In high school I once posted an optimistic rationalist argument calling for replacing more accidental things and institutions with deliberate and designed ones.
My friend Eric Tapley responded:
My only response is this: Butter was found by accident. Is it not good, then?
Our brief in the Verizon case is very, very good.
Did you look closely through the list of parties signing on to it? that's right, it's the National Grange!
While we're on the subject of very fine briefs, there's also the matter of the Aimster brief. Nice work (and thanks, Rick).
In connection with the 2.1 release, I had a party and got a good turnout, not to mention coverage on Crummy. It's too bad that Leonard couldn't remember Riana's pirate jokes. I do want to supplement the record with regard to Leonard's pirate joke -- at the time, I alluded to the fact that a nearly equivalent non-pirate joke was made in a book. I have found the precise citation!
Once in Russia, in a physics exam, the professor wrote the equationE = hν
and asked a student:
'What is ν?' 'Planck's constant.' 'And h?' 'The length of the plank.'
(From Physicists Continue to Laugh, MIR Publishing House, Moscow, 1968, translated from the Russian by Mrs. Lorraine T. Kapitanoff, quoted in R. L. Weber, compiler, A Random Walk in Science, p. 152)
Weber notes: "Astonishingly, this is translated directly from the Russian."
I'm grateful to everybody who turned out to celebrate.
The LNX-BBC got a favorable review in LJ online, but the reviewer spelled the name of the distribution wrong. It's something like doing a review of "RedHat Linux".
About two weeks later, the actual physical discs showed up, and they are really very pretty. If you would like one (and you know me), just ask me. If you would like one and you belong to a LUG, join the lnx-bbc mailing list and wait until we announce what LUGs should do. Otherwise, you can get one through the EFF Store.
When you're in the a store, surrounded by modern technological marvels (as, say, at the Sony Store at the Metreon), and you see the pretty things, and the remarkable things, like a useful portable radio receiver for under $10, remember:
[...] and it is this same joint stock of technology that gives to the modern world's tangible assets whatever use and value they have. Tangible assets, considered simply as material objects, are inert, transient and trivial, compared with the abiding efficiency of that living structure of technology that has created them and continues to turn them to account.(Thorstein Veblen, Absentee Ownership: The Case of America, p. 65)
Then you might contemplate how much greater an accomplishment Sony's litigation victory over Universal City Studios turned out to be than any single one of its engineering accomplishments in the past twenty years.
I went with Praveen to see the play Partition, which is about the life of Srinivasa Ramanujan. The play was beautifully produced by the Aurora Theatre (a smaller theater in the arts district in Berkeley), and the classicist character Billington was just wonderful (approximately: "I daresay that by the twenty-first century, a literary education will consist only of Latin!"). I thought the play was harsh toward G. H. Hardy. It presented him in what seemed to be a very extreme way.
You would get the impression from the play that Ramanujan and Hardy hardly managed to work together at all. In fact, they co-authored several significant papers. They were productive together and they did not fail to do great mathematical work. I'm afraid the play oversimplified their relationship in search of a depiction of one particular aspect of it.
In the play, Ramanujan complains that "this is not about East versus West", yet Partition often seems to stress cultural differences (and interpersonal conflict) at the expense of depicting useful collaboration and exchange. The play almost seems to claim that Ramanujan had to die before Hardy could have a real relationship or collaboration with anyone -- a very strong conclusion about an actual person and one I'm not sure can be justified by history.
In general, the play and the production were brilliant and highly amusing. It could be improved (at a cost of an extra hour or so) by adding some more nuance to Hardy's and Ramanujan's relationship to show that it wasn't completely dysfunctional.
Fermat was the most comical character in Partition, and his presence and incessant discussion of his Last Theorem reminded me that Hofstadter made a joke where you invert Fermat's Last Theorem to say that
na = nb + nc
has no integer solutions for n > 2. (It appears on pp. 334-5 of Gödel, Escher, Bach, where it is attributed to "Lierre de Fourmi" and captioned "Johant Sebastiant [Fermant]'s Well-Tested Conjecture".)
This is easy to show (how remarkably much easier than the trivially typographically different "an = bn + cn"!).
Suppose that n > 2 and na = nb + nc. Then, if b = c, we have na = 2nb, quod fieri nequit.
So if we continue, assuming without loss of generality that b > c, we can divide by nc and get
na-c = nb-c + 1
or
na-c - nb-c = 1
or
nb-c(na-b-1) = 1
or consequently
nb-c = 1
and taking the logarithm of both sides
b - c = 0
contrary to our original assumption that b > c.
I sent this to Sumana as an "in Soviet Russia" joke, but afterward I wanted to see if it generalizes so that we can say that a power of n can be decomposed into n powers of n, but not into any other number of powers of n. This is true for n=2, as above. In fact the same technique works to prove it in general.
Suppose that
na = nb + nc + nd + ne + ... nz
Now (without even making an assumption about whether b, c, d, etc.,are distinct), we simply assume that a>=b, b>=c, etc. This is no loss of generality because any number of integers can be arranged into descending order. Now divide by nz:
na-z = nb-z + nc-z + nd-z + ... + ny-z + 1
and again
na-z - nb-z - nc-z - ... - ny-z = 1
Factor out ny-z:
ny-z(na-y - nb-y - nc-y - ... - nx-y - 1) = 1
So we conclude that y=z, and we also have
na-y - nb-y - nc-y - ... - nx-y = 1
and we can follow the same procedure by induction showing that x=y, w=x, v=w, u=v, t=u, etc. Thus, whenever a power of n is decomposed into any number of other powers of n, they are all equal powers. Thus, there must be exactly n of them.
A place called Mondo Gelato has opened up in Berkeley, right by the BART station. I've been there twice now, and it's very, very good.
I became mildly obsessed with the geeky project of spelling out words in the MD5 checksums of software releases. If MD5 checksums are really random (statistically uncorrelated with every bit of their input), then you would expect that 1/2 of all MD5 checksums end with the bit "0", 1/2 of those end with "00", 1/2 of those with "000", etc. Similarly, 1/(2n) of all MD5 checksums should end (or, if you prefer, begin) with any chosen n-bit sequence. Famously, you can write words in hexadecimal: dead, beef, feed, deaf, EFF. If you allow "0" for "o", you can write c0ffee. Many people say that the longest English word thus expressible is "acceded" (but others claim "fabaceae", a family of beans).
So if you have a way to partially reverse the MD5 hash, you ought to be able to make your software have a hash which starts with a chosen word. (Adam Back calls this a "partial hash collision" in his papers on hashcash. A "hash collision" is a case where two values hash to the same value; a "partial hash collision" is a case where two values hash to related values. Strictly speaking, Back's hashcash and the hashes I'm searching for are not really "collisions" because there is no original hash value with which the desired new hash value can "collide".)
A straightforward approach to this is to take a given existing file and append random bytes to it until the hash comes out the way you want. That is a kind of brute force search and doesn't rely on any knowledge about MD5. (If you want to square the number of operations required, you could try finding something which has the same first few bits in its MD5 hash and in its SHA1 hash -- I did a few experiments with that.) Adam Back mentions that in an ideal cryptographic hash, there is no way of generating collisions faster than brute force. Some other people have pointed out that, since MD5 is not ideal, there might be some known methods of going faster than brute force. You might find mathematical properties of the MD5 function which let you search a little more efficiently. (Maybe there is a correlation between one bit in the MD5 state some time before the conclusion of the MD5 calculation and a bit in the MD5 result?)
I wrote a Python program, and then ported it to C. I originally used Peter Deutsch's MD5 implementation, but Jef pointed out that OpenSSL contains a fast assembly implementation of MD5, so I changed the code slightly so it could link against OpenSSL.
The result of all of this is fun: all of the LNX-BBC builds since the evening May 29 now have checksums starting with "bbcbbc". A typical example is "bbcbbc2a348c82f5f0b907c7a7da4f5d". For official releases, we will try to put the version number in -- for example, LNX-BBC 2.1 ought to have started with "bbc210".
Zack is threatening to write a distributed collaborative partial hash collision client so that people on the Internet can donate computer time to search for how to modify LNX-BBC releases to have checksums starting with longer words. 48 bits is about the most I currently dare to hope for. I can find a 24-bit prefix in a few seconds on one computer!
If you maintain a free software project, why not get my code and spell some words in your next couple of releases' checksums? c0ffee, f00, c0de, maybe "abbe" in honor of my friend Abbey...
I got to go to Tu Lan and to the arcade with Sumana, who just started a new job at Salon and is performing that great ritual of leaving Berkeley for San Franisco. She seems optimistic about this change, and it was great to see her. At the arcade I played Bubble Bobble -- an old favorite of mine from when I had a Nintendo -- and air hockey. The Metreon's arcade has a "Retro Room" where you can play pre-2001 video games. I thought each one should have a sign above it indicating how many bits the microprocessor handles at once. Bubble Bobble would be "8". Air hockey would be "0".
My mom is putting on an international Virginia Woolf conference at Smith College this week. Congratulations, mom!
It looks like the conference will be a great time for all involved.
The Golden Gate Bridge had what I think was its 66th birthday on May 27, and, in keeping with my annual tradition, I walked across it. This year, Riana was my walking companion.
We stopped in the middle to sing "Happy Birthday" to the bridge. Riana observed that, even though nobody else was around, this might be considered a public performance.
The transcript the May 9 hearing at which I testified is now available. It has some errors. For example, the third word of my testimony is misstated. (I said "Register" and the reporter transcriber "Registrar".)
My testimony begins on page 141. One of my favorite moments:
MR. CARSON: Is it more likely than not that the FCC will issue a regulation requiring [the] use and recognition of the broadcast flag?MR. SCHOEN: So it's obviously -- there's never any way to predict what an agency is going to do while they're in the middle of a proceeding.
MR. CARSON: We'll prove that.
(Transcript, p. 159, line 2.)
I'm very glad I got to participate.
I went to court to hear arguments in Bowoto v. Chevrontexaco, in which I filed a declaration in opposition to the defendant's motion. The case is before Judge Susan Illston and could go to trial later this year; as with other cases in which I've been involved, it's had several years of pretrial motions. I got to meet several lawyers for the plaintiffs.
I also went to court (about a week later) to hear arguments in DVD Copy Control Assn. v. Andrew Bunner. It was my second time at the California Supreme Court; the first time was for an argument in a related matter (DVD Copy Control Assn. v. Matthew Pavlovich). Alex Macgillivray, whom I saw there, wrote about the hearing. California's Attorney General, Bill Lockyer, appeared and argued as an amicus for DVD CCA. His remarks mainly centered on the importance of trade secrecy to California businesses, and didn't really address the specific facts of the Bunner case. Lockyer's argument appeared to be that weakening trade secret law would harm businesses in California, and that California had an interest in preventing that from happening. David Greene of the First Amendment Project, arguing for Bunner, responded that there are no reported prior California cases parallel to this one (in which a trade secret claimant seeks a preliminary injunction against a third party republisher of information), so ruling for Bunner couldn't really weaken trade secret law.
My biggest frustration during the argument came when Robert Sugarman, arguing on behalf of DVD CCA, suggested that the Court should not pay much attention to Bunner's first amendment claims, since a large number of "creative" people from the entertainment industries had filed briefs in support of DVD CCA. Sugarman appeared to be suggesting that, if the "creative community" had no problem with the injunction, the injunction must not offend the first amendment. I call this the "we own the first amendment and you can't have any" argument; it is peculiarly common in cases in which copyright is implicated. If a defendant claims a first amendment interest, an entertainment plaintiff will retort that the first amendment is actually about protecting the expression interests of, say, movie studios. The subtext is that other speakers do not have free expression interests worth protecting.
Sugarman actually took a very peculiar position with respect to the first amendment, claiming that the first amendment was meant basically to protect political speech and debates about matters of public concern. Speech on other topics, he suggested, was not really what the first amendment was about, and speech calculated to produce a result was far away from that amendment's purpose. It can only be described as bizarre to hear entertainment lawyers, lawyers for movie studios, maintaining that the first amendment is really about protecting political speech and not necessarily other kinds of expression, that the first amendment is really very narrow. This is, of course, exactly what people seeking to censor sexually explicit or violent expression in the movies always say -- that the first amendment is supposed to protect expression of beliefs, and political and perhaps religious arguments, and pictures of naked people or pictures of people getting shot are neither of those. Why, aside from their duty to maximize profits, movie studios would contemplate applying the first amendment to depictions of sex but not to descriptions of algorithms is simply beyond me. Technical speech -- as an important subject of major political and religious conflict, and a major source of social change in Europe not long before the adoption of the first amendment, must have been more obviously "expressive" at that time than sexually explicit visual images, which law and custom continued actively to suppress for many years. Or are Copernicus and Galileo to receive lessened protection for their expression because they had the "functional" goal of allowing people to calculate the positions of the planets?
I find it offensive that works like those in Touretzky's Gallery of CSS Descramblers are implicitly called uncreative and unexpressive by Sugarman. They contain more artistic and intellectual originality (and, one might say, courage) than some recent studio products.
The California Supreme Court is very informal compared to other courts. Before the beginning of arguments, the clerk came out to chat with the members of the public sitting in the gallery. He started off by giving some of the history of the Court, and at the end asked whether anyone had any questions!
It was odd to see on successive days two EFF clients named Andrew Bunner and Andrew Bunnie (Huang). It was even odder to see each in the company of John Hoy, the president of DVD CCA. We invited Bunnie to the ARDG meeting to give a talk about hardware reverse engineering, and the following day Bunner's case was heard before the California Supreme Court. The parallelism, if only in their names, was eerie.
I was interviewed on The Linux Show, talking about technology mandates and SDMCA. I urged concerned listeners to go talk to someone offline about copyright issues instead of joining another Internet flamewar. Listen to the MP3 archive, if you want. (I was hoping there would be an Ogg archive, since there was an Ogg stream, but it looks like there's no such luck.)
I also got quoted in an IEEE Spectrum article, whose author used my comments to draw a conclusion different from the one I would have drawn. (The article claims that "new standards, like DVD-Audio and Super-Audio CD [...] will contain robust copy protection, putting the issue to rest." I doubt whether the issue will be "put to rest" so easily as that. The DVD-A and SACD business is is a disturbing example of a recent pattern, though.)
David Alpert came by from the East Coast and invited me to help run a puzzle hunt at UC Berkeley put on by friends of his. (This is the genre where you find a series of clues, and each clue gives you information about where to find the next clue.) When the hunt started to wind down, around 10:00 p.m., David and I decided to attempt to solve it for ourselves, so we grabbed the first clue and started puzzling. We finished a little after midnight, having had an unfair advantage on a couple of clues. It was great fun, and I think I ought to look for treasure hunts to try solving for real. It looks like they happen regularly in the Bay Area.
I saw Spellbound with Zack. The Spellbound we saw is the recent documentary Spellbound, not the classic movie Spellbound. This one is about students competing in the national spelling bee. It's really excellent.
When I recognized a word, I spelled along with the competitors during the documentary, and got quite a few words wrong. I was known as a good speller in elementary school, but it seems I'm practically no speller at all compared to these kids.
The parents are the surprise stars in this film, I thought -- their attitudes about what a spelling bee is and why it's important are really interesting. They, and the really energetic kid. (Surely every geeky boy played at being a robot -- but surely not a musical robot.)
Contact: Seth David Schoen
