Foreign Intelligence Surveillance Court, Transcript of Proceedings Before the Honorable John D. Bates, Judge, United States Foreign Intelligence Surveillance Court . September 7, 2011. Declassified.

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 1

FOREIGN INTELLIGENCE SURVEILLANCE COURT WASHINGTON, D.C. SEPTEMBER 7, 2011

[illegible]

TRANSCRIPT OF PROCEEDINGS BEFORE THE HONORABLE JOHN D. BATES JUDGE, UNITED STATES FOREIGN INTELLIGENCE SURVEILLANCE COURT

APPEARANCES:

DOJ: [illegible]

TASHINA GAUHAR

LISA MONACO

ODNI: [illegible]

NSA: [illegible]

CHRIS INGLIS

[illegible]

[illegible]

[illegible]

[illegible]

[illegible]

TOP SECRET//COMINT//ORCON, NOFORN

NYT v DOJ, 16 CIV 7020_000080

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 2 1 PROCEEDINGS 2 THE DEPUTY CLERK: Will everyone please state your 3 names for the record. 4 [illegible]: [illegible], NSA. 5 [illegible]: [illegible] NSA Office of General 6 Counsel. 7 [illegible]: [illegible] NSA Office of General 8 Counsel. 9 [illegible] [illegible]: [illegible], NSA. 10 [illegible] [illegible]: [illegible], NSA. 11 [illegible] [illegible]: Good morning. [illegible] Acting 12 General Counsel, NSA. 13 MR. INGLIS: Chris Inglis, Deputy Director, NSA. 14 MS. MONACO: Lisa Monaco, Assistant Attorney General. 15 MS. GAUHAR: Tashina Gauhar, Deputy Assistant Attorney 16 General. 17 [illegible] [illegible]: [illegible], NSD. 18 [illegible] [illegible]: [illegible] NSD. 19 [illegible]: [illegible] ODNI, Office of General 20 Counsel. 21 THE COURT: Welcome to all of you. Please be seated, 22 and thank you all for coming today. We have a full house, 23 fuller than we usually get in this courtroom. 24 Now, my view of this proceeding is going to be mainly that 25 I'm going to ask a bunch of questions. I hadn't intended that TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000081

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 3 1 anyone give any argument to begin with or opening statement or 2 anything like that. We've been dealing with this issue, the 3 upstream collection and the certifications under 702 for several 4 months, so I don't think there's any need for that. 5 I'm sure that there will be both lawyers and nonlawyers 6 responding to some of the questions. Since this is a formal 7 hearing on the record, we need to swear, particularly those 8 nonlawyers who will be responding to questions. 9 It might be easiest if everybody who's going to be 10 responding simply stands up as [illegible] swears those who absolutely 11 need to be sworn. I think that's probably the easiest 12 procedure. So everyone who's going to be speaking and 13 potentially responding to factual issues, please stand be sworn. 14 (Attendees are sworn.) 15 THE COURT: All right. So first of all, I want to 16 thank everyone not only for being here but for all the very 17 helpful materials that have been supplied over the past few 18 months. We've looked at them closely, taken them all into 19 account, obviously had questions with respect to certain things, 20 and you've followed up with responses, and all of that is very 21 much appreciated and has advanced this matter considerably. 22 We wind up at this point with some continuing questions 23 that I thought it would be useful to have you come in and talk 24 with us about, and they are in many areas. By "many areas," I 25 mean they include minimization concerns; they include some TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000082

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 4 1 questions about the submissions, looking not all the way back 2 but back one or two submissions; they include definitional 3 questions relating to intent; and they include Fourth Amendment 4 issues. There just are a number of areas. 5 As I said already, I understand that different people may 6 be responding to those different areas, but let me just jump 7 right in and start with a couple of questions mainly about the 8 August 16 submission, although the questions obviously relate to 9 some of the earlier and the most recent August 30 submission as 10 well. 11 In the August 16 submission -- I think it's in footnote 5 12 on page 2 -- the government has discussed and indicated that 13 some of the Internet communications that have been acquired and 14 are continuing to be acquired under section 702 were purged 15 prior to the July 14, 2011, time point in dealing with this 16 statistical assessment and therefore were not included in the 17 NSA sample. I have a few questions about that body of 18 communications that were purged. 19 Does the government know how many communications were 20 purged and why they were purged? 21 [illegible]: We do not have an exact account at this 22 point of how many were purged. 23 THE COURT: Do you have any sense of why they were 24 purged? 25 [illegible]: There could be a number of reasons: TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000083

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 5 1 roamers, overcollect of sorts, and things we would have filed of 2 incidents. 3 THE COURT: Do we know if any were purged because they 4 were determined to be wholly domestic? 5 [illegible]: I don't believe any of those were purged 6 because they were wholly domestic. 7 THE COURT: You don't believe, or you're sure that 8 they weren't? 9 [illegible]: In our previous filings, we have indicated 10 that prior to our statistical analysis we had not identified any 11 communications of the MCT type that were wholly domestic, which 12 would require purge. 13 THE COURT: Now, since we don't know much about this 14 purged group -- and this again is just focusing on the 15 statistical sample -- how do we know that the sample is 16 representative of the actual collection of Internet 17 communications? If we don't know what's been purged and what 18 the nature of those purged communications is, doesn't that 19 affect the validity of the sample to the extent that it's a 20 sample of the collection of Internet communications? 21 [illegible]: The sample that we evaluated were the 22 sample that were in there and available to us. 23 THE COURT: I understand. 24 [illegible]: In our evaluation of this sample set, there 25 were some communications that were purged during our evaluation, TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000084

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 6 1 but we believe that given the six-month period, the number of 2 items that were included in that six-month period and then the 3 number that we manually evaluated has a statistical 4 representation of that whole body with a 95 percent confidence. 5 THE COURT: But it seems -- to a fairly ignorant 6 mathematician, it seems that if you're trying to get a 7 representative sample and that representative sample is to be 8 representative of the collected Internet communications but you 9 first take a chunk out of the collected Internet communications, 10 that could affect the validity of that sample. 11 I don't know how it would affect it or whether it would 12 affect it significantly, but I'm just trying to probe whether it 13 does have some impact mathematically on the validity of that 14 sample as being representative of the collected Internet 15 communications. 16 [illegible] Your Honor, if I might try to answer that 17 from the General Counsel's Office. We know that the sample that 18 we took was one specific date, and so it's representative a 19 snapshot of time what was in our system. 20 The reason, Your Honor, that we can't articulate the number 21 of items that had been purged is because we can't put our finger 22 on those in the same way that we can put our finger on what's in 23 our systems at any one point in time. As you may be aware from 24 prior filings, when we discover a compliance incident, it may be 25 several months in time beyond the time that the actual item was TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000085

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 7

collected.

So that 13.25 million sample that you saw is reflective of what was in our systems as of that particular date. As you march forward in time from the time we took that sample, you would expect that other items that were collected during that time could also be similarly identified through our compliance process and purged.

THE COURT: Well, let me -- this is not intended to be reflective of the truth and is simply for demonstrative purposes, but if you purged 13 million from during that time period, then all the numbers that you're presenting would really only be half of the picture in terms of what's collected.

See, I'm interested ultimately in what's being collected, not what happens to be sitting in your data files at a particular point in time. If the purging that takes place as a normal course of business is half of the material, then it changes everything just from that perspective even without knowing what the purged material is, whether it's of a different nature and richer in terms of wholly domestic communications or not. Just in terms of raw numbers, it would really alter things.

So to the extent that you can tell me something about how much has been purged -- 10 percent, .0001 percent, 50 percent -- it certainly is helpful for me in assessing the impact of the statistical presentation you make.

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000086

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 8 1 [illegible]: Right. Your Honor, I think we understand 2 that and appreciate the point. My understanding, though, of the 3 Court's underlying concern was how many of the communications 4 were wholly domestic. 5 THE COURT: Ultimately, yes. 6 [illegible]: Right. It has been an unusual occurrence 7 for our folks to find wholly domestic communications in the 702 8 collections. So, although I can't say it with a certainty, it 9 seems to me that of the communications that would have been 10 purged in the normal course of our compliance regime because 11 either it was a roamer communication or we misidentified a task 12 selector, that those were not likely to have been wholly 13 domestic communications that would have affected the validity of 14 the sample in the sense that we were going through the sample 15 size to actually try to find wholly domestic communications. 16 THE COURT: I understand that based on your 17 presumptions and the presentation you've made, but even 18 accepting the presentation you've made and assuming that the 19 nature of the purged materials we're talking about right now is 20 the same as the sample you looked at, if the volume of the 21 purged materials is equal to what's left, then the numbers that 22 you've given me have to be doubled. 23 [illegible] Your Honor, I think it's fair to say that 24 we don't believe that it would be that high, but we haven't come 25 prepared with numbers in terms of -- TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000087

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 9

1 THE COURT: And you certainly could assess that for 2 the future at some point, because you can look at what's 3 collected and look at what's left during a six-month period, but 4 for now you can't tell me anything more with respect to the 5 six-month period that was analyzed. 6 [illegible] Those are certainly numbers that we can 7 try to get to Your Honor, but we don't have those today. 8 THE COURT: Okay. Let's move on. 9 MR. INGLIS: Your Honor, could we, though, take the 10 action and respond back to you within this week with what we can 11 reconstruct in terms of that purged list? 12 THE COURT: Yes. I think what we should do is at the 13 end of this discussion decide what you would like to present 14 further and talk about a timeline for it. Let's not do it on 15 one item, because there may be four items by the time we're 16 through. 17 Staying on the same vein with respect to the August 16 18 submission, the government states later on in that submission -- 19 I think it's on page 7 -- that NSA cannot determine whether 224 20 of the roughly 5,000 MCTs examined contained wholly domestic 21 communications. Then a little bit later, it's noted that 23 of 22 the 224 MCTs were not further analyzed because they were 23 subsequently purged or placed on the NSA's master purged list. 24 With respect to those 23, do we know why those were purged, 25 or are they just part of this broader category that were purged?

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000088

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 10

1 [illegible]: Yes, Your Honor. In fact, I believe some 2 of the ones that were purged during the sample time were 3 responsive to the [illegible] incident that was previously 4 reported to the Court in the prior filing. If you recall, I 5 believe NSA purged somewhere upwards of [illegible] total 6 transactions as a result of that [illegible] incident. I can't 7 remember the exact number, but a specific number of that 23 were 8 related to that purge. 9 THE COURT: Were any of the 23 purged because they 10 were wholly domestic communications? 11 [illegible]: No, Your Honor. 12 THE COURT: All right. Then also in the August 16 13 submission, we have this sort of unknown category which is the 14 -- 45,359 I think is the accurate -- no, I'm sorry. Not the 15 unknown category, but there are 45,359 of the overall sample of 16 50,440 transactions reviewed that were determined to be single 17 discrete communications. So those were sort of set aside with 18 no further analysis. 19 In an earlier submission, I think the June 1 submission on 20 page 6, you noted that communications are nearly always 21 transmitted from a sender to a recipient through multiple legs 22 before reaching their final destinations, and certainly that 23 seems obvious to all of us who have dealt with these things. 24 Because NSA's IP filters [illegible] 25 [illegible] the government has

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000089

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 11

1 indicated that NSA could intercept the communication [illegible] 2 [illegible] 3 [illegible] 4 [illegible] 5 And then based on a further explanation on pages 7 and 8, I 6 think, of the June 1 submission, should the Court understand 7 that NSA's upstream collection filters wouldn't prevent [illegible] 8 [illegible] 9 10 11 12 Do you understand the question? 13 [illegible]: The filters would not prevent that, and 14 those are examples of the 10 wholly domestic communications that 15 we reference in the August 16 filing. That was exactly the 16 case. 17 THE COURT: So those communications would all -- those 18 about communications would all be subsumed in the category that 19 is identified in the August 16 submission on page 9 as between 20 996 and 4,965, or would they be in one of these other 21 categories? I don't think they're in that group. Aren't these 22 part of the single discrete communications that you didn't even 23 analyze further? 24 [illegible]: That's correct, Your Honor. 25 THE COURT: Aren't there going to be some wholly

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000090

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 12 1 domestic communications in that group that you didn't analyze 2 further? 3 [illegible]: It's possible, Your Honor, but we -- 4 THE COURT: It's more than possible, isn't it? Isn't 5 it likely there will be some about communications [illegible] 6 [illegible], and it's actually a 7 communication that is between two U.S.-located persons? 8 [illegible]: Your Honor, that is possible. However, 9 typically, 10 [illegible] 11 12 13 14 15 16 17 18 19 20 21 THE COURT: How about [illegible] 22 [illegible] 23 24 [illegible]: Yes. It is possible, but the reason 25 why -- TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000091

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 13

1 THE COURT: It's more than possible. Aren't there 2 going to be a fair number of things? 3 [illegible]: Correct. There definitely will be. 4 However, the way they're designed is, as I said, [illegible] 5 [illegible] 6 7 8 THE COURT: The problem that I'm left with is to the 9 extent that there's a statistical presentation to convince the 10 Court of a low volume of wholly domestic communications, this is 11 another category that exists that -- you're using the word 12 "possible," but I have no way to quantify it or know how many in 13 this sample -- just talking about the sample -- knowing how many 14 might be of that category. 15 [illegible]: Your Honor, I think with respect to the 16 sample that NSA conducted, we certainly endeavored to try to be 17 as responsible as we could in the -- 18 THE COURT: Let me state, you've been very 19 responsible. You've been very helpful. I know you're all doing 20 the best you can. I'm just probing the information. 21 [illegible]: Certainly. The focus of the sample -- I 22 think you're correct -- was geared towards identifying MCTs 23 within NSA's upstream collection holdings in order to be able to 24 characterize the nature and scope of that collection 25 specifically with respect to multi-communications transactions.

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000092

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 14 1 So there may be a number of other things that weren't more 2 specifically looked at within that particular sample which we 3 could attempt to take back and answer for the Court at a later 4 time, but the sample that was run was for the purpose of 5 describing MCTs and the nature and scope of that feature of that 6 collection. 7 THE COURT: All right. I think it's fair for me to 8 say to you that I am concerned about this category within the 9 sample of 45,000, which is most of the sample. It's by far the 10 majority of the sample and the fact that it does seem to us that 11 there are going to be some about communications that are 12 probably wholly domestic communications in that grouping, and I 13 don't have any way to quantify or assess how much it is. 14 I know you have cautiously, and appropriately, used the 15 term "possible." It seems to me it's more than possible; it is 16 highly likely, but I can't say what it means because I don't 17 know what kind of volume we might be talking about. So let's 18 put that down on the list of something that we may benefit from 19 further information on. 20 I'm going to allow the legal advisors to follow up with 21 questions on these subjects before I move to another one. 22 [illegible]: [illegible], FISC legal counsel. Just a 23 follow-up. If someone's [illegible] 24 [illegible] 25 [illegible] TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000093

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 15

1 [illegible] Do you have a sense 2 one way or another? 3 [illegible]: It's not necessarily the case that [illegible] 4 [illegible] 5 [illegible] 6 [illegible] 7 [illegible] 8 [illegible]: But other than that sort of general 9 observation, you don't have a sense percentage-wise of how many 10 [illegible] 11 [illegible]: No, we don't. 12 THE COURT: All right. Staying with the August 16 13 submission for just a moment longer, the government has 14 concluded -- I think it's on page 9 of that submission -- that 15 of the 13.25 million Internet transactions acquired via the 16 upstream collection during a six-month period, between 48,609, 17 and 70,168 are MCTs containing one or more communications 18 between nontargeted persons but lacking sufficient information 19 for NSA to identify the location of the sender and all intended 20 recipients of that communication. 21 It's a category that I call "the unknown category," which 22 is fairly large, certainly larger than the category that you've 23 identified as actually containing a wholly domestic 24 communication. A little later in the submission, you indicate 25 that NSA has no basis to believe that any of this category of

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000094

Names of certain hearing participants withheld under b(6) and b(7)(C).
All withheld information exempt under b(1) and b(3) unless otherwise noted.
Approved for public release.
~~TOP SECRET//COMINT//ORCON, NOFORN~~
16

1   transactions contain wholly domestic communications.
2   Now, that's a fairly absolute statement that's made in the
3   submission. Doesn't the conclusion that between 996 and 4,965,
4   which are pretty low percentages but nonetheless an actual
5   determined amount on this sample, doesn't the conclusion that
6   there are those wholly domestic communications acquired every
7   six months, doesn't that undermine the presumption that none of
8   these unknown transactions contain wholly domestic
9   communications? Wouldn't one expect that at least that
10  percentage of that unknown category would be wholly domestic?
11  MR. INGLIS: Your Honor, if I might, I'll defer to [black box]
12  to provide the detailed answer, but in that case, when presented
13  with the possibility that these are either unknown or
14  unknowable, we pushed our analysts further to do two checks
15  against each and every one of these items, first to check to
16  determine whether or not there was any information that might be
17  attributable to a domestic communication, and second, to
18  determine whether there was any information that might lead us
19  to conclude that in fact it was a foreign communication.
20  In each case, both of those checks for each of these items
21  came back showing that the preponderance of evidence -- not
22  absolutely, but the preponderance of evidence which we had
23  before us would say that there were no domestic communications
24  in that pile.
25  THE COURT: So I guess what you're saying is that this

~~TOP SECRET//COMINT//ORCON, NOFORN~~
NYT v DOJ, 16 CIV 7020_000095

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 17 1 grouping, you've determined, is not going to be -- it's going to 2 be less rich in wholly domestic communications than the overall 3 sample is. 4 [redacted] That's correct. 5 THE COURT: I mean less rich even to the point of 6 containing no wholly domestic communication. Why is it that 7 this grouping -- what is it about this grouping that should 8 convince me that it is different than the rest of the sample and 9 will contain no wholly domestic communications? 10 MR. INGLIS: That's more than a fair question. It 11 wasn't that it was wholly devoid of contextual information, that 12 it lacked information, a conclusive statement, but the remaining 13 artifacts led us in every case into a -- if you had to decide 14 yes or no based upon the available information that it was 15 foreign as opposed to domestic. So in each case it wasn't that 16 there was no information. There was insufficient information to 17 say with absolute certainty. 18 [redacted] Correct. And the way we did that analysis 19 stemmed from the data set we evaluated. Of that 224, we did 20 this in-depth analysis that Chris described, and via a 21 statistical analysis we were able to extrapolate that sample set 22 with a 95 percent certainty across the entire 13 million. And 23 there's a certain error associated with that, and that error is 24 expressed in the confidence interval. 25 So based on the data set that we evaluated on page 7 and TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000096

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 18

1 before, our statistician was able to draw conclusions based on 2 that to the true proportion of those type of communication 3 across the entire 13 million set within that certain confidence 4 interval. 5 MR. INGLIS: But, Your Honor, if I understand your 6 question, you would ask why do we believe that that population 7 of data that we would declare as unknowable is statistically 8 different than the larger set from which it was extracted, 9 perhaps on a statistically relevant basis. 10 THE COURT: My guess would be that it goes in the 11 opposite direction because it's a group that has already 12 eliminated all these large portions that clearly don't contain 13 any wholly domestic communications from your view. So why does 14 it then -- 15 MR. INGLIS: In the case of the 10 wholly domestic 16 communications that clearly stood out as having artifacts that 17 said they're wholly domestic. In the case of this pile, there 18 were no artifacts associated with those that spoke to the 19 possibility of domestic that we couldn't rule it absolutely out. 20 We also looked to see whether in each case there were 21 artifacts that would lead us to conclude that if we had to make 22 a judgment that they were foreign, not domestic, and in which 23 case both of those tests led us to conclude that they were, not 24 presumptively, but more likely, to be foreign than domestic. 25 There was no information I'm aware of in the file -- and we

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000097

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release.

TOP SECRET//COMINT//ORCON, NOFORN 19

1 pushed our analysts very hard on this particular pile because we 2 had the same question you did, which is could this be a soft 3 underbelly in our analysis, and pushed them very hard, and they 4 came back with no information that would lead them to conclude a 5 strong possibility that they were domestic. 6 [illegible]: That's correct. [illegible] 7 [illegible] 8 9 10 11 12 THE COURT: Where -- and excuse me for asking the 13 question this way, but I just want all the help I can get. 14 Where should I look in this submission for the explanation that 15 you've given me about the analysis of this unknown group that 16 led you to conclude that it will not contain wholly domestic 17 communications? 18 [illegible]: It's on page 7 at the bottom. 19 THE COURT: Page 7 where, sir? 20 [illegible]: On the second bolder bullet. 21 THE COURT: Of the 5,081 MCTs? 22 [illegible]: Yes. 23 THE COURT: All right. We can read that further and 24 analyze it. 25 MR. INGLIS: Your Honor, we're happy to be more

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000098

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 20

1 responsive based upon what we have and do further analysis as 2 the Court may please. 3 THE COURT: Well, you can look at it too, and you can 4 mark this down as No. 3. If there's further analysis that 5 you've done that supports the conclusion that you reach that 6 you're now expressing that this unknown category is in fact 7 unlikely, and you think highly unlikely to contain wholly 8 domestic communications, then by all means be prepared to 9 provide it, because on first reading, that's not the conclusion 10 that we reached from assessing this submission. 11 All right. 12 [illegible] Your Honor, just to add into this, I know 13 we haven't gotten into the 30 August -- 14 THE COURT: And I'm not going to be going through it 15 line by line. Don't worry. 16 [illegible] But I do think there were at least some 17 portions of the 30 August filing that try to speak to that exact 18 question, and when we get to them, I'll try to -- 19 THE COURT: Okay. Thank you. 20 All right. Just give me one second to see -- all right. 21 The last question in this area of examining submissions goes 22 back early to the June 1 submission which is clear that the 23 scope of the data that NSA actually acquires through the 24 upstream collection is constantly evolving. And you explained 25 that on pages 24 and 25, noting that [illegible]

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000099

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 21 1 [illegible] 2 [illegible] 3 [illegible], and all of that can affect the amount and type 4 of data included in a particular transaction. 5 What I'd like to get a fix on is what this means in terms 6 of any particular sampling at any particular time. In view of 7 that evolution, is it likely that if NSA did a similar analysis, 8 a sampling type analysis of the upstream collection at some 9 later date, two years later than the one done in the recent time 10 period ending in July, is it likely that that analysis would 11 wind up with something significantly or at least somewhat 12 different? This evolution, is it material evolution that really 13 would change things significantly? Or don't you know? 14 [illegible]: Your Honor, a lot of the analysis 15 focused on identifying MCTs and their percentages showing up in 16 our collection. It is possible that let's say two years down 17 the road that the Internet is more rich with MCTs for a number 18 of different reasons. 19 We could have [illegible] 20 [illegible] 21 [illegible] and that would 22 statistically increase the amount of MCTs. 23 Because of that unknowable factor of technology popularity 24 in the future, that could change things. Certainly, several 25 years ago, we've talked in the past about [illegible] TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000100

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 22 1 [illegible] 2 [illegible] and as 3 technologies have evolved, that has become more and more 4 popular. So we could see somewhat different results if we were 5 to conduct a similar study in the future. That is definitely 6 something that could happen. 7 THE COURT: And there's no reason -- 8 [illegible] The other thing I wanted to add, 9 Your Honor, is one of the other factors here that can somewhat 10 limit the evolution is that one of the factors in play here is 11 [illegible] 12 [illegible] 13 [illegible] 14 [illegible] 15 [illegible] 16 [illegible] 17 [illegible] 18 [illegible] 19 THE COURT: I think that last explanation of a 20 limiting factor is important because it affects the question I 21 was about to ask, which is, we use the term evolution, and I 22 take it we would look at this with respect to MCTs and say there 23 has been an evolution to the extent that there are more MCTs now 24 than was the case 10 years ago. Correct? 25 [illegible] I would say that in some cases -- so, TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000101

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 23 1 for instance, 2 3 4 So it's not just so much the MCTs; it's the nature 5 of the MCTs that has evolved as well. 6 THE COURT: But, in light of the limiting factor that 7 you just mentioned, can we make an assumption that the upstream 8 collection -- let's just focus on that -- will become richer and 9 richer in MCTs, or can we not make that assumption? 10 I don't think we could or couldn't with 11 any degree of competence. 12 THE COURT: All right. Let's move on to presumptions 13 more generally here. There are several presumptions that the 14 government has urged the Court to continue to rely upon with 15 regard to the upstream collection. They include that 16 17 18 and that's included in the June 19 1 submission on page 11. 20 There's also a presumption that the vast majority of 21 persons outside the U.S. are non-U.S. persons and most of their 22 communications are with other non-U.S. persons located overseas. 23 That's referenced in the June 28 submission on page 5. And 24 there's also a presumption that 25 We've talked TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000102

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 24

1 about that a little bit, and that's certainly referenced as well 2 in the June 1 submission. 3 In view of the analysis of the upstream collection that 4 you've done -- and I know we're talking about numbers and 5 percentages here, but in view of that analysis, and in 6 particular the conclusion that the upstream collection does 7 contain a certain number of or percentage of wholly domestic 8 communications over a six-month period as analyzed, can the 9 Court rely on those presumptions as really being absolute, or 10 are they imperfect presumptions at best? 11 [illegible] Your Honor, if I may, I think they are, 12 like all presumptions, imperfect. I think the numbers we've 13 generated from the study approximate what we expected; that is, 14 a low number of domestic communications when something we think 15 is aberrational is taking place. So we do not intend to say it 16 can never be the case that, but the presumption is it's not the 17 normal behavior, and I think that's what the study bore out. 18 MR. INGLIS: And, Your Honor, I would add as the chief 19 operating officer of the National Security Agency and having to 20 then essentially sign up to our end of a representation, that I 21 feel the same, that the preponderance of those assumptions has 22 been borne out by the data to be correct but imperfect and that 23 what we then have to apply are a set of procedures to ensure 24 that we are looking for those exceptions and that we act 25 appropriately when we discover those exceptions and that we can

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000103

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 25 1 therefore offer back that the totality of that has a high 2 probability, again not absolute, but a high probability of 3 making the right call in terms of what our presumption should be 4 and having a high probability of catching the exceptions that 5 then ensue and dealing with those appropriately. 6 THE COURT: Let's just turn back the page a moment to 7 something we were talking about a minute ago with the evolution 8 of the Internet. Any sense of what, for example, [illegible] 9 [illegible] might do to these issues and the way that 10 Internet communications are routed? 11 [illegible] 12 [illegible] 13 [illegible] 14 [illegible] 15 [illegible] 16 [illegible] 17 So at this point, we haven't seen any change, and I'm not 18 too sure what we would expect to see just given the nature of 19 [illegible] 20 [illegible]. 21 THE COURT: We're talking about imperfect 22 presumptions. To the extent that NSA is acquiring [illegible] 23 [illegible] 24 [illegible] should the presumptions be reversed? In other 25 words, should we instead presume that communications are within TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000104

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 26

1 the United States and with U.S. persons? 2 MR. INGLIS: Your Honor, I think that question centers 3 on [illegible] 4 [illegible], so I would ask [illegible] to talk to that. 5 [illegible] Just to clarify, the question is, is 6 should we presume at some future date that a U.S. service 7 provider is primarily servicing U.S. persons? Or... 8 THE COURT: It's a little less than that. It's just 9 if the acquisition is [illegible] 10 of someone in the United States, do the presumptions that you 11 rely on and urge the Court to apply hold up, or should we 12 actually think that the opposite would be true, that for those 13 acquisitions, the presumptions, if you're going to apply a 14 presumption, the presumption should be that the communications 15 are within the United States and with U.S. persons? 16 [illegible]: So if the person was in the 17 United States and [illegible] and we intercepted that, 18 would we presume at some future date that that was -- I'm not 19 sure I fully understand. 20 [illegible] If I can address it, Your Honor. I 21 think that that would be an excellent presumption to reverse if 22 not for the presence of the targeted selectors. So remember 23 that the reason that we collect any particular transmission is 24 that it contained the target selector which would be an 25 independent determination that the target is reasonably to be

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000105

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release.

TOP SECRET//COMINT//ORCON, NOFORN 27

1 outside the U.S. So that weakens otherwise what I think would 2 be the natural inclination to reverse that presumption. 3 [illegible] [illegible] If you've got a transaction that is [illegible] 4 [illegible] 5 [illegible] 6 [illegible] 7 [illegible] 8 [illegible] With respect to those individual discrete 9 communications, what presumption, if any, should be applied? 10 [illegible] In that scenario that you've described, 11 if [illegible] 12 [illegible], we still are applying the IP filtering 13 process. So if that user was [illegible] 14 [illegible] 15 [illegible] So that 16 particular scenario wouldn't occur. [illegible] 17 [illegible] 18 [illegible] So that is what happened in the 10 19 examples we have. [illegible] 20 [illegible] We 21 think we're going to very, very rarely see that happen because 22 the overall presumption of the statute, remember, is that the 23 whole world out there is using the services here. 24 [illegible] 25 [illegible]

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000106

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 28

1 2 3 That's the scenario -- and we've seen it play out in this study 4 on 10 occasions -- that's a scenario that you're raising. But 5 it won't be somebody in the United States communicating with 6 [illegible] in the United States. 7 [illegible]: Just to make sure the point is crystal 8 clear for the Court, the concern that the Court has about what 9 presumption to apply to communications [illegible] 10 11 12 13 14 15 16 17 18 [illegible] which was the exception in the study, not the rule. 19 Is that a fair statement? 20 [illegible] That's correct. 21 MR. [illegible]: And I think it's also important to 22 remember that the presumptions aren't the first resort. The 23 presumptions are in many senses a last resort. There can be 24 objective indicia of the location of the communicants that is 25 more reliable than the presumption. The presumptions apply

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000107

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 29

1 ostensibly in the absence of information to the contrary, and 2 implicit in that requirement is a due diligence requirement to 3 actually try to assess whether there's any objective information 4 that to the extent you need to rely on the presumption that 5 would rebut the presumption. 6 I think NSA's manual review bears out that the digging can 7 result in the location of reliable, objective information that 8 is indicative of a person's location. 9 MS. MONACO: Just to press on that one bit, I think 10 that's why the Court is struggling with the 224. NSA only 11 arrived at that 224 because the NSA works through all of the 12 objective indicia that [illegible] mentions, and then, only then, 13 after discarding all of that and identifying all of that and 14 making conclusions from it results in the 224 where there were 15 no reliable indicia. 16 THE COURT: To round this out -- and [illegible] if you 17 have more questions, please feel free to ask them -- but let's 18 talk about the [illegible] situation for a moment that I 19 already referred to. [illegible] 20 [illegible] 21 [illegible]: [illegible] 22 [illegible] 23 [illegible] 24 [illegible] 25 THE COURT: [illegible]

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000108

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 30 1 2 3 4 5 : 6 7 8 9 10 11 12 THE COURT: 13 14 15 16 17 18 19 20 21 22 23 24 25 TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000109

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 31 1 2 3 THE COURT: 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 THE COURT: 21 : 22 23 24 25 TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000110

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 32

1 2 3 4 5 6 7 8 9 THE COURT: 10 11 12 13 : 14 15 16 17 18 19 20 THE COURT: 21 22 23 : 24 25

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000111

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 33

1 [illegible] 2 [illegible] 3 THE COURT: All right. Okay. Let's move on to the 4 question of intent. This is sort of a difficult question to 5 answer. It's an easy question to ask, but I'll throw it out 6 there, and I can be more specific if you'd like me to be. 7 The government's submissions do use various terms -- 8 intentional, unintentional, inadvertent, incidental -- when 9 describing the upstream collection or aspects of the upstream 10 collection. I'm a little confused as to what is meant, and 11 rather than just throwing that out to you, let me ask one or two 12 specific questions. 13 Is unintentional the same as inadvertent? As you use those 14 terms, are they the same? Do both mean "not intended?" 15 [illegible] Your Honor, I think that's correct. 16 "Inadvertent" is a term that's used in the NSA immunization 17 procedures, and "unintentional" has been used in various 18 documents as well. But I think "inadvertent" and 19 "unintentional" can mean the same thing. 20 THE COURT: And "intentional" is the opposite of both 21 "unintentional" and "inadvertent." 22 [illegible]: Yes. 23 THE COURT: And "incidental" means what? 24 [illegible] "Incidental," I think as we've framed it, 25 means something that results as a consequence of an action that

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000112

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 34 1 we take intentionally. 2 THE COURT: Is happening or a byproduct -- 3 [illegible] It's not our intent in taking that action, 4 but as you said, it's a byproduct of that action. So in an MCT, 5 for example, our intent is to acquire a communication to, from, 6 or about a targeted selector. At the time we acquire that 7 transaction, we may not even know that it's an MCT that contains 8 other communications. 9 So we acquire the MCT because we see, hey, that's a 10 communication that has a targeted selector. We reasonably 11 believe that that has foreign intelligence information in it. 12 We acquire it, it's turns out it's MCT, it's got other 13 communications in it. We still intended to acquire that 14 transaction and anything contained within it such that to the 15 extent there are these other communications in it, it's 16 incidental to our acquisition of the transaction. 17 THE COURT: All right. Let's drill down a little 18 more. The June 28 submission states that "acquisition of 19 Internet transactions is intentional," and on the same page, 20 page 6, further states that "given the government's knowledge 21 that such transactions may also include information that is not 22 to, from, or about a task selector, the acquisition of this 23 additional information is not inadvertent." 24 [illegible] I think that's correct. 25 THE COURT: Later in the same filing, and also in the TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000113

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 35 1 August 30 filing, the government seems to be saying that any 2 wholly domestic communication that is acquired as part of a 3 transaction is obtained unintentionally or inadvertently. Is 4 that also correct? 5 [illegible]: Yes. 6 THE COURT: Now, NSA knows with certainty that both 7 will be acquired. In other words, wholly domestic discrete 8 communications are simply a subset of the nontarget discrete 9 communications that you acquire as part of a transaction. So 10 how do you reconcile that with these statements with respect to 11 intentional or unintentional? 12 [illegible]: So we intentionally acquire MCTs because 13 they contain the presence of a targeted selector that we believe 14 is used by a non-U.S. person located outside the United States. 15 However, we don't intentionally acquire all such MCTs. We know 16 that we cannot acquire -- 17 THE COURT: You knowingly acquire them, though. 18 [illegible]: But we don't intentionally acquire them. 19 THE COURT: Well, we'll get to that in a moment. 20 [illegible]: Understood. And we may not know at the 21 time of acquisition that that MCT is something that contains -- 22 THE COURT: With respect to a particular transaction. 23 [illegible]: Exactly. So we are intentionally 24 acquiring MCTs, but we have also implemented [illegible] means to 25 ensure that we are not intentionally acquiring wholly domestic

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000114

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 36

1 communications, but we believe that those [illegible] means are 2 reasonably designed to prevent that. Granted, they are not 3 perfect, but we do believe that they are reasonably designed to 4 prevent the acquisition of those communications. 5 THE COURT: Let's jump to the statute for a second. 6 Under section 1881(a)(1), the Court is required to determine 7 whether the targeting procedures are, in the language of the 8 statute, "reasonably designed to prevent the intentional 9 acquisition of any communication as to which the sender and all 10 intended recipients are known at the time of the acquisition to 11 be located in the United States." 12 Let's look at those two parts of that statutory provision. 13 Intentional acquisition. 14 In the government's view, what's the meaning of 15 "intentional" as used in the statute? And to set the framework 16 for you, I think in criminal law and in tort law it's generally 17 settled that a person intends to produce a consequence either 18 when he acts with a purpose of producing the consequence, which 19 is what you've been focusing on I think in your papers and here 20 for the moment, or, when he acts knowing that the consequence is 21 substantially certain to occur. 22 It does seem to me that you really focus on the first part 23 of that traditional definition, because you're really only 24 talking about the specific-purpose definition and not the 25 knowledge of substantial certainty aspect of intent. Am I

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000115

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 37

1 right? 2 [illegible] [illegible]: I think we need to focus on the fact that 3 -- I mean, granted, the statute says that we are precluded from 4 acquiring communications as to which the sender and all intended 5 recipients are known at the time of acquisition to be located in 6 the United States. I grant you that we may not have focused 7 specifically on that aspect, our knowledge at the time of 8 acquisition. I think NSA's manual review has shown that it can 9 take a lot of drilling down into these communications to 10 determine whether or not they are in fact wholly domestic. 11 THE COURT: Well, it doesn't take that much drilling 12 anymore to determine that there are some that are wholly 13 domestic. It does take a lot of drilling, if you even can, to 14 determine that a particular transaction is wholly domestic or a 15 communication within a transaction is wholly domestic. 16 Indeed, it's especially true or especially difficult when 17 acquired, but it doesn't take much now to conclude, because 18 that's what your analysis has concluded, that there will be 19 wholly domestic communications acquired. Certainly will be. 20 [illegible] [illegible]: No, no. I agree a hundred percent with 21 that, Your Honor. But again, at the time of acquisition, we may 22 not know it, and it's at the time of acquisition that the 23 statute precludes us from intentionally acquiring a domestic 24 communication. 25 THE COURT: Then we're getting a little bit semantic.

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000116

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 38 1 It depends on what you mean by "acquisition," whether you mean 2 acquisition of a particular communication, acquisition of a 3 particular transaction, or the acquisition that takes place 4 during the space of a day, or an hour. If it's either of the 5 latter two, you know that there's some wholly domestic 6 communications that are being acquired, statistically. 7 [illegible] Statistically, yes. Personally, the way 8 that I've always viewed the way that that provision in the 9 statute works is it's on an acquisition-by-acquisition basis, 10 and "acquisition" meaning communication-by-communication basis, 11 because we are targeting to acquire foreign intelligence 12 information. That foreign intelligence information is contained 13 in individual communications. 14 THE COURT: Do you think that definitionally, the 15 Court, when it's interpreting and applying 1881(a) in this 16 setting, should only be focused on the purpose portion of the 17 definition that I went through a moment ago, or should the Court 18 also be looking at the knowing aspect of it -- in other words, 19 that part of the definition of "intention" at to be applied here 20 is "not only acting with a purpose of producing the consequence 21 but also acting knowing that the consequence is substantially 22 certain to occur"? 23 Should I be jettisoning that portion of the traditional 24 "intentional" definition and applying some narrower definition 25 here? Or are you only saying that even applying both prongs of TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000117

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 39

1 that definition you need to look at what's happening and what's 2 known at the time of the acquisition of a specific transaction? 3 [illegible]: I think it's the latter, and I also think 4 that the fact that the statute says that the procedures have to 5 be reasonably designed to prevent the acquisition of 6 communications as to which the sender and all intended 7 recipients are known at the time of acquisition to be located in 8 the United States. 9 I think given that language, the statute contemplates that 10 there's not going to be a perfect system. It has to be a 11 reasonably designed system, and I think the results of NSA's 12 manual review bear out the fact that the system that NSA has . 13 designed, albeit not perfect, is reasonably designed to prevent 14 the acquisition that the statute prohibits. 15 [illegible]: Your Honor, if I could add one thing. 16 If we could take away some of the complication of this 17 discussion by focusing just on other than MCT, for what it's 18 worth, I think the Court has already considered and countenanced 19 the idea that some targets are going to roam into the 20 United States, or we could turn out to be wrong and they would 21 be in the United States. And some statistically, probably a 22 very small percentage of those may have communications with 23 other people in the United States. 24 So, to a certain extent, this may be a threshold the Court 25 has already crossed or at least walked up to and assessed. So I

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000118

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 40

1 would just like the offer that in the context of this question 2 about what the statute means. 3 THE COURT: I think that's a fair point in terms of we 4 all know that systems are not going to be perfect. We've 5 already faced some instances, and this may be another one that 6 is an acceptable imperfection. But that's what we have to 7 explore a little bit more before the Court is prepared to reach 8 that conclusion. 9 We've already talked a little bit about the second part of 10 the statutory term of the "known at the time of acquisition." 11 So it would be your view that the knowledge or the certainty 12 that the collection will result in the acquisition of thousands 13 over the course of a year of wholly domestic communications does 14 not mean that there's a violation of 1881. Set aside the 15 "reasonably designed" language, which I think is important, but 16 just set that aside for a second. 17 [illegible]: I'm not sure I understand the question. 18 THE COURT: Well, you know at the time of acquisition 19 that there are thousands of transactions that are going to be 20 wholly domestic over the course of a year. At the time of 21 acquisition of a particular transaction, you don't know that 22 that transaction is wholly domestic. 23 Your assessment is that even on this knowledge prong of the 24 definition and looking at the "known at the time of acquisition" 25 language of the statute, that the Court should really not be

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000119

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 41 1 troubled by the fact that it's known that there are going to be 2 thousands of wholly domestic communications acquired -- and 3 again, I'm only focusing within the statutory language now; I'm 4 not talking about the Fourth Amendment -- shouldn't be troubled 5 by that because it's not known that any particular transaction 6 is wholly domestic and that the Court therefore need not be 7 troubled by the fact that everyone knows that there will be some 8 wholly domestic communications acquired. 9 [illegible]: It's certainly reasonable to presume that 10 we are going to be acquiring wholly domestic communications 11 despite our best efforts, and that's why, as we've shown in the 12 papers, we're relying heavily on the application of our 13 minimization procedures. 14 So to the extent that the protections that we put into 15 place at the time of acquisition don't work, then we have these 16 substantial back-end protections to ensure that to the extent 17 that one of these domestic communications resides in an NSA 18 system because it couldn't be weeded out at the point of 19 acquisition, and an analyst comes across it during the course of 20 their regular analytical work, that that information is treated 21 appropriately; i.e., we've committed to destroying any wholly 22 domestic communications or MCTs containing even a single wholly 23 domestic communication. So I think the two in tandem work 24 together. 25 THE COURT: I have one or two more questions in this TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000120

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 42

1 intent section, if you will, and then I think it'll be -- well, 2 let me not assume that we won't have a few more than one or two 3 questions. Some of this relates to Judge McLaughlin's 2008 4 opinion, because you do rely on it to a certain extent, back in 5 the June 28 submission, for example. 6 You rely on it for the proposition that NSA's acquisition 7 of transactions continue discrete communications that are not 8 to, from, or about a task selector is intentional because the 9 acquisition of the additional information is a necessary yet 10 unavoidable consequence of acquiring foreign communications to, 11 from, or about a task selector. 12 Judge McLaughlin, in that opinion, also found that a 13 communication would be unintentionally acquired for purposes of 14 1806, if, for example, the acquisition resulted from a technical 15 malfunction or an inadvertent misidentification of a selector. 16 Is the government's argument that its acquisition of wholly 17 domestic communications is unintentional based on NSA's 18 determination that its filters are not functioning properly, or 19 is it instead that the filters have a limited capacity to 20 prevent the acquisition of wholly domestic communications? 21 [illegible]: I think it's both. I think we've 22 previously asserted that to the extent that NSA's filters fail 23 for a technical reason -- and that has happened in the past, and 24 we have reported a compliance incident related to that, and we 25 acquired wholly domestic communications as a result of that --

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000121

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release.

TOP SECRET//COMINT//ORCON, NOFORN 43

1 that is a situation where we've unintentionally acquired a 2 wholly domestic communication. In this instance, it's not -- 3 THE COURT: It's not a failure. Isn't it more a 4 limited capacity? 5 [Redacted]: It's a very small limited capacity. But, 6 yeah, I think it's pretty much -- 7 MR. INGLIS: But, Your Honor, I would offer and add 8 that I think it's more the limited capacity, the limited 9 technical possibility given the way the communications work, 10 that if we had the means to devise it such that it would screen 11 out at that moment in time -- discern, screen out -- we should 12 be expected and would do so. 13 THE COURT: If it's that limited capacity or 14 feasibility, why isn't the acquisition of those communications, 15 to return to Judge McLaughlin's language, a necessary yet 16 unavoidable consequence of acquiring communications to, from, or 17 about a task selector and therefore intentional, under her 18 opinion? 19 [Redacted] Under her opinion, and as we argued to the 20 Court, we were basing our assertion that there were certain 21 types of communications that were intentionally but mistakenly 22 acquired based on our understanding of where the target was 23 also. That's where our primary focus was. 24 This situation is a little bit different than that. In 25 that instance, we were relying wholly on our reasonable but

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000122

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 44 1 mistaken belief that our target was located outside the 2 United States in assessing whether or not, to the extent that 3 that person roams into the United States, 1806(i) applies in 4 that case. 5 We determined that it didn't because we were at all times 6 intending to acquire all communications from our targeted 7 selector under the reasonable but mistaken belief that our 8 target was located outside the United States. 9 In this context, it's a little bit different. Rather than 10 relying on the application of the targeting procedures and 11 relying on our reasonable belief, we are taking active technical 12 measures to prevent the acquisition of wholly domestic 13 communication. That's the distinction that I draw, and when 14 those technical means don't necessarily work, that's when the 15 acquisition becomes unintentional. 16 THE COURT: All right. So if we have an acquisition 17 of wholly domestic communications in a circumstance where [illegible] 18 [illegible] 19 [illegible] 20 [illegible] 21 [illegible] 22 okay, so that's the situation -- would the acquisition of those 23 transactions by the upstream collection be intentional? 24 [illegible]: I just want to make sure we understand the 25 question. TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000123

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 45

1 THE COURT: I'm talking about acquiring wholly 2 domestic communications in circumstances where [illegible] 3 [illegible] 4 [illegible] 5 [illegible] 6 [illegible]. In that situation, wouldn't acquisition through 7 upstream collection be intentional, even though it's based on a 8 mistaken belief that the target was outside the United States? 9 [illegible]: I would argue that it's still intentional, 10 because recall that the IP filtering and [illegible] 11 [illegible] is intended to prevent the acquisition of 12 wholly domestic about communications. In targeting procedures, 13 we rely on the presence of the target to ensure that we're not 14 acquiring any wholly domestic communications. 15 The fact that in practice we apply the IP filters and [illegible] 16 [illegible] to all 17 communications, including those of the target, I think doesn't 18 undermine the notion that we've relied on in the past that to 19 the extent that a person believed to be located outside the 20 United States roams into the United States and we continue 21 acquiring their communications, albeit some of which may be 22 domestic, that that is still intentional but unknowing. 23 THE COURT: All right. [illegible], in particular, did you 24 want to ask anything about the -- I guess it relates back to the 25 [illegible] of the collection?

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000124

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 46

1 [illegible] Sure. So, back in 2009, the government 2 reported an overcollection [illegible] 3 4 5 6 [illegible] overcollection incident. 7 Immediately following that overcollection, the government 8 took it upon itself to purge the overcollected communications 9 from NSA systems in order to ensure that no information had been 10 disseminated by NSA in any form and to deploy improved filtering 11 systems to prevent the future acquisition of such nontarget 12 communications. 13 In Judge McLaughlin's opinion in improving that 14 surveillance following the [illegible] overcollection 15 incident, she relied heavily on those remedial and preventative 16 measures in concluding that the overcollection didn't preclude 17 the approval of the certification and the procedures before her 18 for renewal. 19 The approach being followed here seems quite different in 20 that the NSA's proposing to continue collecting nontarget 21 information as part of Internet transactions and to keep and 22 potentially use much of that information. So can you address 23 the difference in why you're treating them differently? 24 [illegible] I think the key difference is in the [illegible] 25 incident, the overcollection resulted in the acquisition of

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000125

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 47 1 [illegible] 2 [illegible] 3 [illegible] 4 [illegible] 5 [illegible] 6 Here, we are only acquiring transactions because they 7 contain a targeted selector. At the time of acquisition, we may 8 not even know that there are other discrete communications 9 within the transaction that aren't to, from, or about the 10 targeted selector. I think that's a very key difference. 11 Also, another important difference between [illegible] and this 12 situation is, in [illegible], there was a technological fix that could 13 be implemented. The [illegible] was preventable, and NSA did a lot of 14 great work in developing technical means to ensure that those 15 types of overcollections don't happen again. 16 Again, those means aren't perfect and every once in a while 17 something slips though, but by and large, those are technical 18 means that NSA has implemented to prevent the sort of 19 overcollection that occurred in [illegible] Those technical means 20 aren't available here. NSA lacks the technical means to prevent 21 -- or can only acquire whole transactions. They can't unpack, 22 generally speaking, transactions into their component discrete 23 communications [illegible] 24 [illegible] So I think that's another 25 key difference. TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000126

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 48

1 So, just to sum up, it's the nature of the overcollection. 2 In [illegible] there was no nexus to the targeted selector at all, 3 whereas here there is a nexus to the targeted selector that 4 results in the acquisition, presence of the targeted selector in 5 the transaction, and in [illegible] there was a technical fix that 6 could prevent what was overcollection. Here, there's not a 7 technical fix that will enable NSA to conduct the acquisition in 8 a more discrete way. 9 THE COURT: All right. Let's talk about that for a 10 second. Is it clear from your submissions and your assessment 11 of your technology that NSA doesn't currently have or employ 12 technology that would permit it to acquire everything it's 13 authorized to acquire without also acquiring MCTs? 14 [illegible]: That's correct. 15 THE COURT: Is it technically possible -- in other 16 words, within your knowledge of technology, is it technically 17 possible to come up with a means to acquire everything but not 18 MCTs? I know you don't have it right now, but I'm trying to 19 examine whether it's possible. 20 [illegible]: There's sort of two components to that. 21 One is that these technologies are not designed to do that by 22 nature, and as a result, it is very technologically difficult to 23 do that in many cases. Some cases it is impossible, but in 24 terms of us being able to -- let's use [illegible] as an example. 25 Could we develop a technology which could [illegible]

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000127

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 49

1 2 3 However, the viability of that would be very 4 short-lived, largely because 5 . 6 . 7 Compound that by , 8 it becomes a very hard and infeasible task to do, especially 9 considering that NSA is only looking at 10 11 12 13 14 THE COURT: Have you already looked at this issue and 15 made sort of the final determination by NSA that there isn't a 16 means to come up with a technological, as put it, fix? 17 : Yes. We have concluded that it's 18 technologically infeasible to do this. 19 MR. INGLIS: I would go further to say that if 20 21 22 23 24 25 It will necessarily change, and we therefore will always design

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000128

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 50

1 imperfect systems against tomorrow's probabilities. 2 THE COURT: But not to be glib, that's your business, 3 isn't it? Isn't that what you do every day in all of these 4 multitude of settings? 5 MR. INGLIS: That's exactly right, sir, and fairly 6 put. I would say that we cannot perfectly anticipate those 7 changes. So the machinery will fail before we detect it, and 8 then upon detection, we will have to then correct it. 9 MS. MONACO: I might add if I could, Your Honor, in 10 our discussions on this issue preceding this hearing, we also 11 discussed the fact that some of this relies on [illegible] 12 [illegible] 13 [illegible] 14 [illegible] if that's a fair statement for 15 my colleagues. 16 THE COURT: All right. Let's talk about minimization 17 for a second. More than a second. In the most recent 18 submission, NSA has indicated that it will require any analyst 19 who wants to use a discrete communication within an MCT to first 20 perform checks to determine the locations of the users of 21 electronic communications or accounts or addresses, identifiers 22 referenced in that discrete communication "to the extent 23 reasonably necessary" to determine whether that communication is 24 wholly domestic. 25 If the analyst determines that the active user is a task

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000129

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 51

1 selector or is located outside the United States, then no 2 further checks would be done. If the active user is not a 3 tasked selector or is not determined to be located outside the 4 United States, what happens? 5 [illegible]: If the active user is determined to be not 6 in the United States? 7 THE COURT: I think the way I'd actually say it is the 8 active user is not a tasked selector or is not determined to be 9 located outside the United States, what do you do then? Can the 10 analyst use the communication, or are there additional checks 11 that are necessary? 12 [illegible]: There were additional checks. In that case 13 I think there were [illegible] of those that were identified in that 14 filing, and for each of those the analyst went into deeper 15 technical analysis of all those discrete communications and 16 determined that those discrete communications were not wholly 17 domestic. 18 [illegible]: But how can you describe the further 19 checks? 20 [illegible]: They went into the content and verified 21 that all of the [illegible] 22 contained at least one foreign recipient. 23 THE COURT: That was done in the context of the 24 sampling, you mean? 25 [illegible]: That was done in the evaluation of the case

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000130

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 52

1 where the active user was located within the United States. 2 THE COURT: Is that what an analyst is going to do 3 tomorrow when they have this circumstance? 4 [illegible]: If they were going to use that data, that's 5 what they would do. That's correct. 6 THE COURT: And if the analyst is unable to determine 7 that at least one of the communications is outside of the 8 United States, what happens then? 9 [illegible]: If it was inconclusive, it would be likely 10 that they would not use that piece of data. 11 THE COURT: Is that what the protocols and policies 12 require? Can the analyst use the communication, or is the 13 default, if you will, that he or she has to treat it as a wholly 14 domestic communication? 15 (Pause) 16 If you need a second to consult, by all means take it. 17 There's no harm to doing that. You're free to do so. 18 [illegible]: Thank you, Your Honor. 19 (Attendees conferring.) 20 [illegible]: Unless we could confirm, we would not use 21 that piece of data. 22 THE COURT: So that would be the default position. 23 Just for a second, you referred to these [illegible] checks, and 24 my question is really whether those are going to be used going 25 forward, and where do we find in the submissions that you've

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000131

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 53

1 indicated that those are going to be used going forward? 2 MR. INGLIS: Your Honor, I think the answer to your 3 first question is yes, the checks that we have imposed will go 4 forward, and as to where they are documented at that level of 5 detail, we'll determine. 6 [illegible]: Your Honor, I think you're referring to 7 things that are detailed on page 8 and 9 of the 30 August 8 submission which details the very steps that NSA would take. 9 THE COURT: 8 and 9? 10 [illegible]: Yes, Your Honor. 11 THE COURT: Okay. 12 [illegible]: Your Honor, we may be mixing things. 13 THE COURT: Talking past each other or apples and 14 oranges? 15 [illegible]: If you look at the August 30 submission, 16 the first clarification that the government makes is one 17 regarding wholly domestic communications and if an analyst who 18 is confronted with an MCT wants to make use of some discrete 19 communication within it, they will first do the checks that 20 we've just been talking about to determine whether or not that 21 communication is wholly domestic. 22 And I don't know that we got into it at this level of 23 detail, but if you look at page 3, we do there talk about what 24 the analyst will do in order to aid the analyst in attempting to 25 recognize whether or not they are in fact dealing with a wholly

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000132

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 54

1 domestic communication. 2 THE COURT: All right. And we can refer back to 3 those. Thanks. 4 All right. Now, in the August 30 submission, another point 5 made -- I think it's on page 9 -- is the government indicates 6 that if NSA acquires, through this upstream collection, a 7 discrete communication that is not to, from, or about a tasked 8 selector but is to or from an identifiable U.S. person, that 9 communication cannot be used for any purpose other than to 10 protect against an immediate threat to human life. 11 Who's going to make that determination? An analyst? Or is 12 there a process for how that determination is made and who makes 13 that determination? 14 : Your Honor, is the determination you're 15 talking about whether or not the specific item could be used or 16 whether or not the U.S. person could be identified? 17 THE COURT: Whether it can be used. 18 MR. INGLIS: It could only be used to protect the life 19 of a person. We have a defined process at National Security 20 Agency in which we involve our general counsel in in order to 21 make that determination. 22 : And, Your Honor, if I may, that language 23 is similar to language that appears in NSA's minimization 24 procedures at the very beginning, basically a threat-to-life 25 carve-out. But like that threat-to-life carve-out and the

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000133

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 55

1 minimization procedures, there's complete transparency on the 2 part of the government because if NSA needs to take that action, 3 they need to report that to DOJ and ODNI, and we in turn have to 4 report that to the Court. 5 THE COURT: I understand that. I'm still interested 6 in how that determination is made. The fact that it's reported 7 to the Court later on that a determination was made and it was 8 used is important, but it's also important to know that there's 9 an appropriate process for making that determination within NSA. 10 Now, if there's no immediate threat to human life in that 11 same circumstance, does NSA retain the communication? 12 [illegible]: We would, Your Honor. 13 THE COURT: And is it marked in some way to indicate 14 that it cannot be used without that determination being made, or 15 is it just there? 16 MR. INGLIS: Your Honor, I do not believe we marked 17 that communication at that moment in time. 18 THE COURT: Do you think there's a need to, or do you 19 think an analyst in the future, when confronted with this 20 communication and making a decision on whether to use it, will 21 be sufficiently apprised of the fact that it's a wholly domestic 22 communication and that he or she needs to follow this process in 23 terms of a threat to human life? 24 [illegible]: It wouldn't be a wholly domestic 25 communication.

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000134

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 56

THE COURT: I'm sorry. It's not a wholly domestic communication. You're right. [illegible]: I think that it's pretty clear that it would be better if it were marked than not marked if the government knows about it. The feasibility of doing that I don't think we've assessed, so I don't think we can commit sitting at the table to do that. But we will seriously consider whether it's possible to do that, or when we review it, if it's tentatively possible. MR. INGLIS: And so in the absence of that marking the confidence that is placed upon common-training standard that the analyst had, the fact that if it were determined to be wholly domestic a further check that it would be purged in all of its derivatives as well. THE COURT: Let's move from the specific to the more general. Is it possible for NSA to segregate the upstream collection from the rest of its 702 collection? [illegible]: It's possible. [illegible]: The upstream collection in comparison to our PRISM collection, as we've referred to it, they are commingled in the database, but they are marked in such a way such that they can be identified from distinguishing sources. THE COURT: So in a sense, it is separated by being marked. It's all commingled in the same database. Is that commingling in the same database just as good as a complete

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000135

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 57

1 separation in terms of the use of tools to pull out the 2 communications that can easily be identified as to, from, or 3 about a task selector? In other words -- well, I think the 4 question is a little confusing. 5 MR. INGLIS: Is your question, Your Honor, whether it 6 can be better or worse depending upon how you choose to organize 7 the data? 8 THE COURT: If you separate it rather than commingled 9 and marked, would it be easier to deal with the concerns that 10 exist with respect to a portion of the upstream collection 11 technologically? 12 [illegible]: I don't think that the commingling is a 13 factor in determining and marking the data for the analyst to 14 know that this is from upstream versus not upstream or that 15 there is an MCT involved or not. I don't think that the 16 commingling is a factor of that. 17 MR. INGLIS: So, Your Honor, not to extend the 18 conversation into an inappropriate corner of little interest to 19 the Court, but NSA's strategy writ large for its technical 20 architecture is in the face of increased commingling to 21 concentrate on the marketing of such data element such that we 22 can then determine its provenance. 23 Increasingly, what we'll have is many variables with 24 respect to the origins, the policies that pertain to data and 25 many variables with respect to the authorities and privileges of

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000136

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 58 1 individuals, and as that M meets N, the prospect of coming up 2 with M times N, different instantiations of data sets that are 3 physically separate, that's infeasible for us. So our strategy 4 is to go towards the marking and to allow commingling to happen, 5 not sloppily, but there's no real other tenable design. 6 THE COURT: I have a couple more questions on 7 minimize, but go ahead. 8 [illegible]: If I could just follow up on that. 9 I guess what we're trying to get at here is whether it was 10 technologically feasible to separate this data out given that 11 there are concerns about what's in it -- wholly domestic 12 communications, lots of U.S.-person information, nontarget U.S. 13 person information -- whether you physically separate it or 14 separate it via marking rules, access rules. 15 Is it possible to segregate that, such that you could then 16 limit access to it? I guess one of our concerns is, as I 17 understand it, it comes in to your database and pretty much 18 immediately is made available to analysts running queries. So 19 if something is responsive to a query, it will come up and they 20 will see it. 21 We were curious as to whether it would be technologically 22 possible to sort of either segregate it and sort of immediately 23 or soon thereafter pull out the stuff that you know is okay and 24 make that available, or, alternatively, or perhaps in addition, 25 limit the access of the stuff that's problematic to people who TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000137

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 59 1 are specially trained or subject to certain additional access. 2 MR. INGLIS: If I may, I think I hear two questions, 3 but correct me if I got that wrong. The first is can there be, 4 upon the collection and the immediate presentation of that 5 collection to either machine or a person, some distinguishing 6 characteristic, either some physical segregation or some mark 7 that goes along with that, atomically bound to that, such that 8 there's no doubt where this came from. It's either upstream, or 9 it's all other. I think the answer to that question is yes. 10 THE COURT: If you can mark it, presumably you can 11 segregate it. 12 MR. INGLIS: Yes. But I think the further implied 13 question is, is there then some meaningful processing that might 14 occur so that you can then winnow and filter that material 15 before presentation, before some manual application procedures, 16 and we don't yet know of those. We have thought hard about 17 that, and we don't yet know what further processing might occur 18 absent introducing the human into the loop and having that human 19 follow a rule set that would help determine what the further 20 provenance or not of that data might be. 21 [illegible] But you have the capability to put the 22 human in the loop. You have this specially trained cadre of 23 analysts who are working with your manual database. 24 Theoretically, you could have a specially trained cadre -- 25 MR. INGLIS: So given the answer to the first TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000138

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 60

1 question, is it possible to alert the human beings that come 2 into contact with that data that there are certain rules that 3 pertain that you must handle this data, ask questions of this 4 data in a certain way. 5 MS. MONACO: As the Court sees from the August 30 6 submission, that's exactly what the proposal is to do going 7 forward, to apply a banner to this collection so that would 8 effectuate the alert Deputy Director Inglis talked about, and 9 also that coupled with the training and guidance for the analyst 10 working in concert would apply the protections to that 11 potentially problematic material. 12 THE COURT: Let's go on to another minimization area. 13 Under section 3(b)(1) of the NSA minimization procedures, "NSA 14 personnel will destroy inadvertently acquired communications of 15 or concerning a United States person at the earliest practical 16 point in the processing cycle at which the communication can be 17 identified either as clearly not relevant to the authorized 18 purpose of the acquisition or as not containing evidence of a 19 crime," and that's all in quotation marks. 20 But in the June 1 submission, I think at page 22, it's 21 indicated that NSA cannot destroy a discrete communication 22 within an MCT without destabilizing and potentially rendering 23 unusable some or all of the collected transaction including the 24 single discrete communication which is to, from, or about the 25 task selector.

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000139

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 61

1 So, assuming the communication of or concerning a U.S. 2 person is not a wholly domestic communication, would NSA retain 3 the entire MCT for five years regardless of that destruction 4 requirement in section 3(b)(1)? 5 [illegible]: That's correct, yes. 6 THE COURT: And would the communication of or 7 concerning a U.S. person be available to NSA analysts during the 8 entire five-year period? 9 [illegible]: So again, Your Honor, I think this goes 10 back to an earlier question you asked, which is would we mark it 11 once we discover that's true. I think that's a valuable thing 12 to do and to look into. For reasons already described by 13 [illegible] we couldn't eliminate that piece of the 14 communication without eliminating the whole communication. 15 So the line we're proposing here is, if we find a domestic 16 communication within the series of communications, we'll destroy 17 the whole transaction. If we find untargeted person 18 information, we won't destroy, but we won't use that either. 19 THE COURT: Now, these communications will start 20 unminimized and eventually be minimized. Does NSA share 21 unminimized MCTs acquired through the upstream collection with 22 any other agencies? CIA? FBI? 23 [illegible]: No. At this time the only collection 24 that's shared with the CIA or FBI is from the PRISM side, 25 [illegible]

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000140

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 62

1 THE COURT: So the upstream collection is not shared 2 at all. 3 [illegible]: Correct. 4 THE COURT: Except on an individual basis once an 5 analyst has focused in on a particular -- 6 [illegible]: Not shared wholesale. That's correct. 7 THE COURT: How about with foreign governments for 8 translation purposes or analytical purposes under section 7(b) 9 of your minimization procedures? 10 [illegible]: [illegible] 11 [illegible] 12 THE COURT: So just to close out the minimization, 13 unless any of the legal advisors have questions, the manual 14 review process is not going to be continued prospectively. 15 Correct? In other words, NSA proposes to rely on analysts who 16 recognize wholly domestic communications within transactions 17 when they're confronted with them. 18 MR. INGLIS: Sir, by the manual of due process, you 19 mean the process by which we examined the slice of 50,000? 20 THE COURT: Yes. 21 MR. INGLIS: We don't intend to carry that forward. 22 We intend to carry the training standard forward for the 23 analysts who encounter the data. 24 THE COURT: So wholly domestic communications that are 25 never viewed by analysts will remain in the repositories for

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000141

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 63

1 five years. 2 [illegible]: That's correct. 3 THE COURT: And with transactions that are reviewed, 4 should the Court be confident that NSA analysts would be able to 5 recognize wholly domestic communications when confronted with 6 them in their daily work? There's been -- even in this sampling 7 process, there's been some difficulty in recognizing wholly 8 domestic communications. 9 Should we be confident that analysts will be able to do 10 that? Presumably, they're going to have less in the nature of 11 tools, time, and resources than were employed during this 12 analytical exercise. 13 [illegible]: Your Honor, I think the Court should be 14 confident, and really for a couple of reasons. First off, the 15 10 domestic communications that were confirmed during the review 16 of the sample were all communications that when you actually 17 looked at them they were not responsive to any foreign 18 intelligence requirement whatsoever, so it was unlikely that in 19 the normal course of business our analyst would have pulled them 20 up in the first instance. Therefore, by operation of NSA's 21 minimization procedures just sort of by standard, they would 22 have aged off in five years. 23 But secondly, the amount of training, the notices to the 24 workforce, the related efforts to be sure that the workforce is 25 aware of the problem I think will sensitize our analysts to be

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000142

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 64 1 on the lookout for the possibility that an MCT that was 2 responsive to a query they made of a database might be a wholly 3 domestic communication and therefore they should go through the 4 steps that we've outlined in the filings to make sure that 5 they've properly accounted for it and made sure that it is not 6 in fact a wholly domestic communication. 7 MR. INGLIS: Your Honor, if I might add, they're not 8 advised to go through those checks; they're compelled to go 9 through those checks. What we found as a practical matter in 10 this particular activity where we manually examined the 50,000- 11 plus transactions is part of the difficulty in having analysts 12 identify wholly domestic communications is that they were not in 13 practice in terms of finding any meaning or use in them. 14 So we actually had to have them pursue matters that more 15 often than not they would say that is of no interest to me; I 16 wouldn't pursue that naturally. And so then becoming 17 comfortable with the artifacts that are associated with wholly 18 domestic communications that are otherwise uninteresting and 19 otherwise not things that they would daily work with was in fact 20 in my view an opportunity to understand why they then would 21 necessarily go after those things that are directly responsive 22 to their queries, as opposed to those things that were 23 incidentally collected. 24 THE COURT: Now, I have one more of area of 25 questioning that's really under the umbrella of the Fourth TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000143

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 65

1 Amendment and probably 15 minutes maybe at most, but there may 2 be other things that we'll need to discuss and other things that 3 you may want to say. 4 In asking this question I'm concerned about all of you, but 5 I'm most concerned about [illegible] the court reporter, and that 6 is whether we should take a short break now and resume in five 7 to 10 minutes, probably for another -- it depends how much you 8 would to say after I finish with the questions. 9 The questions themselves probably won't take more than 15, 10 at most 20 minutes, but I don't know what further we might need 11 to discuss from your perspective. Anyone think it would be 12 advantageous to take a short break? 13 MS. MONACO: Well, Your Honor, I don't want to impose 14 on [illegible]-- 15 THE COURT: I'll ask him independently. He may answer 16 that question yes, and that'll be the end of it? 17 [illegible] do we need to take a short break? 18 COURT REPORTER: Five minutes would be fine. 19 (Recess from 1:07 p.m. to 1:21 p.m.) 20 THE COURT: Let's continue. I want to move to the 21 Fourth Amendment, but by moving to the Fourth Amendment, I'm not 22 really totally moving away from a statutory assessment of 23 targeting. Both of them have a reasonable component in the 24 statutory targeting assessment that may be a reasonably designed 25 component, but in the Fourth Amendment, there's also much more

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000144

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release.

TOP SECRET//COMINT//ORCON, NOFORN 66

1 of a reasonable aspect. 2 We've assumed, in prior 702 dockets, that at least in some 3 circumstances, account holders have a reasonable expectation of 4 privacy in electronic communications and therefore that the 5 acquisition of electronic communications can result in a search 6 or seizure within the meaning of the Fourth Amendment. 7 Consistent with the position that it's taken in other matters 8 before the FISC, the government does not assert otherwise in the 9 multiple filings in this matter. 10 Can the Court infer, therefore, that the government does 11 not disagree with the proposition that the acquisition of 12 electronic communications can result in a search or seizure 13 within the meaning of the Fourth Amendment? 14 [Redacted] I think so. 15 THE COURT: All right. I'll take "I think so" as a 16 yes. 17 (Laughter) 18 [Redacted] I'll clarify my remarks for the record. 19 Yes. 20 THE COURT: Let's talk about incidentals. The 21 government's position is that the collection of wholly domestic 22 communications as part of the Internet transactions that NSA 23 acquires through this upstream collection is incidental for 24 purposes of the Fourth Amendment. 25 Now, as the government acknowledges, the mere fact that an

TOP SECRET//COMINT//ORCON, NOFORN

NYT v DOJ, 16 CIV 7020_000145

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 67

1 intrusion is incidental does not necessarily render it 2 reasonable and thus permissible under the Fourth Amendment, and 3 I think you've acknowledged that in one of the June submissions, 4 the June 28 submission probably. 5 And then in the August 16 submission, it's indicated that, 6 as we've talked about, that NSA acquires at least 2,000 to 7 10,000 transactions annually that contain one or more wholly 8 domestic communications that are not to, from, or about a task 9 selector, at least that's what this statistical analysis shows. 10 Now, do those numbers alone establish a fairly substantial 11 intrusion on the protected Fourth Amendment interest, 12 particularly when you consider that the actual number of wholly 13 domestic communications may be higher depending upon how 14 convincing you are in a further assessment of the unknown 15 category of communications, also considering that each 16 transaction may actually contain multiple communications, and 17 also taking into account that many of the persons whose 18 communications are being acquired have little or no connection 19 to the user of the task selector? 20 Some persons outside the United States are protected under 21 the Fourth Amendment, and also we don't know what this category 22 of about communications in the four to 5,000 of the 50,000 that 23 were set aside as being discrete communications will turn up. 24 Just in terms of the numbers, it looks like there's a 25 fairly substantial intrusion on protected Fourth Amendment

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000146

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 68

1 interest, doesn't it? 2 [illegible]: I think that's right, Your Honor. I think 3 there is clearly an intrusion upon Fourth Amendment protected 4 interest, but in conducting the Fourth Amendment balance, we 5 also have to consider the protections that are put into place to 6 protect those Fourth Amendment interests. 7 And as we've asserted, we're committed to destroying any 8 wholly domestic communications that, for whatever reason, have 9 run through NSA filters and happen to land in an NSA repository 10 and are recognized by NSA analysts as such. 11 So, yes, even though the Fourth Amendment intrusion may be 12 greater, we are taking basically the ultimate step that we can 13 in minimizing the effects of that intrusion by destroying any 14 MCTs that have wholly domestic communications within them. 15 THE COURT: And part of the assessment, of course, is 16 a balancing assessment under the Fourth Amendment that looks, as 17 we're considering the certifications and the procedures and 18 whether they satisfy the Fourth Amendment, the government would 19 stress the importance of the upstream collection to national 20 security. 21 I want to just assess that a little bit and get a little 22 bit more of a sense of that. Just numerically, the collection 23 of Internet transactions really is a pretty small part of the 24 collection as a whole. We're not talking about telephonic 25 communications, and according to the numbers, the upstream

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000147

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 69

1 collection in the aggregate only constitutes about 9 percent of 2 all Internet communications acquired pursuant to 702. We have 3 the PRISM collections, etc. 4 And then based on the sample and the analysis applied, only 5 10 percent of that 9 percent, which results in .9 percent, or 6 .009 of the overall collection of Internet communications is 7 Internet transactions containing multiple communications. 8 That would suggest that the collection of Internet 9 transactions through the upstream collection, and particularly 10 the MCTs that we're concerned about, isn't a particularly 11 critical national security tool. It's a really small part of 12 what NSA collects. Why should I conclude that it's so vital, 13 looking at that side of the balance? 14 MR. INGLIS: Your Honor, I would offer and then pass 15 to [illegible], perhaps [illegible] from the operational side of the house, two 16 points. One is that it uniquely covers a scene [illegible] 17 [illegible] 18 [illegible] 19 [illegible] 20 [illegible] 21 [illegible] 22 [illegible] 23 [illegible] 24 THE COURT: [illegible] 25 [illegible]

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000148

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 70

1 MR. INGLIS: [illegible] 2 [illegible] 3 [illegible] 4 [illegible] 5 [illegible]. 6 The second is that this collection [illegible] 7 [illegible] 8 [illegible] 9 [illegible] 10 [illegible] 11 [illegible] 12 [illegible] 13 [illegible]. 14 So a significant percentage of our counterterrorism 15 reporting is ultimately attributable to what we do in 702, and 16 albeit the percentages would show that the upstream is a 17 relatively small percentage of that, but it covers a unique 18 scene [illegible] we can't get other ways. 19 [illegible]: With respect to technology parsing out 20 those MCTs, we don't have the technology to eliminate the MCTs 21 without also eliminating the bulk of the discrete 22 communications. We'd lose all of that except for [illegible] 23 [illegible] that we fully understand. 24 THE COURT: This is a unique Fourth Amendment context, 25 as many legal settings will be unique, and therefore, the Court,

TOP SECRET//COMINT//ORCON, NOFORN

NYT v DOJ, 16 CIV 7020_000149

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 71

1 I am having some difficulty in assessing just how the balancing 2 under the Fourth Amendment is applied. But if the primary value 3 of the upstream collection is the acquisition of a fairly small 4 number of communications that have high value to national 5 security, how does the Court balance the possibility of such 6 acquisitions against the known acquisition of at least thousands 7 of nontarget communications each year? 8 Another way of putting it would be, does the possibility 9 that NSA is going to acquire a single piece of valuable 10 intelligence outweigh the privacy interest of these thousands of 11 U.S. persons and persons in the United States whose 12 communications are being incidentally, if you will, acquired? 13 How do I make that assessment and reach that balance? Is 14 the intelligence value always going to rule the day? Is one 15 valuable piece of intelligence enough to justify the thousands 16 of Fourth Amendment intrusions that are involved here? 17 Your Honor, I think what was demonstrated 18 in our sample, and I want to perhaps correct a statement there 19 that we get relatively few communications through upstream 20 collection that are of intelligence value. I think what our 21 sample showed was that we get millions of Internet 22 communications that are of potential intelligence value and a 23 very small number of those which have the potential to be wholly 24 domestic. 25 MS. MONACO: I think I would follow up on that, Your

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000150

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 72

1 Honor, by saying that the Court has posed, I think -- put it in 2 somewhat more of a starker context than I think the government 3 has in its filings. In other words, the Court has already 4 acknowledged that the FI, or the foreign intelligence 5 information sought through the to, from, or about task selectors 6 is of paramount importance to the government and is of the 7 highest order of magnitude. I think that is the one side of the 8 balance that the Court is faced with, and we would not suggest 9 that it should be put in such stark terms. 10 In other words, on the other side of the balance, what 11 we've tried to do in the study that NSA has done is indicate the 12 relatively small portion, understanding the questions that the 13 Court has raised at the beginning of the hearing about those 14 numbers, but we've tried to put in place a series of measures 15 that would enhance the ability of the Court to make its 16 reasonableness finding in terms of the analysis that gets done 17 as the information is seen by the analyst. 18 So I would suggest to the Court that, no, setting one 19 potential piece of foreign intelligence information against the 20 privacy interest of potentially thousands of domestic 21 communications is, I respectfully suggest, not the appropriate 22 analysis but rather the paramount interest of the foreign 23 intelligence contained in those MCTs and our inability to get at 24 that paramount important information other than by collecting 25 the whole MCT.

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000151

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 73

1 Balance that against the procedures and the number of steps 2 that the government is proposing to be put in place to ensure 3 that there is the maximum amount of privacy protection that can 4 be applied to those potentially problematic pieces of 5 communications. 6 THE COURT: Would we all agree that ultimately the 7 assessment or application of this balancing under the Fourth 8 Amendment might be different depending upon whether they're in 9 this fairly substantial upstream collection that is still a 10 limited percentage of the total collection but the outcome of 11 the Fourth Amendment balancing might be different depending upon 12 whether the collection of wholly domestic communications was 13 1,000 a year, 10,000 a year, 100,000 a year, 10 million a year? 14 It would depend upon what those numbers show, if you will, 15 that at some point the collection, even though unintended or 16 incidental, at some point the acquisition of a large volume of 17 Fourth Amendment protected communications would simply be too 18 much under the Fourth Amendment? Would we all agree that there 19 is some tipping point? 20 MR. INGLIS: Your Honor, I would agree from an NSA 21 perspective that there is. In the extreme, there must be a 22 line. It's probably not objectively determinable, but there 23 would be a line. If the preponderance of material that we 24 picked up was in fact wholly domestic or declared incidental but 25 in fact it was the preponderance of what we picked up, clearly

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000152

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 74 1 we'd be in the wrong place. 2 THE COURT: I think that's a good way to put it, 3 Mr. Inglis, that if you have a preponderance that's in the wrong 4 category, then maybe that would be too much. 5 MR. INGLIS: So if I might, given the opportunity to 6 answer the question in a subjective manner, complement 7 Ms. Monaco's answer, I would offer three things as to whether 8 this achieves a reasonable balance between those competing 9 concerns. 10 The first is that it does offer unique material that we 11 believe could provide valuable foreign intelligence. If it were 12 not unique, if we could in fact make up for this some other way 13 and given the problematic nature of this, I think that we would 14 quickly go to that corner, but we haven't found another way to 15 go after what we see as unique material. 16 The second is that we've taken what we believe are all 17 reasonable measures in the technology and a set of then equally, 18 if not more so, reasonable procedures of how we then use the 19 fruits of what that technology provides to address the real and 20 material concerns about Fourth Amendment, statutory, or the 21 Court's authority, have we made reasonable use of that such that 22 we are focusing the majority, the preponderance of our efforts 23 after the greater purpose and at the same time an equal amount 24 of time and effort to make sure that we don't then incidentally 25 or intentionally collect the wholly domestic.

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000153

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 75

1 The last thing I would say is that in equating whether one 2 report is the equal of or is offsetting against incurring 3 intruding upon the privacy of thousands of individuals, that's a 4 hard call, but the one report could in fact protect millions of 5 individuals, depending upon the density of the population it is 6 under, is it at risk, and what that report might pertain to. So 7 that's the great unknowable here is how valuable is that single 8 piece of intelligence. 9 THE COURT: One could use that accurate observation to 10 support the conclusion that one piece of intelligence does 11 outweigh whatever the Fourth Amendment intrusion is, and I think 12 if most reasonable people actually weighed that one piece of 13 known intelligence that would save millions of lives against any 14 number of Fourth Amendment intrusions, they'd say, yes, get that 15 piece of intelligence. 16 MR. INGLIS: I would argue that, sir, but I would not 17 be completely impartial in making that argument. 18 THE COURT: All right. Just one or two small 19 questions, and that will be the end of this inquisition. Is NSA 20 able to acquire [illegible] that are to, from, or 21 about a task selector from a service provider? 22 [illegible]: Yes, we are. 23 THE COURT: So, what would NSA acquire in the upstream 24 collection in those categories that could not be acquired from 25 the provider, anything?

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000154

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 76 1 [illegible]: What could we obtain from, for instance, 2 [illegible] from the upstream world that we couldn't obtain from the 3 provider? 4 THE COURT: Right. Do you get anything more through 5 the upstream collection [illegible] 6 [illegible] 7 [illegible]: Yes. [illegible] 8 [illegible] 9 [illegible] 10 [illegible] 11 [illegible] 12 [illegible] 13 [illegible] 14 [illegible] 15 [illegible] 16 [illegible] 17 [illegible] 18 [illegible] 19 [illegible] 20 [illegible] 21 [illegible]: Your Honor, if I may, just to take this up 22 another level of generality. In general, NSA cannot do abouts 23 collection at the ISP, so that [illegible] that 24 would be entirely off the table, [illegible] 25 [illegible]: I also need to add that we do not obtain TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000155

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 77

1 abouts collection from 2 . 3 THE COURT: Okay. So back to the evolution and the 4 dynamic nature of the Internet, and I'm curious as to how this 5 phenomenon affects the Fourth Amendment balancing given that the 6 upstream collection is going to be constantly changing because 7 of the nature of the Internet. 8 If the government can't predict just what the impact of any 9 changes in the Internet will be on the collection, is it valid 10 for the Court to presume in a Fourth Amendment analysis that the 11 scope and intrusiveness of the upstream collection will expand 12 or contract or stay the same? 13 What should the Court presume in applying a Fourth 14 Amendment analysis? Just take it as it is frozen now, or is it 15 valid to make some presumption or assumption with respect to the 16 future given what the experts see with respect to the Internet 17 and its evolution? 18 MR. INGLIS: I'll leave it to my Department of Justice 19 colleagues to speak on the Fourth Amendment implications, but I 20 would say that the Court can and should assume that it will 21 change. I don't know whether to the greater or the lesser 22 benefit of the interest between the government and the Court, 23 but it will change. And I think the expectation of the Court 24 upon the government is that the government will discern that 25 change and faithfully either stay within the authorities granted

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000156

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. [TOP SECRET//COMINT//ORCON, NOFORN] 78 1 by the Court or come back to the Court and argue that those 2 authorities must be modified in some way, shape, or form based 3 upon those changes. 4 THE COURT: So, in this setting, with the 5 certifications and the Fourth Amendment, if the Court is 6 comfortable, based on the record before it not withstanding the 7 fact that there will be changes, there will be evolutions, the 8 Court should simply rely on the renewal process and the 9 obligation of the government to come forward with any additional 10 information, changes in the technology or what have you, in 11 order to assess a future certification. 12 MS. MONACO: Your Honor, I would add to that that 13 certainly the Court should rely on that, but I would add to that 14 the Court should also hold the government, quite obviously, to 15 the procedures that we've discussed would apply in this case 16 precisely because we understand the evolving nature of the 17 Internet, precisely because we understand the real, and I would 18 say nontrivial, intrusion that we've all acknowledged and 19 discussed here, given the nature and collection of MCTs. 20 And it's precisely because of that that I think the 21 government would propose to put in place these series of steps 22 in what I guess we've termed a multilayered approach to try to 23 do the utmost in terms of being able to enhance the privacy 24 protections that exist in the existing minimization procedures 25 that can be applied to that potentially problematic collection. [TOP SECRET//COMINT//ORCON, NOFORN] NYT v DOJ, 16 CIV 7020_000157

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 79

1 THE COURT: All right. Any other questions legal 2 advisors have? 3 [illegible] Yes. Can I just ask one quick question? 4 Do you have a rough estimate as to the percent of Internet 5 transactions that are actually handled by [illegible] 6 [illegible] just a rough? No? Lots? Few? Anything? 7 THE COURT: Certainly you've said that because of the 8 place of the United States in the Internet, there's a very heavy 9 use of -- 10 MR. INGLIS: So two things remain true, which is the 11 majority of the world's reliable communications continue to flow 12 through the United States, largely because of the investments 13 made over the last 40, 50 years in the creation of the Internet 14 and also because of the inherent stability, resilience of the 15 infrastructures that then underpin that. 16 There's also a degree of innovation that continues to 17 center on the United States, and despite its best efforts, 18 China, Russia, others, have not yet created the engine of 19 innovations that have taken those products and materials 20 offshore, but that has begun to occur. 21 So you do see a flattening of the innovation, you see a 22 slight flattening in terms of the infrastructure. So today, and 23 I think for the foreseeable future, the preponderance of 24 communications are still centered in through the United States. 25 But I can see a world 50 years from now where that might be

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000158

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 80 1 completely flat. I can't predict that far out. I just know 2 over the next year or two we can depend upon the trends that 3 extend from 10 years past. 4 [illegible] And another thing we can say which may 5 be obvious, with respect to our targets, particularly [illegible] 6 [illegible] 7 [illegible]. 8 MR. INGLIS: And I've been pleasantly surprised from 9 an intelligence perspective [illegible] 10 [illegible] 11 [illegible]. I would never 12 have predicted that 10 years ago, and yet it's been 13 extraordinarily lucrative for us. 14 THE COURT: All right. Any other questions? 15 So, let me make one or two observations, and we'll talk 16 about what further you can provide. And I do note that I've 17 kept you here for a long time, and I do know that some among you 18 may have other places to get to. 19 It seems to me that to the extent that either the statutory 20 or the Fourth Amendment analysis turns on the record and 21 statistics, if you will, and numbers of potential or actual 22 wholly domestic communications that are acquired, things of that 23 sort, we have a situation where we have this -- I'll call it 24 fairly low number, at least in terms of percentages, of between 25 1 and 5,000 every six months having identified through your TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000159

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 81 1 statistical analysis as containing wholly domestic 2 communications that are to, from, or about a task selector. 3 We then have a larger category that I refer to as "unknown" 4 that is identified as between 48,609 and 70,168 every six months 5 that you've indicated you've made some assessment of that takes 6 you out of the category of totally unknown and into a category 7 of highly unlikely, at least, or whatever term you want to 8 apply, or can under the facts apply, highly unlikely to contain 9 wholly domestic communications. 10 I'm not yet convinced that the record before me supports 11 the conclusions that have been articulated. So I will look for 12 further information on that to bolster that assessment. 13 We also have this other category, in your random sampling 14 again, that is 9/10ths of the random sampling that was set aside 15 as being discrete communications -- 45,000 out of the 50,000 -- 16 as to which our questioning has indicated we have a concern that 17 some of the about communications may actually have wholly 18 domestic communications. 19 And I don't think that you've really assessed that, either 20 theoretically or by any actual examination of those particular 21 transactions or communications. And I'm not indicating to you 22 what I expect you to do, but I do have this concern that there 23 are a fair number of wholly domestic communications in that 24 category, and there's nothing -- you really haven't had an 25 opportunity to address that, but there's nothing that has been TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000160

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 82

1 said to date that would dissuade me from that conclusion. So 2 I'm looking there for some convincing, if you will, assessment 3 of why there are not wholly domestic communications within that 4 body which is 9/10ths of the random sample. 5 Those things, and anything else that we identified, really 6 was right at the outset of our discussion today, it would be 7 very helpful to receive any further information on those. We 8 don't have much time, though. 9 So I think in fairness to the Court, which is very 10 constrained in the time available now to resolve and write up 11 any resolution, particularly if I conclude that I cannot totally 12 approve the certifications, you need to get that to me as 13 quickly as you can. I have to say even this week. It would be 14 very difficult if we didn't have it this week to formulate and 15 frame the resolution of these matters and get it committed to 16 writing. 17 Anything else you want to say I would be happy to receive, 18 and I don't want to give the impression that this is the last 19 point of communication even in addition to the written response 20 that you may have in the next couple of days. If there's more 21 to be discussed, then the Court is, in the words of Ross Perot, 22 all ears. 23 MR. INGLIS: Your Honor, if I might ask a question 24 regarding your summary, you had said at the outset of the 25 conversation expressed a possible concern that the purged data

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000161

Names of certain hearing participants withheld under b(6) and b(7)(C).
All withheld information exempt under b(1) and b(3) unless otherwise noted.
Approved for public release.
~~TOP SECRET//COMINT//ORCON, NOFORN~~
83

1   might constitute an unexplored piece of the territory, to the
2   extent that it would be useful for us to describe the attributes
3   of those things purged and for what reasons, we'd be happy to
4   provide that as well.
5           THE COURT: Anything you can provide on that, which
6   really goes not to the -- well, it goes to sort of a generalized
7   assessment of the validity of the representative sample.
8           MS. MONACO: Your Honor, if I might suggest, we
9   understand the constraints on the Court's time on this, and we
10  very much appreciate all the opportunities the Court has
11  afforded the government to provide it additional information.
12          I think what we will do is go back and confer and ensure
13  that we can provide you whatever it is along these lines that we
14  can by the end of the week and be in touch if the Court permits
15  with the legal advisors on the precise timing on that.
16          THE COURT: They are always open to such
17  communications.
18          MS. MONACO: And then what I would also say is I know
19  we've had some of this contact I think at the staff level, but
20  looking ahead I think we'll also be in touch on potential
21  coordination of any transition should the Court be poised to
22  issue an order.
23          THE COURT: We understand the significance of the
24  collections and the significance of technological and other
25  issues with respect to any modification in what is currently

~~TOP SECRET//COMINT//ORCON, NOFORN~~

NYT v DOJ, 16 CIV 7020_000162

Names of certain hearing participants withheld under b(6) and b(7)(C). All withheld information exempt under b(1) and b(3) unless otherwise noted. Approved for public release. TOP SECRET//COMINT//ORCON, NOFORN 84 1 being done, and we'll try to be as open as reasonably possible. 2 MS. MONACO: We appreciate that. 3 [illegible] Your Honor, earlier in the hearing as well 4 you had asked where a particular part related to the unknowns -- 5 THE COURT: I'm sorry, what? 6 [illegible] I'm sorry. You had asked where some of 7 the representations we had made here today with respect to the 8 unknowns was in relation to the record indicated that some of 9 those points were captured in the 30 August submission. I would 10 point the Court to the final paragraph on page 6 of the 30 11 August submission there for a few of those representations just 12 for reference. 13 THE COURT: And certainly, I hope you don't think that 14 we didn't look at and take fully into account the August 30 15 submission, but I think you can also conclude that it wasn't 16 fully convincing, so anything more you can do to convince would 17 be appreciated. 18 All right. Thank you again for coming and putting up with 19 these long proceedings. I'll let you get on your way. I'll 20 look forward to receiving further information and to talking 21 with you along the way in resolving this very important, unique, 22 and in some ways difficult matter. I thank NSA and the 23 Department of Justice for all their efforts in this regard. 24 Thank you all. (Proceedings adjourned at 1:52 p.m.) 25

TOP SECRET//COMINT//ORCON, NOFORN NYT v DOJ, 16 CIV 7020_000163

NATIONAL
SECURITY
ARCHIVE

National Security Archive,
Suite 701, Gelman Library, The George Washington University,
2130 H Street, NW, Washington, D.C., 20037,
Phone: 202/994-7000, Fax: 202/994-7005, nsarchiv@gwu.edu