pageok
pageok
pageok
Data-Mining, FISA, and the NSA Surveillance Program:
I'm planning on spending the rest of today at the AALS Annual Conference across town, but I wanted to touch on a few more issues about the NSA surveillance program before I do:

  1. Based on what I have read from Risen's book, it seems less likely to me than it did before that this is a TIA-like data-mining program. It helps to note a distinction between two different methods that the press (and some commentators) often jumble together: packet-sniffing on a packet-switched network, and data mining. Packet sniffing refers to installing a monitoring device on a steam of traffic that looks for specific sequences of letters, numbers, or symbols. Here is how I explained packet sniffing for Internet traffic in my article, Internet Surveillance Law After the USA Patriot Act:
While the Internet uses packets to send and receive information, the packets are really just digital ones and zeroes that computers use to communicate with each other. The ones and zeroes can be reassembled into text to be read by a human, but computers do not need to do this and generally will not. A computer surveillance tool programmed to look for all emails to the Internet account "bob@aol.com" does not actually look for the text "bob@aol.com." To simplify a bit, the tool instead begins by looking for emails, and when it finds an email, it scans the right place in the email for the digital equivalent of "bob@aol.com," which is 0110001001101111011000100100000001100001. If this exact sequence of ones and zeros appears in the right place, the surveillance tool knows that it has found an email to bob@aol.com and will copy and record the block of ones and zeros that represent the email so that someone can later come back, convert the ones and zeros into text, and read the email. If the tool has an advanced filter and is configured properly, the billions of ones and zeros that do not relate to emails or to the exact sequence of 0s and 1s that represent the target account will pass through the device and be forgotten.
Based on what I have read from Risen's book, it sounds to me like that's what the NSA was doing. For those with criminal law experience, this was basically a large-scale pen regsister/trap-and-trace or wiretap, depending on how the filters are configured. (I'm not sure how different telephone traffic is these days, at least inside the provider switches.)

  This is different from a data-mining program. The term "data-mining" is usually used to mean taking an already-gathered database of information, and then performing analysis on the gathered database in lots of ways to identify patterns and characteristics. As best I can tell, the NSA program was not actually recording domestic Internet traffic, putting it in a database, and then "mining" it for key words and the like. Rather, this was a real-time surveillance program focusing on traffic associated with specific phone numbers and e-mail accounts. This is extra-tentative, of course; I'm basing this from snippets in Risen's book, and I'd be happy to change this analysis if we get new info. (Also, while it is true that Nancy Pelosi's letter expressed concern that the program was like TIA, keep in mind that she wrote that letter without any help from her staff; I don't think Pelosi has any background in this area, so I'm not sure her letter is particularly helpful evidence of the program at this stage.)

  2. I know it's going to annoy Armando, but I'm still not yet entirely sure of what to make of the legal issues. If I were confident that the DOJ letter represented a concession that the program violated FISA, I would be happy to bank on that and move on. As I have said before, I find the AUMF and Article II arguments unconvincing, so if that's the right issue to be focusing on, I'm with Armando. But something seems fishy here. For example, the leakers of the story seem focused on the Fourth Amendment instead of FISA. Further, given the extremely small number of people within the government who know the details of the program, it's not clear that DOJ's Office of Legislative Affairs (the office that sent the letter) was briefed on the details of the program. That is, the DOJ memo may have been written by people who knew less about the monitoring program than we now know thanks to Risen's book. (This may seem odd to you if you have never worked in the federal government; my guess is that it will seem less odd to those who have.) So Armando may be right, but I don't think we know enough to be sure of that.

  3. Finally, and relatedly, the details of the program from Risen's book arguably explains the national security interest in keeping the domestic surveillance program a secret. It's not that terrorists may suddenly realize that they may be monitored; that argument never made much sense, as every member of Al-Qaeda must know that they may be monitored. Rather, I suspect the security issue is twofold. In the short term, terrorist groups now know that they can stand a significantly better chance of hiding their communications from the NSA by chosing communications systems that don't happen to route through the U.S. And in the long term, some countries may react to the disclosures of the program by redesigning their telecommunications networks so less traffic goes through the United States. The more people abroad know that the NSA can easily watch their communications routed through the U.S., the less people will be willing to route their communications through the U.S. Cf. Bruce Hayden's comment. No doubt it was a long-term priority of the NSA to ensure that lots of international communications traffic was routed through the U.S., where the NSA could have much better access to it. Indeed, Risen's book more or less says this. The disclosure of the program presumably helps frustrate that objective.

  Anyway, that's it for now. My apologies to readers who want me to have a much more certain answer. My Internet surveillance antennae aren't yet giving me clear enough clues to know for sure where things are going to land.
Joel B. (mail):
I'm all for freedom of the press, but I'm becoming more convinced that there was treason here, which ought to be severely punished.
1.4.2006 1:54pm
AF:
"I'm still not yet entirely sure of what to make of the legal issues."

Well, that's your story and you're sticking to it, I guess. You still haven't explained what national security risk would be created if the Administration simply refused to confirm or deny whether the program was covered by FISA.
1.4.2006 2:06pm
Medis:
Orin,

Of course, one possible explanation of why internal objections might have focused more on the Fourth Amendment, rather than the statutory issues, is that those people might actually have bought the statutory arguments, or at least thought there was no audience for such objections.
1.4.2006 2:11pm
von (mail) (www):
Orin:

Perhaps I've missed something or perhaps I'm just a bit dense, but the following seems to suggest that the real damage to our national security occured with the publication of Risen's book, and not with the New York Times' story (which seems to have omitted the key details). Before Risen's book, it seems that even a learned observer (you) couldn't determine why knowledge of this formerly secret program could inure to the benefit of terrorists.

Finally, and relatedly, the details of the program from Risen's book arguably explains the national security interest in keeping the domestic surveillance program a secret. It's not that terrorists may suddenly realize that they may be monitored; that argument never made much sense, as every member of Al-Qaeda must know that they may be monitored. Rather, I suspect the security issue is twofold. In the short term, terrorist groups now know that they can stand a significantly better chance of hiding their communications from the NSA by chosing communications systems that don't happen to route through the U.S. And in the long term, some countries may react to the disclosures of the program by redesigning their telecommunications networks so less traffic goes through the United States. The more people abroad know that the NSA can easily watch their communications routed through the U.S., the less people will be willing to route their communications through the U.S. Cf. Bruce Hayden's comment. No doubt it was a long-term priority of the NSA to ensure that lots of international communications traffic was routed through the U.S., where the NSA could have much better access to it. Indeed, Risen's book more or less says this. The disclosure of the program presumably helps frustrate that objective.


(Far from me to defend the Times, but perhaps they actually did hold back key aspects of the story to protect national security concerns.)
1.4.2006 3:07pm
Dilan Esper (mail) (www):
One practical question: What is the safeguard that exists to ensure that the NSA is not also reviewing or scanning electronic communications among American citizens or residents that happens to travel (by means of VOIP, e-mail, IRC, or otherwise) outside the US and then back? That, after all, also happens, and I see nothing in Risen's accounts or any of the defenses of the program made by the White House that indicates that there is any such safeguard.

If the White House is asserting the power to eavesdrop on communications between Americans without a warrant, simply because the telecommunications provider routes the communication in a particular manner, that is something that the American people are entitled to know.
1.4.2006 3:23pm
Anderson (mail) (www):
In the short term, terrorist groups now know that they can stand a significantly better chance of hiding their communications from the NSA by chosing communications systems that don't happen to route through the U.S.

Wouldn't a RPT (Reasonably Prudent Terrorist) already avoid systems that don't route through the U.S., on general principle?

In any event, I'm guessing they either eschew all such communications in favor of couriers, or else count on sheer volume to escape notice.

Also endorse what Von said---the book spills more beans than the Times story, it appears.
1.4.2006 3:41pm
Spartikus:
Hmm...don't communications into and out of the U.S. have to route through the U.S.?

And is it not true that foreign to foreign communication, even if it routes through the U.S., are not covered by FISA and the NSA is free to eavesdrop?
1.4.2006 3:46pm
Bob Loblaw (www):
I

n the short term, terrorist groups now know that they can stand a significantly better chance of hiding their communications from the NSA by chosing communications systems that don't happen to route through the U.S.
I don't get this one (although maybe I'm missing something). Aren't the terrorists' communications that don't pass through the US also being monitored, and don't they know that already?
1.4.2006 3:51pm
Bob Loblaw (www):
Oops, Anderson and Spartikus beat me to it.
1.4.2006 3:52pm
Donald B. Le Messurier (mail):
No doubt it was a long-term priority of the NSA to ensure that lots of international communications traffic was routed through the U.S., where the NSA could have much better access to it.

I haven't seen this addressed, but from my limited knowledge of NSA intercepts most of it is done in Europe. I will not mention any specific country(s) but my understanding is that they intecept the signals going to and from communication satellites. One can see the domes from the roadway in some places. They look like radar domes. That the NSA would encourage and receive the cooperation of allies to route their calls through US communication satellites would seem to me to be an obvious thing to do. But does routing these calls to US controlled satellites mean that the call is going "through" the US? Just because it's in synchronous orbit 25,000 miles up makes calling it "through" a little tough to swallow. Would this scenario cause any adjustments in the arguments being discussed?

Le Messurier
1.4.2006 4:00pm
Frank Drackmann (mail):
I wonder if the Maryland State Trooper who let Mohammed Atta go is still patrolling the highways?
1.4.2006 4:01pm
John Lederer (mail):
Much easier than rerouting is encrypting. Clearly they did use plain language post 9/11 and pre NYT.
1.4.2006 4:01pm
Armando (mail):
Professor:

I am sorry my post seemed to show annoyance with you. Actually, I was more annoyed with Risen's book, which seemed to make some fundamental errors.

Now, as for the Justice letter, I assume Gonzales signed off on it. Gonzales, as George Conway noted, said what he said.

For the time being, it seems more than fair to me to assume the Attorney General knows of what he speaks.
1.4.2006 5:51pm
Benzai (mail):
To second Spartikus: Why would the NSA want to route communication to within the U.S., subjecting itself to all sorts of procedural and bureaucratic red tape, when it is free to wiretap/penregister away in foreign jurisdictions (in the same way this Administration has outsourced torture to Gitmo and other locales)?
1.4.2006 6:27pm
Thrax (mail):
More fundamentally, how has the disclosure of the NSA's wiretapping *without a warrant* changed anything from the perspective of an enemy of the U.S.? Said enemy was already in a position to know that calls to the U.S. were being overheard. Said enemy might have thought until now that the NSA, within 72 hours of tapping that call, went to the FISA court and got the wiretap rubber-stamped. Said enemy now knows better. How would this revelation change said enemy's behavior? "Sorry, Ahmad, I can't talk freely. If the NSA had to get a retroactive-justification warrant after overhearing this call, then I'd tell you everything, but as it stands I'd better clam up."
1.4.2006 6:46pm
Karl (mail):

But something seems fishy here. For example, the leakers of the story seem focused on the Fourth Amendment instead of FISA.


And if we make the nutty assumption that the leakers are more likely to know the relevant facts and circumstances than we are, what should we make of the discussion going on here and at other sites concluding FISA was probably violated?


This is extra-tentative, of course; I'm basing this from snippets in Risen's book, and I'd be happy to change this analysis if we get new info.


That should sum it up.
1.4.2006 6:47pm
agesilaus:
I don't understand the idea that the comms have to be routed thru the US at some point. The NSA has a whole array of different SigInt satellites that grab signals every time they pass between microwave towers. The signal beam is aimed at the next tower but most of the beam is not captured by the next tower antenna and is radiated into space. The SigInt satellites capture that data from anywhere in the world.

Also a lot of phone calls are sent up to geosynchronous commercial satellites and beamed back down to a ground station. You can bet that that data is sampled by the NSA. Look at the ECHELON project.

Perhaps this flap is about, what I would guess is a minor fraction, of comms that still go via trans-Atlantic cables. But the NSA has been in this business for decades and is reportedly intercepting 3-4 billion comms a day (a wild guess of course since they aren’t saying).

Most of this was covered in Bamford’s book, The Puzzle Palace which came out years ago. None of this is news in any way so far as I can tell.
1.4.2006 6:53pm
Defending the Indefensible:
Prof. Kerr:

A computer surveillance tool programmed to look for all emails to the Internet account "bob@aol.com" does not actually look for the text "bob@aol.com." To simplify a bit, the tool instead begins by looking for emails, and when it finds an email, it scans the right place in the email for the digital equivalent of "bob@aol.com," which is 0110001001101111011000100100000001100001.

No, really, a surveillance tool just looks for text. We don't program punchcodes and binary these days, programming is done in plain text including search algorithms, though I could get into a little bit more depth and talk about regular expressions (or regexes). A regex is basically just a pattern to be matched, like when you use a command-line on DOS or Windows to find all files starting with BOB and ending with .TXT by typing "DIR BOB*.TXT" -- the "*" substitutes for 0 or more missing characters. Anyhow, still text, not 1001011011...
1.4.2006 7:09pm
Defending the Indefensible:
Prof. Kerr:

As I have said before, I find the AUMF and Article II arguments unconvincing, so if that's the right issue to be focusing on, I'm with Armando. But something seems fishy here. For example, the leakers of the story seem focused on the Fourth Amendment instead of FISA.

Perhaps simply because the leakers aren't attorneys and don't necessarily know what statutes and caselaw might apply, but have a passing familiarity with the Bill of Rights.
1.4.2006 7:11pm
Defending the Indefensible:
Following up on my last comment, both the NY Times article and the Risen book were written before the administration offered a legal defense relying upon the AUMF and Article II. There's no reason these arguments should have been anticipated, so the absence of these justifications is not particularly notable.
1.4.2006 7:43pm
boris (mail):
Wanting to compare NSA surveillance on terrorists to FBI surveillance on organized crime I google-found this in an article from 1999 ...
For example, the FBI would be able to listen in on the cell and ordinary telephone conversations of all parties to a conference call, even if some are put on hold and are no longer talking to the target of the legal wiretap.
So I'm guessing that in regard to criminal investigation a legal wiretap on one person (eg crime boss) does allow monitoring of calls between him and other parties (for whom there is no warrant). If it were proposed that FISA type rules should apply to criminal investigation ...

(1) No monitoring without warrants for all other parties

(2) 72 hour retro warrants allowed "in emergency"

(3) Probable cause for retro warrant needs to be independent of monitored content

I claim that criminal investigators and prosecutors would scream their heads off that FISA style restrictions applied to them would be completely unworkable. If FISA is unworkable then statutory exceptions will be liberally interpreted to apply, or AUMF will override, or Article II will carry the day. That's how it works, constitutional purity after CFR is sophistry.
1.4.2006 7:43pm
subpatre (mail):
DtI wrote "No, really, a surveillance tool just looks for text."

No it doesn't. A) The programming may be text, but the compiler changes the program to binary. B) The surveillance tool --whether a computer or dedicated box-- is looking at a binary data stream; ones and zeros, not text.


DtI tries again: "Perhaps simply because the leakers aren't attorneys and don't necessarily know what statutes and caselaw might apply, but have a passing familiarity with the Bill of Rights."

Perhaps carpenters are unfamiliar with nails. Privacy law is the backbone of NSA internal policy, training, and procedures.

Many people here have floundered around trying to interpret FISA, yourself included.

"This USSID prescribes policies and procedures and assigns responsibilities to ensure that the missions and functions of the United States SIGINT System (USSS) are conducted in a manner that safeguards the constitutional rights of U.S. persons. ... It constitutes a summary of the laws and regulations directly affecting USSS operations. All USSS personnel who collect, process, retain, or disseminate information to, from, or about U.S. persons or persons in the United States must be familiar with its contents."

USSID 18 then goes on and expands on FISA and EO 12333; and USSID 18 Guide, and other similar documents are also required of NSA FI employees.

It's a ludicrous to suggest the NY Times' informants --people who know Chapter 36 and 119 inside and out, and also know the technical details that you don't-- aren't concerned about what you think's important just because of their ignorance.
1.4.2006 8:04pm
Noah Klein (mail):
Professor Kerr,

I know that you will not receive this until your conference is done, but I feel compelled to dispute something you said when you began this thread. In your second point you said:

Further, given the extremely small number of people within the government who know the details of the program, it's not clear that DOJ's Office of Legislative Affairs (the office that sent the letter) was briefed on the details of the program. That is, the DOJ memo may have been written by people who knew less about the monitoring program than we now know thanks to Risen's book.

I am not suprised that the Office of Legislative Affairs would not have operational or classified knowledge of this program. Knowledge of this program has nothing to do with their job description.

What I would be surpised by is the Attorney General not having operational or classified knowledge of this program. It is the duty of the Attorney General to approve of all grants of electronic surveilance by the NSA according to both the president and his executive order. Furthermore, the president said Gonzales approved of the legality of this program when he was White House Counsel and AG. So when he said in his press conference that it would have violated FISA, if not for the 2001 AUMF, I do not see why he would not have the operational and legal understanding to make that claim.

Noah
1.4.2006 8:04pm
Defending the Indefensible:
Even at a "binary" level, modern computers do not truly operate on 1's and 0's anymore. Chips have instruction sets, codes such as "REP SCASB" which match through a buffer bytewise.

Subpatre really doesn't know what he's talking about.

Also, I don't doubt that one or more of Risen's sources may be attorneys, but he is presumably not and neither are his readers. The VC probably has a better handle on the legal aspects of this issue than the NY Times writers and editors.
1.4.2006 8:14pm
Justin (mail):
Boris, the only reason the government doesn't need a warrant is because the 4th amendment doesn't extend to foreigners overseas. If they get a warrant from anyone (including a foreigner overseas) based on probable cause, then any incidental communications they receive lose the REP. But that's not what the government did here, probably because they lacked even probable cause-FISA style on the "foreign" presence, much less the "domestic" one.
1.4.2006 8:17pm
Diversity Hire:
Why assume this is not a data-mining operation just because the "contents" of sniffed transmissions aren't stored?

Packet-sniffing is just the means of collecting data from a digital stream (and it's easy and fun to try at home, at work, or at your local Starbucks; google for "Ethereal GUI").

Sniffing is useless in itself. But the header information and meta-data about the capture, can be aggregated and analyzed to produce invaluable information, even without ever reading the "contents" of the packet. One could infer social-networks from transmission patterns, for instance; or track the movements of a cell-phone user. In some cases the meta-data is more valuable than the contents, e.g., when the contents are indecipherable because of encryption or language barrier.

So, although the distinction betweeen the means of acquisition (packet sniffing with pattern-based filtering) and the use of the data thus acquired and produced is important, it doesn't seem that "detecting" traffic is useful unless it is coupled with a larger effort to analyze that data. If the NSA is empowered to packet-sniff and isn't doing traffic-analysis (i.e., "mining" the resultant data), then they're not doing their job.

BTW, the notion that there is some difference between the "header" and the "content" isn't at all clear to me; it's not like a telephone where the switching info and the voice traffic are easily distinguishable. Instead there are a bunch of symbols (signals on a wire, pulses of light, one's and zeros, 0-Fs, bytes, characters—or however else you want to interpret them) bundled together; the separation of header and content is by convention and is maleable: headers change; packets are tunnelled inside of other packets; payloads are encrypted, fragmented, etc. Protocols are layered upon one another and the contents of the packet reflect that layering: the payload of one layer might begin with the header of the next.

(Incidently, I believe >85% of international communications travel over subsea fiber optic cables and that that percentage has been increasing since 1988 or so. These cables are as secure as their owner, operators, and governments will let them be; which I assume isn't very.)
1.4.2006 8:34pm
Defending the Indefensible:
DH:

There isn't really much difference between header and content, both are part of a single block of text in a standard e-mail message, separated by a blank line.
1.4.2006 8:40pm
Charlie (Colorado) (mail):
Even at a "binary" level, modern computers do not truly operate on 1's and 0's anymore. Chips have instruction sets, codes such as "REP SCASB" which match through a buffer bytewise.

You know, this would be funny, except I'm afraid that DtI may actually have a technical degree.

No, DtI, computers still do work in binary. No, having multiple bits in the instruction set doesn't change that. Yes, you can do "wildcard" matches. No, whether you consider them as binary versus some other coding scheme is not germane to the basic issue.

Yes, you've made a complete fool of yourself. But we'll let you off the hook this time.
1.4.2006 8:42pm
Charlie (Colorado) (mail):
In any event, I'm guessing they either eschew all such communications in favor of couriers, or else count on sheer volume to escape notice.

Anderson, sadly I think part of what this may have done is clarify that sheer volume isn't sufficient protection.
1.4.2006 8:44pm
Charlie (Colorado) (mail):
Why assume this is not a data-mining operation just because the "contents" of sniffed transmissions aren't stored?

For the same reason 2 isn't an odd number: because that's the definition.
1.4.2006 8:46pm
John Lederer (mail):
agesilaus:

I have a 2004 FCC report that shows available international circuits by submarine cable and satellite in a graph. There are about 2,500,000 64 Kbps equivalent circuits as of 2002. The graph is too small to tell precisely, but I would say 2,400,000+ are submarine cable and 100,000- are satellite.

Cable dwarfs satellite, so cable is the game for spying.
1.4.2006 8:47pm
Diversity Hire:
DtI, I agree: at some level there is no difference between headers and contents, they're "just bits" and "bits are bits". But there are arbitrary lines drawn between them by both protocol developers and attorneys (acting independently, of course); aparently, it is legal to intercept header information but not content in the United States, without a warrant.

I didn't mean to speak about email headers, I meant headers more generally. One protocol layer's header starts the next layer's header; your 802.3 packet has a header and a payload, the payload may be an IP packet with a header and a payload, the IP payload might be a TCP packet (segment) with a header and a payload, and so on, like Russian dolls. Where do the headers stop and the content begin?
1.4.2006 8:53pm
OrinKerr:
Noah,

The AG didn't write the letter.

Diversity Hire,

I have a long dicsussion of the content/envelope distinction in the article linked to in the main post.
1.4.2006 8:53pm
Mark Buehner (mail):
It's a ludicrous to suggest the NY Times' informants --people who know Chapter 36 and 119 inside and out, and also know the technical details that you don't-- aren't concerned about what you think's important just because of their ignorance.

I havent read Risen's book yet, but is there any proof the leak came from the NSA, as opposed to the WH, DOJ, FISA, or (most likely imo) Congress? Or a variety of leaks from a combination of sources? I think you are right about the NSA in general, if someone there leaked they knew exactly what they were doing and (assumedly) why. OTH, a Congressional aide would look at this thing and have a conniption fit at face value (as plenty of others have since), and might leak not knowing exactly what was wrong with the program but being sure it was illegal somehow. Does Risen point to any inside information from the NSA that gives it away?
1.4.2006 8:57pm
David Resnick:
Defending the Indefensible:

Subpatre is correct. I am not a lawyer, but I am a programmer. Computers execute machine language, which
is binary. Assembly language (e.g. REP SCASB) is converted
to that for execution...

1.4.2006 8:59pm
Medis:
Orin,

Noah was referring to AG's statement in his press briefing, not the Moschella Letter.
1.4.2006 9:13pm
subpatre (mail):
Mark Buehner wrote "I havent read Risen's book yet, but is there any proof the leak came from the NSA, as opposed to the WH, DOJ, FISA, or (most likely imo) Congress? Or a variety of leaks from a combination of sources?"

The Times draws from a variety of sources. Of course there's no "proof" of anything at this point, but "beyond reasonable doubt" may describe it. According to their article, the Times used WH, DOJ, NSA, and Congressional officials among many others.

...some officials familiar with the continuing operation...
...a former senior official who specializes in national security law.
Nearly a dozen current and former officials...
...program remain secret, officials familiar with it say.
...according to the officials familiar with the N.S.A. operation.
...according to several officials who know of the operation.
...officials familiar with the program said.
...officials knowledgeable about the [Rockefeller] letter said.
One government official involved in the operation said...
1.4.2006 9:36pm
subpatre (mail):
While searching on the word 'official' in the original NY Times article for the reply to Mark Buehner (above) several intriguing passages jumped out. Here's the first:

"Aside from the Congressional leaders, only a small group of people, including several cabinet members and officials at the N.S.A., the C.I.A. and the Justice Department, know of the program."
Intriguing indeed. Only a small group knew of the program, yet the NY Times got a more than a couple to leak to the paper. Is this unrelated to NSA; but part of the ongoing CIA v. Administration conflict?
1.4.2006 9:52pm
Bruce Hayden (mail) (www):
For the life of me, I can't find a law review article I read in the last week on ECHELON that I think was posted here originally. In any case, everyone on this thread seems to have gotten the technologies confused a bit.

We are really talking voice, and not Internet / email right now. And, as Orin does point out in his article, the addressing is different for them. Voice (excluding VOIP of course) is still circuit switched (see below), while the Internet is packet switched. Big difference. It is quite hard to effectively tap the Internet except near one end or another of a session (and please don't get into the fact that much of Internet traffic is really datagram and not session based). The problem is that data blocks are chopped up at one end into packets and reassembled at the other end. And the packets don't have to, and often don't follow the same route. And without all the packets making up a block, it is most often impossible to decode the message. Thus, in Orin's bob@aol.com example, "bob@a" might be in one packet, and "ol.com" in another, and the first packet might go from Wash. D.C. (Orin) to L.A. (Eugene) via Dallas and Phoenix, while the second packet goes there via Denver and Salt Lake. And, if you aren't trying to capture this in either Wash. D.C. or L.A., then you are unlikely to get the whole message. BTW, this is part of the story behind the FBI's abortative CARNIVORE project of a couple years ago. Oh, and unless you stand on your head (by explicitly specifying all routing, which is extremely rare), you have no control over any of this.

In any case, voice is different. Instead of being dynamically routed like the Internet, it is statically routed. A circuit is put up at the start of a call, and resources reserved throughout the extent of the call. If a call from Orin to Eugene starts via Dallas and Phoenix, that is the way it will continue throughout the call, at which time the resources are released. If Dallas is nuked during that time, the call is lost (whereas the Internet will reroute automatically through, for example, Denver and Salt Lake). Traditionally, this static routing meant reserving trunks all along the route for the duration of the call. More recently, many longer calls are now digitized for transmission over, for example, ATM. Nevertheless, the reservation paradigm still rules.

Orin points out in his article that the destination telephone number is required for voice call setup and routing (technically, a bit more than a telephone number - see E. 164 as an example). My memory is that source telephone number and other billing information is also typically transmitted along during call setup for accounting reasons. Thus, at each node through which a call is routed, this information is most likely available, and is part of what is presumably utilized by the NSA program.
1.4.2006 9:54pm
Bruce Hayden (mail) (www):
Let me also point out that the present NSA kerfuffle is primarily about optical (and to some much lesser extent, copper), and not radio, microwave, sattelite, etc., transmissions. The NSA has had legal methods of tapping those for years. I was surprised to find out in that ECHELON article that they use satelites to capture surface microwave transmissions. In any case, the fact that those all transmit through the ether makes tapping them legally much more straight forward.
1.4.2006 10:06pm
Bruce Hayden (mail) (www):
Donald B. Le Messurier:

I would think that any call that goes through satellites in geosynchronous orbits would not be considered going through the United States for the simple reason that there are no satellites in geosynchronous orbit over the United States. Why? Because they are all over the Equator. By necessity, they have to be.

Note though that not all telephone traffic that goes by satellite goes by satellites in geo-synchronous orbit. One notable exception is Iridium.

But as I keep pointing out, this is most likely irrelevant to the present discussion. The U.S. has been legally tapping satellite calls for a long time.
1.4.2006 10:20pm
subpatre (mail):
Here's another weird fact, from the original NY Times article, that's avoided any discussion on this blog.

The legal opinions that support the N.S.A. operation remain classified, but they appear to have followed private discussions among senior administration lawyers and other officials about the need to pursue aggressive strategies that once may have been seen as crossing a legal line, according to senior officials who participated in the discussions.


If the legal underpinning was classified, is it still? Unless declassified, the AG can't release it, reveal the contents or the substance of the opinion. Meaning Gonzales couldn't describe the legal basis for the program, and was misleading the public in his press appearance.

Some people are scrutinizing §180x with a fine-toothed comb; yet are adamant the AG's press conference must be interpreted "in context". I say that isn't so.

If legal justification for a program is classified, the public has no expectation of the direct truth from Gonzales on that particular issue. In two previous comments --here and here-- the Attorney General's press speech and comments are examined from a defensive standpoint.

The summary is that Gonzales was faced with the practical imperative for a press conference and the classified nature of the subject. He formatted his speech so he appeared to talk about "the program", but in fact he talked about several different subjects.

Many people believe Gonzales 'confessed to FISA violation' in that speech. I'm not sure he could be convicted if he later claims he was describing several different programs.

Be that as may, the legal underpinning was classified, and may still be.
1.4.2006 10:24pm
Medis:
subpatre,

The fact that a document is classified doesn't mean all of its contents are classified.
1.4.2006 10:28pm
Just an Observer:
subparte, et al:

Assuming this mess ever got ito court and was litigated up to SCOTUS, can anyone seriously believe that the court would permit the major part of the government's case to be redacted?

I envision the oral argument:

Solicitor General Clement: "Our legal reasoning is impeccable, Mr. Chief Justice. But if I told you what it is, I'd have to kill you."

Chief Justice Roberts: "That sounds more than reasonable, general."

In their dreams.
1.4.2006 10:41pm
Medis:
JaO,

And one obvious reason the legal opinions are still classified is that they are undoubtedly full of details about the program. So I doubt the government would ever claim in court the need to make secret legal arguments--but they might claim the need to keep operational details secret.
1.4.2006 10:57pm
subpatre (mail):
Medis - I know you're one who's scrutinized the code, disregard the President's statement, yet (forcefully) insist the AG's speech be taken in context as 'a confession'.

It's logically incoherent; but it's a free country and you can do that.

Facts are sparse. The few people who have talked about the program --some of them intelligence law experts, almost all of them knowlegable of FISA-- don't allege FISA violations; even fence issues. They express concern over 4th Amendment infringement.

Of all the people who are in-the-know, whether defending the special collections program or leaking it because they object, none of them say "This violates FISA". Only us bloggers with no clue of the details are saying that.

Medis asked earlier why AG Gonzales would conceal the truth about legal justifications for the program. The NY Times says the legal justification's classified.


Just an Observer - Huh? I hope you weren't addressing that to me, as I indicated nothing of the sort.
1.4.2006 11:17pm
Medis:
subpatre,

Out of curiousity, what do you mean by "disregard the President's statement"? To which statement are you referring?

Anyway, I find it interesting that when I take the AG as telling the truth about their legal claims, you claim I am taking the AG as making a "confession". I guess you really don't believe their 2001 AUMF argument if you think the AG was making a "confession".
1.4.2006 11:21pm
Just an Observer:
subparte: "Just an Observer - Huh? I hope you weren't addressing that to me, as I indicated nothing of the sort."

I was addressing it to anyone who suggests that the administration has a stronger outline of legal reasoning than they have stated publicly. It is one thing to withhold operational details supporting a particular line of argument. It is another to suppress the whole line of argument.

As far as the facts go, there is much they haven't told us.

As far as the law goes, the only reasonable reading of events is that what we see is what there is.

I don't completely discount the possibility that the tooth fairy exists, but I generally try order my life around the opposite proposition.
1.4.2006 11:43pm
Medis:
subpatre,

I just read back, and I wanted to clarify one point. I hope you are aware that when the word "opinion" is used in a legal context, that usually means an actual document, not some viewpoint. So to say that some "legal opinion" is classified would usually mean to say that some particular document is classified, not some viewpoint.
1.4.2006 11:55pm
subpatre (mail):
Medis - Thanks, I'm fully aware that a particular document that gives the legal reasoning for this program is classified. Until it's declassified, revealing the contents or substance, in whole or in part of the declassified portions is illegal.

On the same subject and going a bit further, the article states:
"The legal opinions that support...", not
"The legal opinion that supports..."

It may just be a typo, but it indicates there could be more than one approach to why the President believes the program's legally sound.
1.5.2006 12:23am
Defending the Indefensible:
David Rasnick,

It's a distinction without a difference, if you reduce to the transistor level then yes, computers are binary, but even within the physical hardware of modern CPUs, instructions are converted to "macro-ops" and a whole lot of abstraction occurs before you get to the level of bitwise operation.

But a similar argument could be made by reducing the function of mental processes to the neuronal level, if you hold a mechanistic view of such things. The point is that this is not the level at which people operate, it is not the level at which algorithms are created and software written.

It is more correct to speak of patterns being matched bytewise than bitwise, anyhow, because modern processors cannot access less than a byte of memory nor possess registers of less than a byte in size. It's inefficient.

Bitwise reads can still be performed, by loading in a data word, and using a special bitmask with an OR instruction, but I guarantee you that this is not how text scans are performed.
1.5.2006 12:50am
Defending the Indefensible:
Sorry for spelling your last name wrong David.
1.5.2006 12:51am
Smithy (mail):
Data mining is a crucial tool in our war against terror. And it does not violate the Fourth Amendment.

Quite simply: I don't see why anyone would care about NSA surveillance of their communication unless they were in league with the terrorists themselves. If the NSA read all of my email, all they would learn is that I love American and hate terrorists. I have a strong distrust of those who are concerned about data mining.
1.5.2006 1:09am
Noah Klein (mail):
Professor Kerr:

You are right he did not write the letter, but it must have approved of it.

2) and most important he made the same case as the letter in his press briefing. This would indicate that the letter and the briefing were the DOJ's and the administration's opinion that the program violated FISA, but was legal because of the 2001 AUMF.

Noah
1.5.2006 2:30am
Noah Klein (mail):
Subparte:

Your close reading of the article exposes an interesting fact and I now do not find it so incredible that the AG would mislead the public. Yet, as I pointed out in other threads, a lie or dissembilization (not sure if that is even a word) of the legality of the program, may expose the AG to a credibility problem with the Congress, which would hurt the administration over the next three years.

It is important to note that the article came before the confirmation and thus declassification of elements of the program. It is still most likely that the president declassified the legal opinion behind the authorization for the program. This would allow the AG to inform Congress and the public of the administration's legal justification of the program and thus he was telling the truth in the press briefing.

Noah
1.5.2006 2:44am
Bob Loblaw (www):

there are no satellites in geosynchronous orbit over the United States. Why? Because they are all over the Equator. By necessity, they have to be.
Actually, only geostationary orbits need to be over the equator. Some geosynchronous satellites may spend some portion of their orbits over the United States (depending upon the angle of inclination of their orbits.
1.5.2006 2:45am
K Parker (mail):
Indefensible and subpatre,

You're both off the mark. In reality 'bob@aol.com' and '0110001001101111011000100100000001100001' are just two different (but equivalent) visual representations of the same sequence of bits.

Bruce Hayden,

please don't get into the fact that much of Internet traffic is really datagram


Will you permit me to go as far as to say, "Oh really?" Other than DNS (which does not appear to be of any interest here) the things we are talking about here--primarily email--are exclusively TCP-based, are they not?
1.5.2006 2:51am
Defending the Indefensible:
K Parker,

Yes, the same sequence of bits, and the same sequence of bytes, same thing, that's largely my point. In reality what we see on our respective monitors here is not "written" communication at all, it is just bits and bytes, but to try to draw a legal distinction from pen marks on paper is ridiculous as the fundamental construct is just a medium of communication.

Not that posts to a publicly accessible blog like the VC are privileged communications, but that's a matter of venue, not mechanism, as if we were writing letters to the editor of a publication (which, in a real sense, we are, but with a more automated system).
1.5.2006 3:00am
Defending the Indefensible:
Smithy amuses me.

Quite simply: I don't see why anyone would care about NSA surveillance of their communication unless they were in league with the terrorists themselves. If the NSA read all of my email, all they would learn is that I love American and hate terrorists. I have a strong distrust of those who are concerned about data mining.

All Smithy ever talks about in e-mail is that he loves America and hates terrorists. He doesn't have any other conversations about any other subjects, ever.
1.5.2006 3:08am
Defending the Indefensible:
K Parker,

Well, TCP layers on top of IP, as does UDP, and IP is essentially just datagrams (or packets, the terms are really synonymous).
1.5.2006 3:11am
Bob Loblaw (www):
DtI - He'd be amusing if it weren't so scary that that's actually what he thinks
1.5.2006 3:21am
18 USC 1030 (mail):
First of all, I am a bit confused as to the applicability to the binary representation discussion as well as the format of packets in a packet switched network. Unless it is being argued that the communications intercepted were traveling via the internet, such arguments are moot. I have yet to hear that the NSA intercepts were anything but voice communications which I presume, perhaps incorrectly that these are communications traveling over a static routed system. However, if this is proven untrue, much of the debate thus far on the basis of technology is incorrect. Perhaps it is merely due to simplification, but much of the representations and explanations are not entirely correct. Though Prof. Kerr, you are correct in your binary representation, practically speaking, as I believe has already been mentioned, that string would most likely be broken up over multiple packets. However DtI, I am not sure what you are talking about, but computers cannot understand anything but for ones and zeros. It has been suggested this is the case only at the transistor level; however, that is the level in which calculations take place. Our being unable to program specific transistors does not remove the fact that the computer translates everything to binary prior to computing. This discussion however, is a bit misleading. We are not talking about the programming side of this; but rather, the transmission of data across some type of network. If one accepts the argument that computers understand a level higher than binary code; such a belief does not conflict with the argument that the data being transferred is transferred as ones and zeros. If one were to look at a packet sniffer, generally two representations are provided hex and ASCII. The packet sniffer is translating the binary data intercepted into text or hex. It could easily display the binary, without any alteration; however, I don't know many people who can understand binary to a degree that such a representation would be prudent. However, one cannot send anything but binary equivalents in a transmission.

I would suggest anyone confused about the difference between envelope data and contents to read Prof. Kerr's article. In short, envelope data is that which you would ordinarily share if sending a letter through the mail (i.e. To name, From name, To address, from address, postal routing number). Thus, for an email, the email addresses and IP addresses of the mail servers as well as time/date stamps and rout between senders mail server and receivers mail server are all included in envelope data. The two things that are not included are the subject line and the body of the email. This data can be separated from the other data, first of all because of the method of transmission, the packets of the header would be sent first. In order to send the email, first, sender email server would send header packets in order to ensure the receiver exists. Also, though most email clients do not allow the separation, if one were to use, even a traditional packet sniffer such as ethereal, one could see the header packets along with the contents. There are two types of filters one can utilize, one filters the data prior to it being saved, as in it looks for particular packets and saves them if found. The other is applied to collected data, so long as the intercept is operating with a filter prior to saving packets in order to prevent investigators from seeing the contents, the header information is separate. The only remaining question is whether or not that tap and trace would be legal in the first place.

Once again, I am a bit perplexed as to why the computer communications are of concern, unless I have missed something.
1.5.2006 3:55am
Bruce Hayden (mail) (www):
Defending,

Not quite sure what you are considering macro-ops. Maybe some of the microcoded instructions in CISC architectures? The trend seems to be going in the opposite direction in the RISC architectures - what you see is what you get. I just reviewed some IA-64 documentation again, and the trend there is to move what used to be microcoded CISC into compilers and let them optimize (which is really necessary in IA-64 and IA-128 given their parallel execution instruction packing).

Still, your basic point is valid - that comparisons are not done at the bit level, but at a higher level, in IA-64 selectively at the 8, 16, 32, or 64 bit level. And the hardware itself is now actually doing these comparisons in parallel. Yes, that takes a lot of gates, but there are millions of transistors on a processor chip these days, so that is no longer a concern, and comparing two bits only takes a handful of them.

My guess is that either eight or sixteen bit (parallel) comparisons are utilized when comparing addresses like bob@aol.com or phone numbers, depending on whether an 8 bit or a 16 bit character encoding is being utilized, as a result of C type string comparisons most likely being used.
1.5.2006 4:01am
Bruce Hayden (mail) (www):
18 USC 1030

I think I made the same point above. The illustrative use of email over presumably TCP/IP just confuses things. The debate here is not about packet switched TCP/IP, but rather circuit switched voice, which is typically routed utilizing extended telephone numbers.

It confuses things for a number of reasons. Earlier I pointed out the differences in trying to intercept the two - circuit switched voice is really the more tractable problem, which is probably why it is what is at issue here.

Another reason is that, except at the local loop (i.e. DTMF), with circuit switched voice calls, the routing information is really somewhat separate from the actual transmissions. That is partially because the entire circuit (almost always consisting of multiple hops) has to be reserved during call setup before any voice data can be transmitted. So, there is no real searching through messages for "bob@aol.com". Rather, the nodes typically know the start and end addresses (i.e. telephone numbers) of each call, based on the call setup information they received.
1.5.2006 4:21am
Bruce Hayden (mail) (www):
Let me add to that last post. Because the routing information has to preceed the actual voice transmissions, by necessity, given the requirement in circuit switching for reserving the facilities required during setup, merely monitoring transmissions is worthless, since once setup, a specific logical circuit need not (and does not) carry any identification of where it is going or where it came from, since it always goes the same place, until reallocated to another call. Thus, for example, ATM slot#4 will be allocated to the same call until complete. Oh, and a lot of set up is done out-of-band (i.e. on a separate command channel or slot).

So, in order to intercept circuit switched voice, you either have to monitor an entire bundle of circuits over a period of time, saving off the routing information whenever a call is set up, or go to the nodes where the routing information is maintained. Needless to say, the later is probably a lot easier, though the NSA is obviously capable of the former, and apparently has a dedicated submarine for just that sort of thing.
1.5.2006 4:35am
Medis:
subpatre,

But if you understand that we are talking about documents, not viewpoints, then you should understand that more than one legal opinion does not have to mean there is more than one set of legal arguments. Instead, it could be the same set of arguments provided by different people, applied to different facts, and so on.

In any event, as I noted before, the fact that a document is classified does not mean that all of its contents are secret. For example, I suspect these documents contain statements equivalent to "The United States was attacked by Al Qaeda on September 11, 2001" as part of their factual summary. But obviously the fact that the documents are classified does not mean that those with access to the documents cannot discuss that particular fact.

Moreover, the fact that the opinions cited the 2001 AUMF as authority for this program was one of the things reported in the original NYT article. So, even if the government had originally intended to keep its legal reasoning secret as well--as opposed to merely keeping the actual program secret, which is a sufficient reason to classify these documents--that secret had already been revealed.

Nonetheless, the government continues to have a reason to keep these documents classified--they are undoubtedly full of operational details. But the mere fact that these documents are classified does not imply that there are legal arguments they are still keeping secret.

Noah,

Again, I'm not sure I follow you. The President does not need to declassify the relevant legal opinions in order to discuss his legal reasoning, nor to authorize others to discuss their legal reasoning. Again, that is because a legal opinion is a document, not a viewpoint, and not all of the contents of a classified document are necessarily secret.
1.5.2006 6:46am
subpatre (mail):
K Parker - Although your point about visual representation is correct, it's not my claim. It's normally a nitpick, but DtI's erroneous and extremely misleading claim needed negation before 'text' slid into the 'readable text' territory.

See DtI's reply to you where that anticipated claim is almost completed.

No doubt not the last word, but binary isn't actually ones and zeros anyway, but two distinct levels. Like flipping a light switch on and off; there may still be other light in the switch's off-state, but it's a distinct difference from the on-state. Inside computers the binary is two voltage levels, often different levels depending on what part of the computer; inside optic cables it's two visible or infrared light levels.


18 USC 1030 - Indeed, you have missed something; the crux. If a small part of data stream is examined through a digital filter and put in a database, one (good) argument is that no privacy violation can have occurred until a particular stream is kept.
1.5.2006 6:59am
John Lederer (mail):
"So, in order to intercept circuit switched voice, you either have to monitor an entire bundle of circuits over a period of time, saving off the routing information whenever a call is set up, or go to the nodes where the routing information is maintained."

I think if you have access to the switch/router that "serves" the submarine cable you have the whole ball of wax.
1.5.2006 9:35am
Bruce Hayden (mail) (www):
John Lederer said:

I think if you have access to the switch/router that "serves" the submarine cable you have the whole ball of wax.

As I noted above, the NSA (or some agency it utilizes) does have a submarine that can tap fiber optic cables, that is becoming technologically harder and harder to do as technology progresses. One problem is that fiber works by essentially cladding the light transmitting core with an effective mirror to almost eliminate light loss. Thus, getting at the light between a switch and a repeater would almost by necessity result in increased signal loss. Not good, and presumably detectable. Thus, I am theorizing that they actually monitor the repeaters, 80-100 of which are needed to get a signal across the ocean. But still somewhat problematic from a technical point of view, and thus, again, why the bulk of the surveilance logically has to be at the switches.

Just to amplify this, a quick primer on cable optic communications. Voice is typically sampled at 56kbs (56x10^3 bits per second). Copper wire would typically support 24 (T1) or 32 (E1) 56kbs channels simultaneously, for an approximate 1.5 mbs (1.5x10^6) capacity. Optical fiber works utilizing (obvious) monocromatic light, which can be switched at a much higher rate, somewhere on the order of 40 gbs (40x10^9 bits per second). Thousands of calls are simultaneously carried over one fiber strand at one optical frequency using a technique called Time Division Multiplexing (TDM), where each call gets allocated a time slot, with the time slots constantly rotating. In the case of T1, for example, you would have a constant rotation of channels 0,1,2..22,23,0,1,2.. (and channel 0 is the supervisory channel that carries the call setup information for the other 23 channels). Each time slot is typically a fixed length, whether one bit, or 100 bits long. (And note that these TDM time slots are reserved for a given call at call setup across the entire length of the call).

Two things complicate this further. First, multiple fibers are bundled together for cable runs under the ocean. And secondly, in recent years, multiple frequencies of light have been implemented to multiplex multiple TDM channels on one fiber - typically now 4, 8, or 16 frequencies (or colors), with higher numbers expected in the future.

The purpose of this quick primer is to point out (again) why it is much simpler to conduct surveilance at the switches. This is where all of this multiplexing, both through TDM and utilizing different colors (or frequencies of light), is done and undone. It is thus far easier to look at or capture one given conversation at that point, than during actual transmission.
1.5.2006 12:07pm
18 USC 1030 (mail):
Subpatre

No I did not miss that, perhaps I didn't touch on it as strongly as I should have, it was late, I was tired, I didn't make much sense. This is plausable; however, there are a few concerns. First, in order to obtain the information related to the account (i.e. who owns it and from where), they would, in a criminal investigation need to obtain a subpeona for such information. This never occured. Also, is one really to assume that all the NSA implemented was a tap/trace device? I find it had to believe that they did this and did not intercept any contents. In fact, it seems to me that they have said as much in the rhetoric that says that attacks have been prevented by this implementation. That means one of three possibilities occured: contents were gathered allowing authorities to prevent a specific threat as outlined in a message; authorities obtained only envelope information then detained those involved in the converrsation and somehow got the information out of them; or; no specific threats have actually been prevented and that it is pure rhetoric. I think it likely that either contents were gathered or there is fun with rhetoric. I do not believe a tap/trace was utilized.

Also, information I reference earlier, I'm afraid is a bit misleading, some of that information also is guided by a subpeona. See, Hill v. MCI, 120 F. Supp. 2d 1194 (S.D. Iowa 2000). See, also H.R. Rep. No. 103-827 at 10, 17, 31 (1994). The amendments to 2703 and the case law dictate a difference between "true envelope information" and "information pertaining to a customer or subscriber." Thus, with a subpeona, authorities can find the owner of the account, their address and other like information. In order to obtain logs and transactional records as discussed previously as information pertaining to a customer or subscriber, the authorities must obtain a 2703(d) court order. The burden for a court order is significantly higher than that of a subpeona which obviously would not have been granted. The records could have been obtained via subpeona had the authorities provided advanced notice to the customer which without question did not happen.

This issue though, matters only for cases where US persons were being investigated. Communications in which an outisde party was being investigated and merely obtained incoming data from the US would, under the ECPA be legal. However, if US persons were the target of the investigation, I am not sure how the ECPA would not apply. Thus, requiring subpeona with advance notice or 2703 (d) court order; this is especially true for the "mistakes" that involved obtaining communications between two US persons.
1.5.2006 1:49pm
Chris Gwinn (mail):
With the advent of unicode, a string can map to several different sequences of bits/bytes, depending on the character encoding used (unicode isn't always 16 bit, despite popular opinion). So matching a single bit pattern could be foiled by something as simple as using unnecessary byte continuation or a different encoding.

You would want to write your program to look for sequences of characters, or it wouldn't work correctly with unicode. And presumably the NSA isn't spying only on people writing in English.
1.5.2006 2:14pm
John Lederer (mail):
Bruce Hayden,

New submarine cables can be laid for 2000km+ before needing a repeater. That is expected to increase.

Either NSA has a technology breakthrough on tapping fiber optics (which may be the case -- there are secondary effects of light passing through matter) or it is the landside installation that is the focus.

My comment was made because of CALEA.
1.5.2006 4:49pm
Jim Erickson (mail):
Wonder if there might be any connection with Orin's guess about NSC's methods, and the recent attempt to wrest control of the Internet out of the US and into the UN.....
1.5.2006 5:56pm
Bruce Hayden (mail) (www):
John,

Thanks for the update. My information is obviously a bit stale then. Not surprising. I read somewhere about the NSA submarine, but it again may be old, used to tap into previous technology. As I think I indicated, I think that it would be quite hard to effectively tap into a current generation fiber optic cable without giving away that you have done so. My hypothesis that if it is being done, it has to be done at the repeaters is just that, a hypothesis, given the physics and engineering of fiber optic cables between repeaters. But if the distance between repeaters is increasing (from 2,000 km), we are very near the time repeaters won't be necessary.

Much more likely and useful to be tapping at the switches. But looking at the CALEA stuff I could see (obviously, you can see more), their FBI requirements don't appear to be able to support the high level of simulataneously being monitored at any one time supposedly being carried out by NSA.
1.5.2006 6:05pm
Bruce Hayden (mail) (www):
Jim Erickson

My guess is no, this has nothing to do with foreign govts. trying to wrest control of the Internet from the U.S. That seems to be more a debate about control over top level domains, such as .com, .org, etc. Also, I think that a lot of their concern is internal - how to keep their citizens from communicating with the rest of the world.

My understanding of the relevant technologies is that the U.S. govt. really has little control over routing of packets over the Internet. As far as I can tell, that routing is pretty transparent. I have several tools installed on my computers that can be utilized to trace packets as they move from here to there. One thing that is interesting is that at times, routes can change dramatically from one minute to the next. This is a function I think of IP protocol being designed to be survivable in the face of nuclear war.
1.5.2006 6:22pm
Noah Klein (mail):
Medis:

I may be wrong, but I always thought that if something is classified both the document and the information contained in the document cannot be legally released until it is declassified. I think this because even when a document is released portions of it can remained redacted.

I was also under the impression that any document the president or others in the administration feel needs to be classified can be classified. This why legal opinions, histories and other such documents can be classified. This was the case in the Pentagon Papers.

Noah
1.5.2006 7:54pm
Medis:
Noah,

That is not quite my understanding. As I understand, it is actually specific pieces of information that are classified, and a given document may contain different portions containing different levels of classified information (marked according to certain marking rules, which I believe go down through at least parts, sections, and paragraphs), as well as portions marked as containing unclassified information. Each page of the document will be marked with the highest classification level of the information on that page (and so some pages could be marked unclassified), and the whole document will be marked with the highest classification level within the document.

So, the overall classification level of the document is the same as the highest classification level of the information in the document, but that document could also include pages and specific pieces of information marked as unclassified.

Incidentally, I believe that something like a legal argument incorporating classified information (eg, the operational details of this program) would be given a "derivative classification", meaning it would be marked consistently with the classification of the information it contains.

On the classification rules--I believe there are a series of Executive Orders governing all this. And we don't have a general Official Secrets Act like Britain--a law criminalizing the disclosure of government information--although we do have a set of laws covering specific situations.
1.5.2006 8:35pm
Polaris (mail):
Medis,

You said:


subpatre,

The fact that a document is classified doesn't mean all of its contents are classified


That isn't quite true and it seems clear that you haven't handled classified information. While it is true that not all sections of a document might have equal classification (and in fact you will see various paragraphs and subparagraphs with different security designations), you are required by law to treat the entire document as though it was the highest clasification of any information in it. Certainly both the DOD and NSA regs explicitly state this, and it is known by anyone with OPSEC training.

Only those with originating classification authority (which would *not* be the AG in a case like this) can seperate them out in the way you are suggesting.

-Polaris
1.5.2006 11:02pm
John Lederer (mail):
Bruce Hayden,
http://en.wikipedia.org/wiki/USS_Jimmy_Carter_(SSN-23)
1.6.2006 1:30am
Bruce Hayden (mail) (www):
John Lederer,

Thanks for the Jimmy Carter reference. The article seemed to indicate that this wasn't the first, just the biggest and most sophisticated submarine for that can be used for serveilance. Wonder what he thinks of the uses being made of his namesake? (which is why, I think, that most ships are named after dead people).

It may be possible to use this ship to tap fiber optic cables outside our territorial limits, and thus put the surveilance outside of FISA. But I still don't think that that would be that effective, esp. as compared to the volume that is potentially being done through switches tied to undersea cables.
1.6.2006 1:58am
Medis:
Polaris,

I don't see how you are disagreeing with me. As I noted myself in the post before yours, the document itself receives the highest classification level of any information within the document. Which makes perfect sense--if you hand an entire document to someone, you hand over all the information in that document.

My only point was that subpatre seemed to be assuming that since these legal opinions were still classified, all the information in those legal opinions was classified. As I think both you and I have noted, that is a false assumption, because a classified document can contain unclassified information.

Again, consider a simple example: suppose one of these legal opinions contains a background, single-sentence, paragraph stating, "On September 11, 2001, the United States was attacked by the foreign terrorist group known as Al Qaeda." This would not be classified information, and it would not be marked as classified information. Accordingly, people who had seen these documents would not be barred from discussing that particular piece of information.

Indeed, it would be absurd for it to work any other way. Information that is unclassified--that we were attacked on 9/11, that the President of the United States is George W. Bush, that the sky is blue and 2+2=4--can't become classified simply because it is incorporated into a classified document.

All of which is just to say that one cannot assume--as subpatre seems to assume--that any information in a document receives the same classification level as the document itself. Again, as both you and I point out, the opposite is the case--the document itself receives the highest classification level of any information in the document, but that does not convert the classification level of all the information in the document to that highest level. Indeed, such documents logically must, and in fact do, contain information which is unclassified.
1.6.2006 10:16am
Challenge:
"All of which is just to say that one cannot assume--as subpatre seems to assume--that any information in a document receives the same classification level as the document itself. Again, as both you and I point out, the opposite is the case--the document itself receives the highest classification level of any information in the document, but that does not convert the classification level of all the information in the document to that highest level. Indeed, such documents logically must, and in fact do, contain information which is unclassified."

So?
1.6.2006 10:42am
Just an Observer:
I am coming the the tentative conclusion that the NSA program in question may fall through a large loophole in FISA on technical-definition grounds.

The loophole, described most usefully in posts by John Lederer in several related threads on this site, obtains when the intercepted communication is from a fiber-optic cable and not a "wire communication" under 50 USC 1801 (l) because the company owning or operating the cable is not legally a common carrier.

(As an excerise, I invite you to compare the narrowly defined provision of 50 USC 1801 (l) with the analogous definition governing purely domestic criminal wiretapping, 18 USC 2501 (1). The FISA definition contains the "common carrier" restriction; its domestic cousin does not.)

As John notes, the bulk of international traffic is carried on fiber cables, and many (most?) submarine cable operators are not common carriers. Some major domestic U.S. telecom cables also are not owned by common carriers. John correctly reasons that it would be technically easy to tap most international communications at or near U.S.-based communications switches where the submarine cables terminate on shore, with the cooperation of the private-sector owners.

Such a scenario is consistent both with the reportage of the Risen book and with the NYT story of 12/24 (not the original story of 12/16), which disclosed that the NSA had arranged with telecommunicatons companies to tap into such arterials in the United States.

Notably, the disclosure of this key technical detail came after the Bush administration's press briefing 12/19 or the DOJ letter 12/22. In neither forum was a legal justification that included such a factual scenario addressed. The technical details were unknown to the public at that time.

To me, this apparent loophole has two ramifications for discussion:

1) When discussing whether the NSA program "violated FISA" — no matter how weak the case now made by the administration based on the 2001 AUMF — we should remember the caveat that such technical loopholes seem to exist.

The hardest thing to reconcile on that legal point is the literal wording of Alberto Gonzales on 12/19, which seemed to say that the NSA program would have required court orders under FISA but for his assertion of an override by the 2001 AUMF. I can't reconcile that, but neither can I base my whole analysis on parsing his garbled sentence in the face of what newer facts on the ground, and the apparent black-letter loophole, seem to indicate.

2) For purposes of congressional oversight, I would argue that the FISA defitions need to be fixed.

One possible statutory fix would be to replace the "common carrier" clause restricting the definition or "wire communication" of 50 USC 1801 (l) with language from the more inclusive definition of its domestic counterpart 18 USC 2501 (1) — "furnished or operated by any person engaged in providing or operating such facilities for the transmission of interstate or foreign communications or communications affecting interstate or foreign commerce."
1.6.2006 10:57am
Medis:
JaO,

Maybe I am missing something, but I don't see the loophole. If this is not wire communications because the cables are not common carriers, then 1801(f)(4) potentially applies. It covers:

"the installation or use of an electronic, mechanical, or other surveillance device in the United States for monitoring to acquire information, other than from a wire or radio communication, under circumstances in which a person has a reasonable expectation of privacy and a warrant would be required for law enforcement purposes."

When John L. was talking about using submarines to tap cables on the seafloor, (f)(4) did not apply because that would be outside the United States. But if this tapping is occuring inside the United States, then (f)(4) does apply, at least as far as technical details are concerned.
1.6.2006 11:10am
Just an Observer:
Medis,

1801(f)(4) is iffy, as John Lederer notes himself. I failed to include this in my previous post. I stopped writing with the analysis of 1801(l), and didn't work through the 1801(f) analysis specifically.

1801(f)(4), refers to a "installation or use" of a "device ... for monitoring to acquire information, other than from a wire or radio communication."

My own common-sense reading of that is that is was intended to cover such things as hidden microphones or other bugs unrelated to telecommunications. I'll bet that is what the legislative history would show. The definition certainly would have excluded an old-fashioned telephone tap, because the old-fashioned telephone call would have been on a wire communication over a common carrier.

It may be possible to construe 1801(f)(4) to cover a tap at a modern telephone switch which is not owned by a common carrier, and thus does not technically involve a "wire communication," but I don't think that is clear. One could argue the point either way.

(Note that none of this discussion of 1801(f)(4) does anything to close the underlying loophole in "wire communication", which continues to screw up the application of 1801(f)(1) and 1801(f)(2). These provisions had meaning before the industry shift away from common carriers.)
1.6.2006 11:49am
Medis:
JaO,

I've learned not to guess at legislative history, but it seems to me that (f)(4) by its plain terms is designed as a catch-all. In the part you note, it refers to "monitoring to acquire information, other than from a wire or radio communication." It seems to me the clear intent is to have (f)(4) cover any sort of communication not defined as a wire or radio communication, whatever that other communication might be.

Moreover, in the part you didn't note, it refers to "the installation or use of an electronic, mechanical, or other surveillance device . . . ." That is a very broad sense of device (electronic, mechanical, or other), and a very broad sense of what can be done with a device (installation or use).

And, of course, reading f(4) as a catch-all makes sense given the structure of the rest of 1801(f). The other three provisions all explicitly apply either to radio communications, or wire communications, or both, but nothing else. So, including a final catch-all for other cases makes perfect sense.

Of course, I have no doubt they were not imagining all that could be done in the future that would fit within category (4). But that is the precise point of a catch-all provision--to relieve them of having to predict every sort of circumstance that could arise.
1.6.2006 12:04pm
Medis:
Challenge,

You can look back at subpatre's posts for an explanation of why that erroneous assumption mattered to his argument.
1.6.2006 12:18pm
Just an Observer:
Medis,

I think your more expansive reading of 1801(f)(4) is not unreasonable. I mean no disrespect when I say it is not a slam-dunk, either.

Since no court except FISC is generally in the business of interpreting FISA, I suppose finding any public caselaw on the point is hopeless.

The legislative history, of course, would be the next thing to research. I can't find anything online that far back.

To me, it remains an open question right now. It leads into the larger structural issues about of how open or arguable legal questions are determined in this environment.
1.6.2006 12:42pm
John Lederer (mail):
Medis,

No tangible object would need to be installed at the switch, just software changed or duped.

A switch in general principle works much like a mail sorter in an old fashioned post office -- a mixture of mail comes in and he sorts it into different cubbyholes for transmission further along. In a switch a mixture of addressed packets come in, and are sorted into outgoing ports, which in turn connect to cables.

All is controlled by software. Changing the software or its data tables changes the result.

Return to my mail analogy. If we trick the fellow with the green eyeshades about which country code is which cubbyhole, have I put in a device?
1.6.2006 12:45pm
Medis:
JaO,

I absolutely agree that this gets into a larger structural issue. And for that matter, those larger structural issues are often lurking behind statutory interpretation disputes even in less pressing cases. So, I'll admit to having a strong preference in favor of resolving issues on the basis of the plain text without reference to legislative history when possible, and that such a view is motivated in part by structural concerns. And I understand a lot of people do not share that preference, or at least not so strongly.

John L.,

In line with what I just said to JaO, my first instinct is to look at the text. The actual language is "electronic, mechanical, or other surveillance device". So, I don't think this language requires a tangible device, because a nontangible device like a computer program could be "electronic". Indeed, a quick dictionary.com check yields this definition of "electronic":

"e·lec·tron·ic

1. Of or relating to electrons.

2. Of, based on, operated by, or otherwise involving the controlled conduction of electrons or other charge carriers, especially in a vacuum, gas, or semiconducting material.

3. Of, relating to, or produced by means of electronics: electronic navigation; electronic books.

4. Of or relating to music produced or altered by electronic means, as by a tape recorder or synthesizer.

5. Of, implemented on, or controlled by a computer or computer network."

Computer software is thus "electronic" under definition #5, and I would suggest a program implemented on a computer for the purpose of surveilling communications passing through that computer is an installed electronic surveillance device.

Indeed, to build on your example, suppose someone crafted a virus (excuse me if this is not the correct technical term, but you will get the idea) that infected a host computer and instructed it to copy and send communications passing through that computer to some destination. I think this virus would likely count as an "electronic" surveillance device.

Moreover, even if we think "electronic device" meant something tangible, there is still "other device". Again, "other" seems to be a deliberate catch-all provision, and I see no reason why software would not fall into this provision if it was not already "electonic".
1.6.2006 1:18pm
Medis:
John L.,

Sorry, I forgot one crucial point. This language--"electronic, mechanical, or other surveillance device"--runs throughout the definitions of electronic surveillance in 1801(f). So if software never counted, that would imply a TRULY gaping hole in FISA--eg, the government could infect every computer in the US with the virus I described, instructing our computers to send all our emails and other electronic communications to the NSA, without ever conducting electronic surveillance within the meaning of FISA. Of course, maybe that hole exists--but I somewhat doubt that would be considered a reasonable reading of FISA.
1.6.2006 1:23pm
Medis:
John L.,

Sorry again, one other forgotten point: I'm not sure I saw your mail sorter argument before, but obviously tricking a human mail sorter may not involve an "electronic" device in the sense I described. Indeed, it may not involve a device at all--eg, if a human being gave him bad information.

But I do think this raises the question of what a "device" is. Again, a quick dictonary search reveals this common definition:

"A contrivance or an invention serving a particular purpose, especially a machine used to perform one or more relatively simple tasks."

As noted, the definition in this case does not require the device to be mechanical. But is a computer program a "contrivance or an invention serving a particular purpose"? I would say that is an excellent fit.
1.6.2006 1:30pm
John Lederer (mail):
I just talked with an old friend, more knowledgeable than I, who made what seems to me a very logical set f inferences:

(1) We have a pretty strong indication that the UN delegations in New York that were interceted during the Iraq runup were intercepted in England because of the English gal that broke security.

(2) We know that submarine cables are intercepted in England -- they even run them to the interception station.

(3) Wouldn't the trick be to simply tell the switches in the US to route traffic of interest through the cables that go through England rather than some alternate route like the cable to France? We know the computer power, cabling, specialists, cyprologists, etc are already there.
1.6.2006 1:48pm
Just an Observer:
Medis,

Let me explore some byways of the structural issue. My purpose is to raise questions for which I have no ready answers. Wiser heads may.

I think we both agree that in our system the traditional way of resolving matters of statutory interpretation is for a court to decide. The court may or may not look to legislative history. What matters most is that the court gets to decide.

Suppose, hypothetically, that the NSA/DOJ lawyers theorized that the common-carrier loophole does exist, and further that 1801(f)(4) does not apply. So they proceeded with wiretaps without requesting any order from a FISC judge, but notified the chief FISC judge that was what they were doing.

(A similar hypothetical can be constructed for most any alternate legal theory justifing the NSA program.)

Did the chief FISC judge somehow "uphold" the administration theory? There was no case. Was any order issued confirming assent? Was any information withheld from the judge?

Another hypothetical: The FISC judges reportedly are due for some sort of "briefing" from the adminstration about the NSA program, but the issue reportedly is a narrow one — were any FISC orders bootstrapped from information derived from the warrantless wiretaps?

Suppose the answer is yes. What it is remedy, and what could the judges do about it? There are procedures for cases when retroactive approval of wiretaps was sought but denied, but not for cases where orders turn out to have been wrongly issued. There also are no procedures for wiretaps where no warrant was sought, but now may turn out to have been required after all.

I can find no answers in the FISA statue itself. What authorities do the judges have to cure such problems if they now find them? Does the judiciary have any inherent power to effect such cures?

Are there precedents from the Article III courts that would help answer these questions?
1.6.2006 2:02pm
John Lederer (mail):
Medis,

Your "virus" example brought to mind a curious and never definitively explained incident (here insert the "dum-de-dum-dum" music of the "Twilight Zone"):

http://en.wikipedia.org/wiki/NSAKEY
1.6.2006 2:02pm
Medis:
John L.,

As an aside, I've had some interesting discussions in other contexts about whether the US government should be given secret backdoors in software (perhaps which they could only use with some sort of warrant). One notable problem is that if it truly got out that US companies were doing that, they might lose their stranglehold on the market.

Anyway, on your hypotheticals:

One way this might come up in court is if any such information (straight from the program, or as evidence for a FISA warrant which then generated information) was used in a criminal prosecution. A court may then have the occasion to look back down the chain in order to see if the evidence in question was admissible.

Another possibility is that someone actually gets prosecuted under Section 1809 of FISA, and raises this as a defense. That may seem far-fetched, but it is early days yet (eg, we might get an independent prosecutor at some point, or maybe a new President--indeed, we will certainly get a new President in 2008).

A third possibility is that someone could sue under Section 1810 of FISA, in which case this again could all come up as a defense. Again, that might seem far-fetched, particularly in light of the state secrets privilege that wiped out the surveillance cases from pre-FISA, but we again do not know yet what information might become publicly available. And, again, things might change with a new President.
1.6.2006 2:42pm
Just an Observer:
Medis,

(I assume it was my hypotheticals you were responding to.)

I wasn't thinking about how someone might have standing to bring a civil action, or that someone surveilled might be prosecuted and thus have a remedy, or some other route into the Article III courts. Those are all interesting issues, and there is nothing wrong with exploring them.

The particular issues I am wondering about are what FISC judges can do in their own right -- especially since they are scheduled to address the issue in their so-called briefing. They may even have secret case law that resolves some of the FISA technical-definition issues.

Looking at the FISA statute, there are actions prescribed in analogous (but admittedly different) situtations. If, for example the attorney general applies retroactively for a court order under the 72-hour emergency window and the application is denied, the government is supposed to purge and make utterly no use of all the information gathered, and the court is to notify the surveilled parties unless the government shows cause not to.

What if, after this briefing, the FISC judges say it was all illegal, and find that the illegal surveillance tainted court orders they previously issued? Can one or more judges force the issue on their own by ordering some sort of remedy for two classes: those who were unlawfully surveilled in the first place and those whose own court-ordered surveillance was the fruit of the unlawful surveillance.

Nothing in the FISA statue grants the FISC judges any such authority. Can it be inferred from their inherent judicial powers or similar judicial practice?

For that matter, nothing in the FISA statute covers such a "briefing" of the FISC judges, or the prior "briefing" of the chief judge.

When Judge Robertson resigned, some reports in the Washington Post quoting sympathetic friends and colleagues were not so much about the NSA case, but about the way the court was beeing used. One such source raised the image of a "Potemkin court."

To whatever extent FISC judges have such institutional concern, do they have authority to do anything about it?
1.6.2006 3:22pm
John Lederer (mail):
"What if, after this briefing, the FISC judges say it was all illegal, and find that the illegal surveillance tainted court orders they previously issued?"

One of the things that bothers me about the FISC court is that its warrants really aren't "warrants" in the classical sense. Instead the court seems to be acting as a supervisor of prosecutors. Where is the "case or controversy"? Is this really what we want courts to do?
1.6.2006 6:03pm
Just an Observer:
John Lederer,

You are correct that it is not called a "warrant" but rather an "order approving electronic surveillance." I don't really have a problem with the fact that the court has to approve that.

What to me is strange about FISC is not that it is a "supervisor of prosecutors," but rather the reverse -- it is effectively subordinated to the DOJ.

What I am wondering aloud about in my prior post is to what extent these judges are able to assert control the way an Article III court can. Sometimes that is because of the gravitas that accompanies the bench in a U.S. District or Circuit court.

If we are sure we need such dedicated, secret court proceedings, I wonder if we would not be better off letting the same specially designated judges preside while wearing their District Court hats (er, robes). If the government jerks around a District Court judge, there can be consequences.
1.6.2006 7:26pm
Medis:
JaO,

First, I apologize for the identity confusion.

Anyway, I'm not really sure what the FISC could do. Normal courts have the power to craft evidentiary rules, impose sanctions, and so on. But I actually don't know what the FISC is able to do. I do doubt, however, that they have any power to order some sort of remedy on behalf of the targets of this surveillance.

John L.,

If I'm not mistaken, it is considered a "controversy" when the government is seeking something like a warrant because it is assumed the target of the warrant is adverse to the government.
1.6.2006 8:13pm
Just an Observer:
Medis,

No sweat on the ID mixup. I think that makes us even from a prior thread. :-)

I have a feeling you are right about the FISC being powerless, but I pose the question nevertheless.

It is worth noting that, if the chief FISC judge when "briefed" was powerless to block such surveillance, then it is quite preposterous to cite the briefing as an endorsement of its legality.

Still, as a public-policy matter, this part of FISA seems ripe for reform. The negative label of "Potemkin court" does not seem far-fetched.
1.6.2006 9:12pm
Just an Observer:
Medis,

I have just read the recently surfaced (leaked?) Congressional Research Service legal memorandum analyzing the NSA surveillance issue, and I have to say the legislative history it cites tends to support your expansive interpretation of 50 USC 1801(f)(4).

Footnote 78 on page CRS-20 states:

With respect to the ability of FISA to keep pace with the rapidly changing level of communications technology, it is possible that 50 U.S.C. § 1801(f)(3) and (4) may provide some or all of the needed statutory flexibility. See, e.g., S. REP. NO. 95-604(I) at 34-35, 1978 U.S.C.C.A.N. at 3936, discussing the congressional intent that subsection 1801(f)(4) was intended to be "broadly inclusive, because the effect of including a particular means of surveillance is not to prohibit it but to subject it to judicial oversight." Thus, it was intended to include "the installation of beepers and 'transponders,' if a warrant would be required in the ordinary criminal context. . . . It could also include miniaturized television cameras and other sophisticated devices not aimed merely at communications." Id. See United States v. Andonian, 735 F. Supp. 1469, 1473 (C.D. Cal. 1990), aff’d and remanded on other grounds, 29 F.3d 634 (9th Cir. 1994), cert. denied, 513 U.S. 1128 (1995).


I'd still like to see the Senate committee report the footnote references, since the examples of beepers, transponders and miniaturized video cameras are still the kind of devices I envisioned to be like microphones, and have less to do with telecom communications analogous to "wire" or "radio." What I'd like even more would be that a court decide the applicability in the particular set of facts we are discussing for non-common-carrier wire transmissions!

The possibility of a "common carrier" loophole in the FISA definition of "wire communication" was not addressed by CRS, so far as I could tell.

Interestingly, the CRS memo suggests there may be an additional layer of legal of coverage of any surveillance that falls through the sieve in the FISA definition:

On page CRS-21:

The interception of wire, oral, or electronic communications that is not included within the definition of "electronic surveillance" for the purposes of FISA may nevertheless be prohibited by or subject to a warrant requirement pursuant to 18 U.S.C. § 2511 (Title III).


Overall, I think the CRS memo, produced by two attorneys in its American Law Division, comprises a thorough rebuttal of the Bush administration's legal case advanced so far. (Much of it will not be news to close readers of VC.)

For purposes of public discussion, we are lucky to have the document. CRS, a nonpartisan research arm of the Library of Congress, works only for members and its reports are not routinely published.
1.6.2006 11:59pm
Medis:
JaO,

Yeah, that legislative history is basically what I expected: an expression that this is a catch-all, but a limited imagination of what we might later be doing. Which, as you point out, is not quite conclusive.

Incidentally, I also found the Title III discussion interesting--we have, of course, talked extensively about 18 USC 2511(2)(f), but not so much about the rest of Title III and how it might apply.

Anyway, I would love to believe this memo will convince some particular people that there is a nonpartisan problem in this case--but I somewhat doubt it.
1.7.2006 12:16am
jukeboxgrad (mail):
Medis,

"whether the US government should be given secret backdoors in software"

Something somewhat like this seems to already be happening: "In a purported effort to identify counterfeiters, the US government has succeeded in persuading some color laser printer manufacturers to encode each page with identifying information" (link). Sorry if you already know this.

When I think of this alongside the idea of phone companies inviting government spooks into the inner sanctum, I think the general pattern is of government and business becoming indistinguishable from each other. This also seems to be a theme of the Abramoff story.

I'm not sure if the right word for this is fascism. Maybe it's corporatism.
1.7.2006 12:36am
Medis:
jukeboxgrad,

I actually did know that about laser printers, but it was certainly worth mentioning.

Incidentally, on your last theme, I am also reminded of Einsenhower's farewell speech, in which he warned of the dangers of a growing military-industrial complex. Oh well.
1.7.2006 12:45am
jukeboxgrad (mail):
Good point. For the sake of posterity, it might as well be a part of this thread:

"In the councils of government, we must guard against the acquisition of unwarranted influence, whether sought or unsought, by the militaryindustrial complex. The potential for the disastrous rise of misplaced power exists and will persist.

We must never let the weight of this combination endanger our liberties or democratic processes. We should take nothing for granted. Only an alert and knowledgeable citizenry can compel the proper meshing of the huge industrial and military machinery of defense with our peaceful methods and goals, so that security and liberty may prosper together." (link)

They don't make Republicans like they used to.

Maybe an alert and knowledgeable citizen AJ will now chime in that Ike was a pinko.
1.7.2006 1:36am