My Comments on:


Bill Heuisler's comments about computers and telecommunications


in response to

William C. Kashatus,

Is Bush Just Following Lincoln's Example?
,

http://hnn.us/articles/19928.html


HNN , before Dec. 31,2005

Andrew D. Todd

 a_d_todd@rowboats-sd-ca.com 

http://rowboats-sd-ca.com/




(My Responses)
HNN posr, William C. Kashatus,  Is Bush Just Following Lincoln's Example?

01/03/2006 09:23 PM

If You Really Want It That Way?

http://hnn.us/articles/19928.html

http://hnn.us/board.php?id=19928
http://hnn.us/readcomment.php?id=73447&bheaders=1#73447

So, how do you know if your phone has been tapped?  With digital telecomm switching, phone tapping is not something  people do on a street corner. The "bonus services" like caller ID, conference calling, and call forwarding  implement all the raw technical capabilities required for telephone tapping. The CALEA act required telephone companies to incorporate telephone tapping into the computer programs which run the network. The number of people, in the telephone companies,  outside of the NSA,  who have "need to know" for massive telephone tapping  is approximately on  the same order as the number of Air  Force and Navy officers who have "need to  know" for nuclear weapons launch codes.  I know enough about telecommunications engineering to know how little I know. It's not my engineering specialty. Most of the HNN conservatives have never programmed a computer in their lives, and they think they know all there is to know!!! They know nothing and they think they know everything. I used to have conversations with la-di-da, too-good-for-this-world, academic Marxists who were like that. It's a strange reversal. I don't know whether it's simply a matter of the Trotskyite roots of Neo-Conservatism, or whether it's just a matter of critical technological lag.

You have  no way of knowing if your telephone is being tapped. What you can do is to assume that your telephone has been tapped, and automatically take such actions as you would consider justified in that eventuality. There  are various technical measures you can take, such as encryption.

http://yro.slashdot.org/article.pl?sid=05/12/20/221204&tid=158&tid=126&tid=219

http://yro.slashdot.org/article.pl?sid=05/12/25/0029204&tid=158&tid=215&tid=219

http://ask.slashdot.org/article.pl?sid=05/12/18/1456210&tid=158&tid=123&tid=4

In  1774-75, in response to events, large  numbers of  villages in New England organized "Committees of Correspondence and Defense." Something similar  seems to  be happening among the computer people on the  internet.

[Responding to Bill Heuisler's claim that  (1/4/2006) message traffic could be easily intercepted and  anlyzed, I replied:]

(01/04/2006 06:30 PM)

Well, matters have rather changed over the last forty years or so.

Physical Plant:

In the first place, there's much more optical fiber in place nowadays. The telecommunications glut of the 1990's means that people are practically giving bandwidth away. Fiber optic cables are preferable to satellite links because their  time lag  is less. It is something like 20,000 mile up to a geostationary satellite, and 20,000 miles down again. That is, about a fifth of a second at the speed of light. I know that doesn't sound like much, but it does play the devil with machine feedback, and you have to design special mechanisms to  get around it.  By contrast, an optical cable to Europe  is only about three or four thousand miles long, and even allowing for the diminished speed of light in glass,  the time lag might be about one  sixth of that in a satellite  link. The better grade of optical fibers, used in long-distance telecommunications, are usually "graded index single-mode." To tap in, you  would have to cut the cable, and this would cause alarms to ring in the telecommunications company's switching center. The whole  point of the CALEA act was that the FBI was becoming afraid because it was losing its traditional wiretapping ability due to technological changes.

Additionally, packet switching is in widespread use. Packet switching tends to convert messages into waves of packets, each traveling independently along the momentarily least congested route. Telephone tapping in a packet regime practically requires that packets be steered to a point where they can be picked up. 

Sattelites and radio generally are primarily useful for communication with mobile units. The tendency is to incorporate enough cryptography to give at least as good privacy as a landline, because any kid can use generic components to build a receiver tunable to any frequency.

Encryption: 

The gold standard of encryption  is the so-called  "once-only-cipher." You take a stream of true random numbers, produced by an electronic "noise generator," XOR them against the plaintext to get  the ciphertext, and XOR them against the ciphertext to get the  plaintext back. And  you never use the key again, which is why the  cipher is called "once-only."  You can  put something like 4.7 gigabytes (4700 books) of once-only-key on a DVD  for about a dollar, so the difficulties of once-only ciphers are not what they once were. Once-only-ciphers are theoretically unbreakable-- that is, there is no logical or even statistical basis for asserting that one putative plaintext is the correct solution of a given ciphertext of the same (padded) length. 

If  you know someone to the point of  conspiring with him, the only sane thing to do is to exchange DVD's, and go to once-only cipher  in the interests of peace-of-mind. One can take for granted that all the more critical communications of the  Pentagon are on this basis. Lesser ciphers are useful for dealing with people you don't know well enough to set up a special arrangement with. And of course, people you  don't know very well are precisely the people you can't trust not to publish  messages sent to them.

Now, for ordinary encryption:

There  is something called Bremmerman's Conjecture, an argument from quantum physics about the  ultimate limits of  how fast a computer can ultimately go. The consensus is that a "complexity" much in excesss of ten to the three hundredth power is  beyond the limits of any computer which can be  built. The computational difficulty of cracking a cipher increases much faster than the computational difficulty of encrypting and decrypting. A typical desktop computer, which would have been worth perhaps a hundred million dollars in 1975, can shove a message into Bremmerman's Conjecture territory without undue difficulty.

Now, of course there is this buzz going around about "quantum computing," which is hard to assess. However, it is agreed that there are certain operations which quantum computers cannot perform. Private key ciphers are designed to methodically use just about every possible operation. They are designed rather like the thicket in which Brer Rabbit was  born and bred.  Furthermore, they use "autocoding," an operation analogous to carrying in addition,  so they can only be attacked at the start of  the  message-- after that, the autocoding gives the cipher the properties of a once-only-cipher. 

Public-key ciphers, at present, are not thicket  designed. They _may_ be vulnerable to quantum computing, and they seem exposed to advances in mathematics. It is presently unclear whether one can design a public-key cipher on the thicket principle. A partial solution to the weaknesses of public-key ciphers is the so-called  "keyserver," eg. Kerberos. You have an ongoing relationship with the party operating the keyserver, and they give you "cryptographic introductions" to people you want to talk to. That is, you use  your private-key cipher to communicate with the keyserver, and the other party uses his private-key cipher to communicate with the keyserver, and the keyserver generates a  random number and gives it to both of you to use as a private key in  communicating directly.

The main ultimate practical usefulness of public-key ciphers is not for secrecy as such, but for digital signatures. Someone can have enough information able a public-key cipher to determine that a signature is good, without being able to forge it. So the real threat of quantum computing is primarily towards signatures and electronic payments.

Incidentally, with the rise of "bot-nets,"  the NSA is no longer top dog in code cracking. The author of a computer virus can steal computer time worldwide faster than the NSA can  buy computers. Basically, everyone in the  telecommunications business knows that they have to take certain cryptographic  measures, or the Russian mafia will own all their subscribers' credit card numbers,  simply by tapping the phones of the  major mail-order firms.  A practical side effect of this is that if the NSA wants in, it cannot simply tap  lines, but has to  physically get into the telecommunications company's control room, and practically, that means inducing the telecommunications company to cooperate. 

Now, of course, if you don't trust the telephone company, you can superimpose your  own cryptography. That is what people are discussing doing. Telephone tapping is ultimately futile,  in  much the same sense that the British march to Lexington and Concord was futile. It merely broke down trust, and caused people to start acting in terms of the logic of force. One can say  with the benefit of hindsight that  General Gage was probably a rather stupid man who did not understand what America was all about. The logic of force meant that within a couple of days, he was besieged in Boston by 15,000 minutemen,  twice the total number of British troops in North America. The  logic of force as applied to telecommunications means that every  little girl burns a disk full of random numbers, and exchanges  same with her best friend, so that they  can giggle over the phone in perfect privacy.

[order of material shifted for clarity]

Let's finish off my field first. Fiber optic cables do not cease at the water's edge.  Another basic point you have to understand is this: the  information capacity of an electromagnetic wave-- that is radio, microwave, or light, is  proportional to its frequency. Satellite  radio is broadly speaking in the gigaherz range, that  is,  billions of cycles per second. The signal has to drill up through twenty miles of atmosphere before reaching space,  so you can't go too far up-frequency before it starts behaving like a weather radar. Weather radars are very  nice in their place, but for long-distance communication, they present certain problems...  By contrast, an optical cable runs on light in  the hundred teraherz range.An optical cable will typically have ten or so optical fibers. In  other words, an optical cable can carry as much information as a thousand  or more communications satellites.   As you may infer, the relation between an optical cable and a satellite is very much that between a truck and a mule. Undersea cables were  built to all kinds of improbable places. In particular, Global Crossing built a long  way  out into the Third World.

http://www.cellular.co.za/news_1999/news-06071999-africa_one_plans.htm

http://marketplace.publicradio.org/shows/2002/08/09_mpp.html

http://www.globalcrossing.com/xml/news/2002/january/11.xml

http://www.lightreading.com/document.asp?doc_id=51872

http://www.globalcrossing.com/xml/network/net_map.xml

http://home.singtel.com/about_singtel/network_n_infrastructure/submarine_cable_systems/networkinfra_submarinecablesystems.asp

http://www.convergedigest.com/DWDM/dwdmarticle.asp?ID=16871&ctgy=

If you want to get somewhere which is only  mule-accessible, what you do is to drive your truck, pulling your horse trailer to the road access point which is closest to where you want to get to. I think the same principle applies to cables and satellites.  Even it the cable doesn't go all the way, you might want to get within the footprint of a comparatively underutilized satellite, perhaps  over the  Indian Ocean, and avoid competing with mobile services and satellite broadcasting  in the comparatively congested and  expensive  North Atlantic region.


---------------------

Parenthetically, computer voice recognition does not work very well. As one Slashdot humorist remarked: "Voice recognition is AI complete." Unless a surveillance agency knows exactly what it is looking for,  it has to record a large swath of telephone conversations,  file them away, and eventually accumulate enough so that the telephone numbers which are called form a pattern. They need to know that little Jenny is little Brenda's best friend, simply for purposes of filtration.  You see the potential invasiveness.

You might review the Brandon Mayfield case, incidentally. It's a classic case of that general sort of thing going haywire. The FBI issued an apology, but the apology was not accepted, and the last I heard, Mayfield was suing them. His "theory" will be in effect that the FBI intentionally  attempted to isolate militants by interfering with their marital relations, and that he  got picked up on account of his work as a divorce lawyer. The truth is probably more subtle. Artificial Intelligence programs are like badly behaved dogs. They tend to surface and act on their owners' suppressed feelings and motivations. You don't like someone, but  you know you are not allowed  to bite him. Your dog knows that  you do not like the person, but cannot understand why biting is not permissible. So he  bites the person for no better reason than that you dislike  him.


(01/05/2006 12:23 AM)


Now then as to the Mayfield case, I take it your source is this, or perhaps the derivative Daniel Pipes column. 

http://www.thetribonline.com/archview.cgi?id=24584

http://www.danielpipes.org/blog/330

I'm going to tell a shaggy dog story first. Some year ago, the Guinness Book of World Records looked into odd card hands that people claimed to have dealt themselves. Computing the probabilities, the McWhiter brothers, who edited the Book, came to the conclusion that the entire world population would have had to have been  playing cards continuously for considerably longer than the earth has existed. No one seriously contends that brontosauruses played either  bridge or  poker. The editors'  conclusion was that  virtually all such claims had to be phony. 

Now, let's  look at the FBI's story in that light:

Quoting the  article:

"The FBI found 15 potential matches for the fingerprint found near the scene of the terrorist train bombings in Spain. But the bureau only arrested one man — local attorney Brandon Mayfield, a convert to Islam."

"
U.S. Attorney Karin Immergut and Portland FBI spokeswoman Beth Anne Steele both insisted this week that Mayfield was not targeted because of his faith.... 'That really had nothing to do with it,' Steele said. 'It was based on a computer analysis. The computer had no idea whether he was Muslim. (The fingerprint) was looked at by fingerprint examiners who had no idea he was Muslim. It was sent to us, and we had no idea who he was, much less the fact that he was Muslim.'"

If this account were true, it would imply that about twenty million people were terrorist affiliates of the same sort that Mayfield allegedly is, that is, about one fifteenth of the population. The FBI claims to have selected down to 15 people on the basis of fingerprints alone. The  FBI admits that there  is no connection between Mayfield's fingerprint and the fingerprint on the bomb. It is agreed that this belongs to a North African who was subsequently arrested in Spain, and the FBI claims that the fingerprint transmitted to them was so bad that it was confused with Mayfield's. Therefore, the political associations of those fifteen people should be a random distribution of the  political associations of  all Americans. Muslims are in fact  only about  two percent of the population, and even remotely  militant one must only be a fraction of a percent. 

My reaction is essentially a statistical one-- the FBI claims to have drawn four aces, ten times running. There's just no way you can do that without fuzzing the cards. I don't  know what the Arizona  method is for dealing with someone who draws four aces, ten times running...

I don't know if you have  ever heard of the "Six Degrees of Separation" experiment, carried out some years ago. People were given a letter to a random stranger, and told to deliver it, without simply sending it through the mail. Rather, they were to pick someone they knew who  might be in a  better  position to  hand-deliver the letter, in effect to create a chain of acquaintances leading from themselves to the unknown stranger. It turned out that people could do it in an average of six hops. If you know a hundred people, and they know a hundred, and so on, that works out to a  trillion  people, and the  world  population is only six billion. So that kind of indirect association means nothing.  For example, anyone who is Irish-American has probably talked to someone who has talked to someone who has talked to someone who is an IRA fundraiser. 

My guess is that this Perouz Sedaghaty character, if he were raising money, would have gone about  it like any other businessman. He would have collected directories from mosques, made a list of the people who seemed likely to have money (doctors, lawyers, executives, etc.), and called them up. He would have tried to avoid stating his business to a secretary or wife, because no one really wants to talk to a fundraiser. However, a fundraiser makes his living by getting people, in essence, to pay him to go away. It's easy to see how  some one might have been  induced to write  his telephone number down.

The computer does not know the difference between scientific fact and political prejudice. The most obvious explanation for Mayfield's arrest is that someone stirred a "watchlist" into the computer, either intentionally or inadvertently. I think the FBI were trying too hard to find a man who couldn't be found because he didn't exist. There was no American who had  left his fingerprint on the bomb, because a North African had done so. In looking for this  mythical American, someone faked something. Effectively, the FBI's computer was feeding their ideas back at them, and it was a kind of cyborg Folie a Deux. Such an event has been postulated in science fiction, but this is probably the first real  live sighting.
I suppose Karin Immergut and Beth Anne Steele will eventually be brought to trial for kidnapping, or conspiracy thereto. I don't know if there's a statute of limitations, but if so, it must be a long one. Sooner or later, a Democratic U. S. Attorney will  open charges. This will raise an  interesting legal question: does the McNaughton Rule defense apply to a delusional relationship with a computer?

Parenthetically, Daniel Pipes' comments reveal a complete failure to grasp the distinction between dependent  and independent events. 
You are effectively demanding that the mathematics of probability and statistics be abolished in support of your political agenda.


[Bill Heuisler (1/5/2006) acknowledged my debating skills, and wanted to know if I was a professional lawyer. He also made the assumption that I was postulating some sort of conspiracy. So I clarified.]

(01/05/2006 10:50 AM)

Well,  I should state that I  am an engineer and historian, not a lawyer. And, no, I don't postulate planning in the FBI. I  postulate recklessness, or rather, a "culture of recklessness." Too many people egging each other on,  too many people afraid to say no to the boss. Oh, and as I said, there  is this weirdly dangerous relationship of  people who rely on complex machines, but don't understand them well enough to exercise sound human judgment. As to intent, no robber who shoots a shopkeeper in the course of a stickup really intends to do so. The law imputes intent,  as a means of taking the more dangerous crooks out of circulation. 

There really does not seem to be even the  beginning of a case against  Mayfield. There  is no evidence that he ever contemplated any action save purely legal and constitutional protest, and considerable  evidence that he is a stable citizen, who does things like teaching English to  immigrants.  I was not able to find a  copy of the original Immergut memorandum, but [only] as reproduced by Pipes. [I]t seems a thoroughly  disingenuous document,  evasive about details.

If the tables were turned, how would you defend yourself against an accusation that you intended to organize a systematic massacre of all Muslim Americans?  Turn about is fair play, and the standards of evidence you set for other people will inevitably be used against you. Give the Devil benefit of law for your own sake!

[Bill Heuisler's response was in effect "that there are no sanctuaries for theenemies of Christ." I responded that:]

(01/05/2006 06:01 PM)

Well,  put this in perspective. The noted trial lawyer ("barrister") F. Lee Bailey made a specialty of representing "gruesome murder suspects," typically  men who were accused of strangling their wives, eg.  Sam Shepherd. Then there's Otto Schily  in Germany. Back during the  1970's he defended the various members of the Baader-Meinhof gang, and was subject to official harassment. He eventually became the Interior  Minister  of Germany  (equivalent to our Attorney General). By definition, defense lawyers have dubious clients.  In the last analysis,  if you allow lawyers to be harassed for taking clients the authorities disapprove of, it effective works out to denying the constitutionally guaranteed right to counsel.

[Bill Heuisler (1/7/2006" claimed that: "We're at war. Of course there have been declarations. Three in fact." I responded that:]

(01/08/2006 06:28 AM)
The term "state of war," or "declaration of war," has a very specific and technical meaning. It is what the lawyers call a "term of art."  Various and sundry laws are formulated in terms of the state of war. Provisions become operative  when a declared state of war exists. Congress, made up almost exclusively of lawyers, declined to declare war, but merely  made various lesser resolutions. This was their way of saying that they did not permit the courts to be closed down, that they did not authorize the introduction of conscription, nor the introduction of price controls or rationing, etc. You may have seen the picture of Sewell Avery, president of Montgomery Ward, being forcibly carried out of the company offices by two soldiers,  upon the orders of President Roosevelt during the  Second World War. Avery had refused to deal with  trade unions. Roosevelt had reached a concordat with the trade  unions, guaranteeing against strikes in wartime industries, and Avery's actions represented a threat to the bargain.

http://www.suntimes.com/photos/galleries/realchicago/1940s/20.html
http://www.suntimes.com/photos/galleries/realchicago/1940s/index.html
http://www.suntimes.com/photos/galleries/realchicago/index.html

Avery does look rather smug, don't you think. The GI's, on the other hand,  seem a bit nervous.


[Postscript: The government eventually settled litigation with Mayfield by paying him a reported two million dollars.]



  Index   Home