather, they're collecting all the call metadata of a small fraction of subscribers; the number of people targeted may be relatively small (though not small in absolute terms), but each person targeted has their privacy significantly invaded, as the NSA learns quite a lot about their life. And to be targeted, you only have to have once talked to someone for whom the NSA has a "reasonable suspicion" they might be involved in terrorism. No need to do anything suspicious yourself.
- SomeStupidPoint 9y ago
  Yes, I'm aware. But we're talking... one in a thousand people, likely targeting mostly those involved heavily with international relationships. Even if you confine it to the Arab population, they're still not tapping everyone (closer to 1 in 6, if it were all Arab targets, which is unlikely). If you confine it to immigrants, it's 1 in 200. This suggests that they're using some actual basis for their selection (as you seem to be aware). It's unfortunate, but if they're collecting metadata for network analysis and if their selection criteria is that it's foreign related (and preferably tracing back to known networks), then it doesn't really strike me as a problem. Also, your last line isn't true -- talking to someone who is strongly suspected of being a terrorist is suspicious until analysis reveals the nature of the relationship, particularly if you go on to contact others. I suspect that people who only ping once (or even very intermittently) are quickly pruned as not being useful to observe.
MichaelGG 9y ago
A couple years ago, I handled dialer traffic for a modestly-sized dialer wholesaler. The total number of bookkeeping entries due to their traffic was 2-3 billion per week, representing perhaps 500M-1B call attempts. And I was basically a nobody. 151M call records is incredibly tiny, though they will be targeted and thus probably represent near 100% of the calls for some group of people.
- SomeStupidPoint 9y ago
  Someone in a different part of the thread puts it at 82% makes 5 per day. So my estimates are off by at least an order of magnitude (since outliers are likely to be targeted for a number of reasons). I'm just too lazy to fix them.

brewdad 9y ago

One would presume that SMS and MMS messages get caught up in this too. I send an order of magnitude more texts than phone calls. The % of users caught up in the dragnet must certainly be smaller than your estimate.

comex 9y ago

It's not a random sample of 0.1% of calls, though. Rather, they're collecting all the call metadata of a small fraction of subscribers; the number of people targeted may be relatively small (though not small in absolute terms), but each person targeted has their privacy significantly invaded, as the NSA learns quite a lot about their life. And to be targeted, you only have to have once talked to someone for whom the NSA has a "reasonable suspicion" they might be involved in terrorism. No need to do anything suspicious yourself.

SomeStupidPoint 9y ago
Yes, I'm aware. But we're talking... one in a thousand people, likely targeting mostly those involved heavily with international relationships. Even if you confine it to the Arab population, they're still not tapping everyone (closer to 1 in 6, if it were all Arab targets, which is unlikely). If you confine it to immigrants, it's 1 in 200. This suggests that they're using some actual basis for their selection (as you seem to be aware). It's unfortunate, but if they're collecting metadata for network analysis and if their selection criteria is that it's foreign related (and preferably tracing back to known networks), then it doesn't really strike me as a problem. Also, your last line isn't true -- talking to someone who is strongly suspected of being a terrorist is suspicious until analysis reveals the nature of the relationship, particularly if you go on to contact others. I suspect that people who only ping once (or even very intermittently) are quickly pruned as not being useful to observe.

MichaelGG 9y ago

A couple years ago, I handled dialer traffic for a modestly-sized dialer wholesaler. The total number of bookkeeping entries due to their traffic was 2-3 billion per week, representing perhaps 500M-1B call attempts. And I was basically a nobody. 151M call records is incredibly tiny, though they will be targeted and thus probably represent near 100% of the calls for some group of people.

SomeStupidPoint 9y ago
Someone in a different part of the thread puts it at 82% makes 5 per day. So my estimates are off by at least an order of magnitude (since outliers are likely to be targeted for a number of reasons). I'm just too lazy to fix them.