Semantic Compositions

On a non-trivial substring recognition problem

.coCourtesy of SC's favorite information technology scandal rag, The Register, comes news that Yahoo! has banned the registration of usernames containing the string "allah", presumably in order to avoid offending Muslims. This was discovered not by someone wishing to express an opinion on the Islamic faith, but rather by an individual bearing the last name "Callahan". Seeing as this demonstrates that Yahoo's filter must be case-insensitive, for the remainder of the post, SC will continue to spell out the string in all-lowercase, as no reference to anything but a particular string is intended.

At first, your host's instinct was, "That's an idiotic oversight; couldn't they have just done something to ban patterns like "allah2000" or "myopinionaboutallah", where it clearly isn't part of another word?" They've got the data on what sorts of names people are trying to register; surely they could construct a regular expression that would catch the bulk of it, and maybe use a dictionary to filter out the rest?

But actually constructing such a filter is harder in practice than it sounds. Suppose we assumed (to take just one of many possibilities) that "allah" at the end of a proposed usermane is probably it, and wrote an expression along the lines of "*allah". This would shut down "ILikeChallah", an expression of opinion on a wholly unrelated matter. Of course, this is why you would then need to follow up the expression with a dictionary, to try to match increasingly-long substrings to existing words, which would verify that the occurrence of "allah" was a false positive. But this would be complicated by the potential of having to expand the substring in both directions as part of a search; after all, "LivinInTallahassee" ought to pass as well, not to mention the name that started this whole line of inquiry (Callahan).

And then, even if the two stages of the filter concluded that you probably did have a match to the Muslim deity in particular, not just an Irishman or a Jewish baked good, there would have to be some additional processing to determine if the reference was actually offensive. Which would then require some way to parse out the rest of the proposed username, taking into account the tendency of people to abbreviate words in coming up with said usernames. If someone came up with "flwrallah" as a username, is that "follower" or "flower"? One can use finite-state machines to come up with potential word lists from abbreviations pretty easily -- your host used this program for exactly such an application as part of a statistical NLP course -- but suddenly, we're talking about a fairly complicated piece of software just to check for all of the possible words that could be in someone's e-mail address.

In that light, it's not too hard to understand why Yahoo! might! have! just! decided! to! ban! the! string! outright! (to use a frequent Register affectation when writing about the company). But it's actually fairly clear that they didn't go through any such engineering analysis to consider the trade-offs. You see, we know not only that this was their policy as of two days ago, but how the story ends. Rather than just admit that they took the straightfoward approach to dealing with a problem they didn't want to have (namely, complaints from irate Muslims), they instead claim to have "recently re-evaluated" the issue, and concluded that it's simply gone away, which strikes SC as rather implausible in light of current headlines (think cartoons).

Of course, they could be telling the truth insofar as they've got the data on what usernames people have actually tried to register (and when they've done so), and SC doesn't. However, if we take their word, then we're also supposed to believe that they independently came to a decision to unblock the string right around the time they got a complaint that they should have easily anticipated -- and thus had a clear policy to deal with -- if they devoted even as much effort to this issue as SC has in writing this post.

A translation/search engine for Wikipedia

Every now and then, SC browses Google News for language-related stories, and comes up with something like this, a search engine for Wikis. The interesting linguistics hook here is that Qwika, the search engine in question, performs machine translation of Wikipedia articles into a variety of (admittedly fairly common) languages.

So how does it do? Well, here's the first sentence of the Wikipedia article on semantics:

In the main, semantics (from the Greek and in greek letters "σημαντικός" or in latin letters semantikos, or "significant meaning," derived from sema, sign) is the study of meaning, in some sense of that term.

And here's their Spanish translation:

En la cañería, semántica (de Greek semantikos, o el "significado significativo," derivó de sema, la muestra) es el estudio de meaning, en un cierto sentido de ese término.

El estudio de meaning? It's not bad for gisting purposes, though, which is about as good as you could hope for with large-scale machine translation, especially with so many subject areas to cover. Considering what a useful resource Wikipedia is in English, this looks like a pretty nice start to getting some of that work available in other languages.

Weird Al's song of love

Your host was worried he might have a hard time coming up with an idea for Valentine's Day this year. Not for Mrs. SC, mind you -- those plans are always taken care of well in advance ([where by "well", he means multiple whole seconds prior to the order cutoff on the 13th -- FTD]). No, Valentine's Day demands something special for SC's long-suffering readers. Fortunately, a post by Neal Whitman got your host thinking about Weird Al Yankovic, which in turn has inspired today's effort. To celebrate Valentine's Day, we're going to go through a rigorous analysis of the metaphors in Weird Al's unique ode to love, One More Minute.

The song begins with a straightforward accounting of recent events:

Well, I heard that you're leavin'
Gonna leave me far behind
'Cause you found a brand new lover
You decided that I'm not your kind

So I pulled your name out of my Rolodex
And I tore all your pictures in two
And I burned down the malt shop where we used to go
Just because it reminds me of you.

Poor Al seems to be going to fairly great lengths to suffer for this relationship. As a California resident, his act of arson is subject to between three and eight years in jail. But there's no metaphor at issue yet.

That's right, you ain't gonna see me cryin'
I'm glad that you found somebody new
'Cause I'd rather spend eternity eating shards of broken glass
Than spend one more minute with you

Now things get more interesting. Obviously, Al isn't going to live forever, but we can reasonably interpret "eternity" to mean "the rest of his life". So how much glass would Al eat in his lifetime?

The average American lifespan is around 78 years, and the song came out when Al was 26, so we're looking at 52 years of eating glass. Typical human stomachs range in capacity from 1.5 to 4 liters. If we assume that Al eats three square meals of broken glass each day, and that each meal comes in at the low end of capacity, that's 4.5 liters per day, or 256,230 liters of glass in his lifetime. How many bottles would Al have to smash to get that glass? Let's figure that a 16-oz. bottle (roughly 0.57 liters) can be shattered into glass representing about half as much volume, or 0.285 liters, meaning that Al will need 899,052 typical soda bottles to meet his commitment. Let's hope he can buy them wholesale!

I guess I might seem kinda bitter
You got me feelin' down in the dumps
'Cause I'm stranded all alone in the Gas Station of Love
And I have to use the self-service pumps

If Al is stranded at the self-service pumps, assuming he drives a car with about 20 mpg fuel economy, and does about 15,000 miles per year, with an 18 gallon gas tank (assuming he fills up every time he hits 16 gallons), he'll fill up 46.9 (call it 47) times. Assuming he saves about $0.30/gallon by not using the full-service pumps, he'll pocket $281. So this doesn't work out badly for him -- actually, it's just about perfect, as we'll see in a moment.

Oh, so honey, let me help you with that suitcase
You ain't gonna break my heart in two
'Cause I'd rather get a hundred thousand paper cuts on my face
Than spend one more minute with you

This sounds pretty horrible, but it's a lot less of a sacrifice on Al's part than you might think at first glance. How much blood do we get out of a paper cut? This proved hard to estimate, but one site promises that you'll get more blood from a papercut than from a particular diabetic test kit. Since it draws about 0.3 microliters, we'll guess a paper cut comes in at about 0.5 microliters. If Al gets 100,000 paper cuts on his face, and each one bleeds out .5 microliters, he'll bleed about .05 liters. A standard unit of blood for transfusion is 495 mL, so since Al has only lost about 1/10 the amount of blood needed for a transfusion, the odds are good he'll live through this one.

I'd rather rip out my intestines with a fork
Than watch you going out with other men
I'd rather slam my fingers in a door
Again and again and again and again and again

As for ripping out his intestines with a fork, the procedure for seppuku involves a sword called a wakizashi, which has a blade about 20 inches long, rougly three times the length of typical forks (as measured by several found in SC's kitchen). The tines of said forks are about 1 1/2 inches long, so if Al tries to commit seppuku with a fork -- even one specially serrated along the tines for the occasion -- he's in a lot of trouble. This may be his intent -- since Al doesn't mention having a second, he's probably trying to commit jumonji-giri (see the article on seppuku), which is an enhanced version that's even more painful than the standard version.

I'd rather have my blood sucked out by leeches
Shove an ice pick under a toenail or two

The leech issue is complicated, because your host found two fairly divergent measurements of how much blood a leech can suck out. An article from the Journal of Rehabilitation Research and Development indicates that a leech can suck about 4.7 mL over 3 hours, or roughly 1.6 mL/hour. But another article, admittedly a more popular-science treatment of leeches in medicine, suggests that the correct figure is more like 15-30  mL over a period ranging from 20 minutes to an hour. We'll go with an estimate of 15 mL/hour for the remaining calculations.

At an average of 14-18 pints of blood per human adult, it would take one leech (working on 16 pints, or 7.5 liters), about 4687 hours to completely drain Al. Assuming Al doesn't eat or drink anything, since loss of 30% of blood volume can lead to irreversible shock, it will take the leech about 1406 hours, or 58 days, to finish Al off. He'd die of thirst in just 15 days, though, so 4 leeches would be needed for the job. To do it in one day, he'd need 60 leeches, and at $4.75 a pop, that would mean Al would need to add just $4 to the money he saved from using the self-service pumps at the Gas Station of Love. So he shouldn't act on this for at least a year.

Actually, once the leech gorges itself (according to the pop-science article), it's done eating for up to 18 months while it digests, slowly. Assuming reasonably quick changing of leeches by a wholly unscrupulous attending physician, that 30% of blood volume works out to 2,250 mL, or 150 leeches. Which means he'll really need to spend about 2 1/2 years at the gas station to save up the money for his irrational act, during which time he might reconsider.

I'd rather clean all the bathrooms in Grand Central Station with my tongue
Than spend one more minute with you

This always sounded to SC like the second-most impressive claim in the song (trailing only the grand finale), but the truth appears to be far more depressing -- it's actually not nearly that much of a sacrifice. According to this article on toilets at Grand Central Terminal (station is apparently a misnomer), there are only 32 ladies' stalls, and 6 men's stalls, in the public area, plus 4 more of each in private areas which can be hired out for receptions according to a FAQ at the GCT website. Another caveat is in order, however, regarding the sourcing of what follows: while the author of the cited article, Donna Porstner, really does write for a paper called the Stamford Advocate, which seems like a reasonable source (being a Connecticut newspaper), the provenance of the article is a newsgroup -- the article doesn't turn up in a search on the paper's website. Having said that, the only bathroom listings shown in the GCT website's directory are on the lower dining concourse, just like the article said was the case prior to the remodeling job it discusses, so it's entirely possible that these figures are accurate. Admittedly, 46 toilets is 46 more than SC is willing to clean with his own tongue, but the lyrics make it sound like we're talking about several hundred.

Yes, I'd rather jump naked on a huge pile of thumbtacks
Or stick my nostrils together with Krazy Glue
I'd rather dive into a swimming pool filled with double-edged razor blades
Than spend one more minute with you

How many thumbtacks could get stuck in Al if he jumped into said pile? Well, Al is 6' 0" (from his website), and 170 pounds, so we'll figure that he's got a waist size of about 32", which allows us to get a rough estimate of his surface area. Treating Al as a cylinder, which has a surface area of 2*pi*r*^2 + 2*pi*r*h, and dropping the first term since he doesn't really end in neat circles, we get a rough surface area of 7234 square inches. We'll take off about 20% to deal with deviations from non-cylindricality, for about 5780 square inches. Thumbtacks come in a range of sizes, from 5/16" to 1/2" heads. We'll go with the 5/16" size, assume no overlap, and also assume that he gets completely covered with tacks from rolling around after the initial impact, which means that at 0.31 square inches per thumbtack, Al will be on the receiving end of 18,645 thumbtacks, which works out to $658 worth using the nice nickel-plated ones which top the list in the last link. That's at $3.53 per box, but he probably qualifies for an even bigger volume discount than the one cited, since we're talking about 187 boxes here.

We can also investigate the number of razors it would take to fill Al's hypothetical swimming pool. A standard Olympic-sized swimming pool runs 50 meters by 25 meters by 2 meters, or 2500 cubic meters. Browsing the web for straight razor dimensions yields mostly dimensions specified in terms of cases, not blades, but reasonable figures seem to be about six inches long, two inches wide, and a half-inch thick, which means that at .0034 cubic feet per blade, or .00009 cubic meters, it would take 25.4 million blades to fill the pool.

As for the ultimate act of self-destruction, Al's pledge that:

I'd rather rip my heart right out of my rib cage with my bare hands
And then throw it on the floor and stomp on it 'til I die

we can't really speculate on how many stomps Al would get, but we know that you can rip a man's heart out of their rib cage and have it continue to beat, as documented in Indiana Jones and the Temple of Doom, where the act was performed by Mola Ram.

As Weird Al has helped us to see, being miserable for love can be quite expensive, and fairly painful to boot. So make sure to give your significant others a hug today, and thank them for saving you several thousand dollars in leeches, thumbtacks, and razor blades. Then take them to Charlie Palmer's at Grand Central Terminal to thank them for saving you from having to clean the place. Happy Valentine's Day!

The worst sentence of 2006

Your host rarely singles out bad writing for mention solely on the grounds of the sheer awfulness of the text. But this line from French philosophical rock-star Bernard-Henri Levy in today's Wall Street Journal (from the free part of the site) is so jaw-droppingly redundant, so repetitious, so intending to provoke by making the ears bleed from hearing the same thing over again...ugh, this is such an atrocity, it actually can contaminate your mind. Without further ado:

And it is hard not to link this provocation, the deliberate circulation of these cartoons, the quasi-home-delivery of a Danish paper that no one could have guessed had so many readers in the Muslim world, it is hard not to link this self-inflicted blasphemy, this calculated offense (calculated, mind you, by the organizers of the distribution of the cartoons), it is hard not to link this blasphemy to a new planetary configuration, itself determined by three recent and major events.

Technically, the English version of this sentence was perpetrated by Messr. Levy's translator, according to the biographical note at the bottom of the article. But SC has no reason for supposing it to be anything less than a faithful rendering of the original text. Especially since so much of the rest of the article is written this way, only with periods where about half of those commas are. Also since Garrison Keillor tore into Levy's recent book on America for some of the same reasons. SC has seen a number of articles over the last few years indicating that this man is enormously popular in France. Why?

Mighty silly

Apropos of recent posts both here and at Phonoloblog on the new TV network to be known as "The CW" comes a disturbing name change to SC's favorite radio station.

First, a little background. For almost 20 years, the preeminent sports talk station in Southern California was XTRA 690, a San Diego-based station with a Mexican broadcast license. This was done both because getting to start the call sign with an "X" made branding them as "eXtra Sports" easier, and because they could have a much high-powered signal ("from Baja to the Canadian Rockies" as one of their hosts used to start off every show). Both due to the higher-powered signal and the opportunity to have a name dripping with machismo, the other accepted name for the station was "The Mighty 690".

Some time ago, the station was bought by Clear Channel, which decided that what Southern California really needed was a station completely and totally focused on Los Angeles, and what better way to do that than by moving the broadcasters to L.A., purging most of the staff associated with San Diego-area sports, and changing the signal to broadcasts on 570 AM? (Not all of this happened at once.) The original owner of 690 must have gotten an abnormally lenient noncompete agreement, because not long after they did this, he got another Mexican broadcast license with the rights to use 1090 AM, and opened up shop as "The Mighty 1090", with the call sign XPRS. This resulted in an additional slogan, "Express sports talk for San Diego", a play on XPRS. (Disclaimer: if you click on the 1090 link, you'll see the logo of SC's employer. This is due to an advertising deal, and SC's company has no ownership stake in the operation of 1090 AM.)

Sometime last week, but news to SC until just yesterday (he didn't listen to the radio for a few days), they acquired a second license in order to simulcast on an FM station, XHBCE. To commemorate the event, they now call themselves "XX Sports Radio", and the approved tagline is "The Mighty Double X" (two Xs, get it?).

SC doesn't claim that the same heuristics which appear to govern TV station names govern radio stations as well, but a similar observation nonetheless applies.

Call signs often are intended to spell out something and serve as the excuse for claiming that the station's name is really some other word that they'd like to be associated with. XTRA meant "Extra", XPRS is supposed to be read  "Express", and perusing a list of LA-area FM stations gets us plenty more such examples. KOST, "Coast; KSPC, "The Space"; KTWV, "The Wave"; KXOL, "El Sol"; KFRG, "K-Frog"; KRTH, "K-Earth"; KBUE, "Que Buena"; KPWR, "Power 106"; and KROQ, which goes by "K-Rock". Calling a station "Double X" is at most evocative of Dos Equis beer. Actually, it's so evocative not only in literal meaning, but in the similarity of the logos, that SC would expect someone at Cervezas Mexicanas of White Plains, NY (that's actually the importer) to be calling the station any day now. And not to chat about San Diego State athletics.

SC would try to comment on whether or not "Double X" works phonologically in the same way that "Mighty 1090" did, or in the manner of FM stations with jingles incoporating both their call sign/derived-from-call-sign name and number, but the truth is that he listens to very little on the radio, and isn't familiar with enough examples to derive a pattern from. Sorry, Prof. B.

Can Google be export controlled?

SC was absolutely thrilled over the weekend to receive an e-mail from Thomas Lipscomb, whose actions in response to Soviet censorship of a book fair were discussed in a recent post about Google's Chinese business dealings. Mr. Lipscomb wished to direct your host to a follow-up piece he wrote for the magazine Human Events, advocating the updating of export control laws to penalize American companies that actively aid and abet tyrannies. It's a provocative idea, and we'll offer just the outline of response because the subject is so complicated.

Your host has some minor experience with export control laws, having prepared an ontology of controlled dual-use materials for an ARDA project a few years ago (in case anybody is worrying about security issues, the project was Unclassified, and also not Sensitive, so it's OK to have mentioned that). The current export control list contains sections for  NBCR (nuclear, biological, chemical and radiological) materials, computers, encryption software, sensors, lasers, and other sorts of hardware or production tools -- in other words, it's essentially the same as it was when SC worked with it 3 years ago. It doesn't really address services, where the actual products/technologies remain under the control of the developer, which is the case with search engines, web-based e-mail, and other offerings from Google/Yahoo!/Microsoft/etc. This of course only demonstrates that Mr. Lipscomb is on target when he writes:

Since those export controls were originally imposed back in smokestack America, the United States economy changed from an industrial economy to an information economy. Today more than half of America’s GDP comes from information technology and information products. Not surprisingly, the Federal Government is often the last to understand changes like this. It can be hard to see that in today’s world information technology might be more dangerous than military equipment.

Regardless of the specific merits of updating export control for dealing with some particular government's policies and behaviors, application service providers have been around for a few years already, and to still have no approach to that business model is insane. Section 730.5 of the Export Administration Regulations addresses this to the extent that it recognizes "export" can mean things other than the physical transfer of hardware or media, and therefore covers things like trying to make nuclear blast simulations a service offering rather than delivered code, but there is still no recognition of the technologies which are best suited to a service model. Filtering technologies provided as a service are functionally equivalent to denial-of-service attacks, in that they have the same outcome of blocking information from being exchanged between legitimate users. The fact that no content provider's servers are being directly vandalized by the filter does not change the outcome. Export controls would come into play with any hardware device designed to provide denial-of-service capabilities to the Chinese government; why mandated filtering should be treated differently is a matter that at least deserves scrutiny.

Mr. Lipscomb also calls for Congress to recall the Jackson-Vanik amendment, which linked trade status to emigration rights, and act in that spirit. The principle here strikes SC as basically sound -- a foreign policy with no actual implementation mechanisms isn't much of a policy at all -- but also hard to sell. Aside from the fact that the specific compliance mechanism provided by Jackson-Vanik is no longer available, there are material differences between the present situation and the Cold War. Whereas trade with the Soviet Union was limited, U.S. trade with China is pervasive. Whereas a specific grievance could be readily targeted beforehand -- the emigration of dissidents being restricted -- it's harder to see how restrictions on information could be dealt with. Chinese crackdowns on dissident journalists, bloggers, or other media sources, are often couched in terms of "state secrets", and while this is often plainly untrue in any sense that would be recognized in the U.S., a dialogue that consisted of "We have a right to protect our secrets! No you don't! Yes, we do!" would bog down with few results. Trying to gain popular support for restrictions on our biggest source of cheap goods is best undertaken when there is no opportunity to obfuscate the source of conflict.

There are all sorts of arguments about whether or not export control is even a good idea, the specific goals of present U.S. regulations notwithstanding, and SC won't pretend to consider them in any sort of detail here. However, as the actions of the American software industry come into increasingly sharp conflict with American foreign policy, it would be a grave mistake to simply ignore the issue. Thomas Lipscomb deserves our thanks for his efforts to keep this from happening.

Super Bowl XL pick

Since your host doesn't know how to quit when he's ahead, and also because having a topic to segment off game predictions eems like a good idea, we'll use today to introduce the SC Sportswire, which will never actually be a pay service.

The consensus pick among ESPN's commentators (not to say experts) is the Steelers, with 15 of 24 reporters picking them. The betting lines are hovering around a 4-point advantage to the Steelers. Sometimes, the consensus is wrong, as when USC was favored so heavily to win the college football championship a few weeks ago, but this time, SC thinks it's basically correct. The AFC is a much more competitive conference, and Seattle's 13-3 record doesn't impress SC as much as Pittsburgh's 11-5. The Seahawks aren't as soft as their detractors think they are, but in SC's opinion, the Steelers already have faced and beaten three playoff teams as good or better than the Seahawks. So it's the Steelers to win, with SC taking them to cover any line short of 7 points.

Just garbage

Today is Super Bowl Extra Large -- er, XL -- and so to join in the hype in the only way a linguist knows how, today SC is going to write about trash talk.

It has long been alleged that by saying insulting things to opposing players, it's possible to "get inside their heads", and make them distracted. Deion Sanders was alleged to be a master of this in both his college and pro football careers. The Denver Broncos' Shannon Sharpe claimed to be able to draw opponents into committing personal fouls (the maximum penalties in football) with nothing more than his mouth. Defensive tackle Warren Sapp believes trash is so integral to his reputation that his website is named "QB Killa", parodied by the Philadelphia Daily News (when Sapp played for a Philly rival) as "Twinkie Killa" (link is to Google's cached version, as the article seems to no longer be in their public website).

Another claim is that trash talk provides motivation to the recipient. Coaches supposedly live in fear of the dreaded "bulletin board quote" that some careless player on their team will drop, which the other coach will then use to fire up their opponent. The Pittsburgh Post-Gazette's Mark Madden wrote several years ago that "No athlete gets more motivated in response to trash talk. If he/she does, that's a pretty sad athlete." On the other hand, Michael Jordan allegedly used to be so fired up by it that he would tell himself opposing players had trash-talked him even if they hadn't (SC is certain he read an interview with Jordan discussing this some years ago, in either GQ or Esquire, but has been utterly unable to find such a reference online).

Both of these claims were on display over the past week, as the Seattle Seahawks Jerramy Stevens made a comment about spoiling Pittsburgh Steelers' running back Jerome Bettis' final game; this in turn drew a fiery reaction from noted Steelers trash-talker (and occasional linebacker) Joey Porter. Porter brags about his research efforts in preparing trash talk for each game, and claims responsibility for an NFL rule mandating at least 10 yards of separation between teams during pre-game warmups. Porter seems to believe he really will be inside at least Jerramy Stevens' head:

"When a guy says something who lines up in front of me on every play, I have to like that," Porter said. "He has to see me. There's no way he can hide from me. We have to meet -- over and over and over. ... I'll remind him every time I put him on his back."

On the other hand, the coaches are more concerned about the motivational factor:

Steelers coach Bill Cowher subsequently cautioned Porter and other players about making comments that opposing teams might use for motivation. Cowher also called Porter's officiating comments "ridiculous," and Porter subsequently had little to say before Pittsburgh's 34-17 victory over Denver in the AFC championship game.

...

Both coaches also tried to play down any effect on the game by Porter's constant talking. Seattle coach Mike Holmgren talked to his team about not creating any more distractions.

"I used to lecture my children, too. As much as I'd like to think they're perfect, they're not," Holmgren said.

It seems rather unlikely to SC that such talk has both effects on the same individuals, although he's prepared to believe that for some players, hearing taunts causes self-doubt, while for others it works as motivation. On balance, though, he's prepared to give the benefit of the doubt to the coaches, who know that there aren't too many professional athletes who got there by thinking they might not be among the best.

Me want cookie -- now! OR ELSE!

Arts and Letters Daily catches SC up today on a super-important Wall Street Journal article (no subscription required; it's on the free "Opinion Journal" site) from Wednesday that he otherwise would have missed. Even if you don't agree after reading this post, at least click on it to have a look at an illustration that had SC in tears (it's about halfway down the page). The opening sentence is a trip:

While the extreme branch of heavy-metal music known as death metal is defined in part by often-vile lyrics about violence, catastrophic destruction, nihilism, anarchy and paranoia, its singing style is associated with a beloved goggle-eyed, fuzzy blue puppet.

This is a setup for the revelation that the death metal community has managed to one-up all amusing terminological gestures of linguists, ever. Forget all the Haj Ross coinages like "pied piping", "sluicing", and "squish". Forget Geoff Pullum's "Duke of York gambit". For you see:

Death-metal vocalizing is also known as Cookie Monster singing, if not in tribute to, at least in acknowledgment of, the "Sesame Street" puppet that blurts in a guttural growl, his words discharged so rapidly that they tend to collide with each other. [emphasis added -- SC]

No responsible linguist would simply accept an analysis without some supporting evidence, so your self-sacrificing host has ruined his Amazon recommendations for the next month in order to investigate some suggested examples from the article. Napalm Death's album "Scum" does indeed sound like "a virtual Cookie Monster tribute"; check out the second half of this clip from their song "Multinational Corporations" (the first half is instrumental only). The title song from Suffocation's "Effigy of the Forgotten" is accurately described as "an especially malevolent Cookie Monster", although a more precise description might be "Cookie Monster aided by a vocoder". On the phonological principle that "once is an exception, twice is a rule", and also the fact that much more of this would make SC's ears bleed, we'll accept this as adequate evidence that "Cookie Monster singing" is an apt description of the vocal style in question.

Now, if only one of these bands would take on the true abomination, Cookie Monster's new theme song: "A Cookie is a Sometimes Food".

Google, China, and the root of all evil

Today, Google shares have fallen some 9% after reporting earnings of $1.22 per fully diluted share -- a slight surprise to the Wall Street analyst community, which had been looking for $1.76. Of mildly cold comfort to the analyst community -- and to your host, who was net long -- was that excluding options expenses and charitable donations, the figure would have been $1.54, a mere 12.5% shortfall instead of the actual 31% that it represented.

In apparently separate news, Google came under heavy criticism last week for its decision to comply with Chinese government censorship in order to launch a local edition of their otherwise-dominant search engine. When this latter story broke, it was on the heels of news that Google had just refused to comply with a U.S. Department of Justice request for records -- sans personally-identifying data -- of pornography-related search terms. At the time, these stories were generally perceived to illustrate hypocrisy, or at least inconsistency, in Google's claims to "do no evil", a story exposed in these pages previously here and here.  Radagast wrote about it then as well, and wondered what your host might have to say about it. The earnings release does much, however, to clear up the confusion about Google's business practices.

Let's start with two observations about Google's business in general:

1) Google is a private company which provides an enormously, mind-bubblingly useful service for no cost to end users. The terms on which they provide that free service are completely up to them. So long as Google's paying customers believe that Google's management of their service provides them a better return on investment than any other online advertising medium, there is no reason for them to listen to anyone else about how to run their service. And this is how it should be so long as "anyone else" doesn't own a stake in the company, buy advertising services from them, or work for it. People who attempt to define themselves in as "stakeholders" when they are not in one of these roles lack the credibility of someone whose livelihood really does depend on the company's operation.

2) It is an utter fantasy to expect that businessmen making rational calculations about profit-maximizing behavior will voluntarily cede markets to other businessmen, especially when they know that no net gain in ethical behavior will occur as a result. Google's staying out of the Chinese market does not mean that Baidu will behave in a more honorable manner.

This brings us to the fundamental tension in Google's business practices, a tendency for management to act as though they are aware of these issues, while behaving publicly as though they are not. The leadership of the company -- inventors Larry Page and Sergey Brin, and serial technology company CEO Eric Schmidt -- is comprised of two individuals with a disturbingly immature, college-student-sloganeering view of the world, and a man who has worked hard to be as distrusted on Wall Street as he is respected within the technical environments he works in. Don't take SC's word for it, take Jim Cramer's (subscription required):

Google is one of the worst-run companies when it comes to telling its story and dealing with Wall Street. That's in part because of Eric Schmidt, who is not well-liked on Wall Street.

It is hard to be certain which of these three individuals bears the most responsibility for Google's general refusal to disclose material information except at quarterly earnings calls, but suffice it to say that Google's management team has established an arrogant attitude towards the outside public, shareholders and disinterested observers alike, whereby they will frequently make sophomoric moralistic pronouncements about their business, while providing little solid basis for estimating their ability to run it in accordance with their principles. To give just one example of the immature outlook of Messrs. Page and Brin, consider their recent purchase of a jet for the company (admittedly, funded through their private wealth, and not from corporate coffers). Rather than simply saying, "Look, executives like to one-up each other, and buying a Boeing 767 trumps all those other billionaires with their little Gulfstreams", Larry Page justifies it in terms of occasional charitable endeavors, and says of his purchase "I think that can only be good for the world." Google's leadership simply cannot perceive that some of their acts are merely morally neutral, the same as buying a newspaper, because they are so certain of the basic goodness of their intentions.

Having made these points, let's turn to a recent Editor and Publisher editorial by Thomas Lipscomb on the subject of Google's deal with the Chinese government:

Some years ago, as the Soviet Union was headed for its demise, a Moscow Book Fair was announced and publishers in the United States and throughout the world flocked to gain access to a huge potential new market. The Soviets promised an open market at the Fair to display what publishers felt were their best books most suited to the market. But as soon as the Fair opened, Soviet police moved in on publishers and confiscated books they felt might "feed agitation."

Other publishers, fearing this kind of action, had already self-censored the books they displayed or quickly removed them on the spot. Times Books, the general book publisher owned by the New York Times Company, immediately withdrew from the Fair, arguing that it was difficult to maintain First Amendment standards in the United States while conceding them elsewhere. A lively debate ensued, and the Moscow Book Fair was seriously diminished as a marketplace thereafter.

Reserved to the biographical note at the end of the piece is the fact that Mr. Lipscomb was the head of Times Books who made the decision to withdraw.

Mr. Lipscomb's position -- that sometimes, defending the conditions that make your business possible outweighs short-term profit goals -- is the crux of the current controversy. Google's behavior in China does not wholly ignore that idea -- when they provide censored searches, they indicate that censorship has occurred through a notice on the results pages provided to the user. Google's executives also recognize that as they introduce services which result in ever-greater quantities of personal data being stored on their servers, their business is dependent on assuring user confidentiality, a point made ably by John Battelle. As a result of this, they are not offering Gmail or Blogger service locally to Chinese users. These steps contrast admirably with those taken by Yahoo! and Microsoft in launching services, as SC has noted before. And even those comapnies may be regretting the totality of their knuckling under to Communist dictatorship. Just today, Microsoft has changed their own blog censorship policies in response to Western outrage over their behavior in a case involving a blogger working for the New York Times.

The problem is that Google's management is aware of their responsibility to shareholders to grow the company, aware of the problems created by censoring their search results, and yet utterly unwilling to deal with the public relations disaster that occurs by their inability to reconcile hard decisions with their public pronouncements.  Google may actually have taken the most honorable approach of any company seeking to do business in China -- without any Blogger information to turn over to the government, we won't be hearing about journalists jailed because the company compromised their privacy rights, like we did with Yahoo!. They may have taken the most principled position in the U.S. pornography case -- while SC believes the government is acting within their rights in the present circumstance, the precedent of turning over records in the absence of clear wiretapping-type laws that deal with search data is a bad one to set.

Having considered all these conflicting pressures, we can at last be clear about the true purpose of the controversies over Google's apparently inconsistent behavior in the last two weeks. Messrs. Brin, Page, and Schmidt knew well before yesterday that their earnings report would fall shockingly short of expectations, even taking into account the various charitable deductions and option expenses (the latter of which many people are inexplicably opposed to calculating corporate earnings). Therefore, they needed to demonstrate clearly to investors that they were working hard on not being left behind in the Chinese market, a fact evidenced by China and "international markets" being mentioned in ratings reiterations from stock analysts today. The simultaneous announcements of expansion into China with compromised data and an unwillingness to compromise data at home were meant to indicate that while Google is doing everything they can to grow their revenue base, they do not intend to compromise their existing sources of revenue. Pornography is a significant part of their American revenue base (although no exact figures are available, some research estimates say that Google is responsible for over half of all search referrals, and somewhere around 25 percent of all searches -- not just on Google -- are for pornogaphy; SC notes that he has not found a primary source for this latter figure, just several references to it), and so "do no evil" aside, investors can be absolutely certain that Google is moving to protect that revenue. Because Google's executives will not simply acknowledge that their business is a business first, and a charity/public defender second, these announcements come off as the products of a singularly capricious arrogance, and we are left to sift through the fine print to see that perhaps they are doing less evil than things appear at first glance.

But then SC thinks of this experiment suggested by Jonah Goldberg, contrasting image search results for "Tiananmen" from Google's U.S. and Chinese sites, and wonders if even the controls they have put in place really represent change.

At the time of publication, SC was net long Google. Under no circumstances does the information in this column represent a recommendation to buy or sell stocks. SC cannot provide investment advice or recommendations, and the preceding article shall not be construed as a solicitation for the purchase or sale of any mentioned security.