Steve Reads

A friend yesterday tried to order Boston Red Sox postseason tickets at the moment they went on sale (6 p.m.), so I helped out. The lines were flooded and the Sox’s web server was jammed. Hitting redial over and over again only revealed a number of failure modes that I was hitting all at once: sometimes an Operator Voice told me that the circuits were full, sometimes I got a busy signal, etc. As for the website, it had an elaborate series of choke points built in to stop traffic before it reached the actual commerce server. The net result is that we spent a couple hours redialing and refreshing pages (using several simultaneously open IE windows). By the time we were done, the tickets had sold out anyway.

This points out an annoying design that we run into constantly. If you call a tech-help line, or you’re on the phone with any kind of operator, you always have to sit on hold, and you shouldn’t. You should be able to leave your phone number, which the operator will then call back as soon as your number reaches the front of the queue. (If you want to stay on hold, of course you should have that option.) This would be no more elaborate than a pager with smart queueing. It would free up the phone lines, and would free you to go about your business.

Ditto the website. You should be able to put yourself in queue for tickets on the Sox’s site, or in queue for any busy resource (say, a page that’s been Slashdotted). HTTP seems a bad protocol for this, given that it seems geared to real-time interaction. Email seems perfect for this sort of thing, actually: you send an email, and so do 200,000 of your closest friends. Your mail server waits in line to send, and keeps trying to connect until it succeeds. When it does, the receiving mail server puts your message in queue for processing behind your 200,000 friends’ email messages. This needs to be tweaked somewhat, because it gives the advantage to sending email servers which back off the most slowly. But you get the idea.

Basically, I think we all have much better things to do than wait in lines, and we have the technology to get out of lines. So why are we still stuck in the old model?

Does anyone know of a way to create collapsible lists without using JavaScript? By a collapsible list, I mean a hierarchical list whose subtrees you can hide until you want to see what’s in them. I can only find JavaScripty ways of doing this, which seem suboptimal; a lot of people turn off JavaScript in their browsers. If there were a way to do this using only CSS and HTML, that’d be ideal.

The reason I ask is that ye olde list of links on the left is looking rather ungainly. A little grepping suggests that it contains 95 links. I think it’s time to break that down into expandable-and-collapsible sublists. Even better would be a cookie that allows you to choose which you want: the fully expanded, 95-link list, or the collapsible JavaScripty one. Soon, my pretties; soon.

Even within the set of all JavaScript interpretations of a collapsible list, what I linked above isn’t very good. I shouldn’t have to change the HTML to a JavaScript-only mode; I should instead be able to wrap JavaScript around an HTML list, and have the script figure out the hierarchy. If I’ve already defined a hierarchy within the HTML, I shouldn’t have to define it again within the JavaScript. Can anyone suggest a clever JavaScript way to do this?

I started writing a bit about my evolving views on copyright law, and it turned rather long (about 1,100 words). I’m interested in what people think about it. It slowly morphed into some thoughts on what the law means to those who use the Internet, which I think could make for an interesting publishable essay.

My friend Seth made a great point today. I was in the car with him, proposing something like what I’ve proposed before: namely, that there ought to be some distributed way to categorize web pages. I like the earlier version better, actually; I wrote that we should be able to attach metadata to every link, so that we can say “the destination of this link is a page about philosophy.” Then search engines could decide how much they trust my links about philosophy. In this way, a meta-tagged link could provide a more refined route than the way we link today; today when we link, Google assumes that we’re casting a vote for the trustworthiness of the destination. In the future, I think it would be better if a search engine knew our trustworthiness on a particular subject.

I gave a variation on that today. The idea is that I should be able to tag my own web page with something like the Dewey Decimal System: “this is a page about mathematics, specifically about topology, specifically about the Jordan Curve Theorem.” This is no more a controversial idea than the idea of metatags themselves.

Seth’s response was just great: the Dewey Decimal system only exists because we didn’t have the technology to find out the contents of a book in any automated way. We didn’t have computers that could look at the contents of a lot of books, figure out their subjects solely from the words inside, and continue to learn as they processed more books. The Dewey Decimal System, in other words, makes up for human deficiencies that computers may very well eliminate.

What’s more, Seth points out that a Dewey-like categorization for a web page might actually be harmful. The subjects that a web page deals with should be revealed by the search-engine query. We can’t expect web page authors to know all the uses to which their pages might be put, so we shouldn’t artifically constrain search engines ahead of time by pre-categorizing our pages. If it’s a page about topology, in other words, we should let the search engine figure that out. If the search engine can’t figure that out, then there may be something wrong with our page. One final way to say this is that the contents ought to dictate the categories. A fixed taxonomy might end up being more harmful.

I might propose one modification: there’s nothing wrong with metatags, so long as the search engine doesn’t rely on them too heavily. It might be very helpful for me to label my page as relating to topology. If the search engine finds that the labels I assign to my own pages are consistent with what the search engine itself discovers, then perhaps the search engine ought to label me a “good librarian” — i.e., reliable as a labeler. The search engine might weight the metatags from reliable labelers more highly than it rates metatags from unreliable labelers. On the other hand, if the search engine tends to disagree with my labeling, it might check the scoring with people who link to me. That is, if I say that it’s a page about topology, and the search engine disagrees, but many users who link to that page say that it’s a page about topology, then the search engine should consider changing its labeling. I can imagine a very clever, intricate, and powerful back-and-forth amongst users, authors and search engines. The net result for the web could only be positive.

Speaking of the end-to-end argument (see immediately below), I notice a company linked off BoingBoing called CloudShield. The idea seems to be that we’ll break the end-to-end nature of the Internet to increase efficiency. It’s worth repeating: The Net works as well as it does, and has grown as much as it has, because it is a dumb network. If the network only knows that it carries bits — not phone bits, not HTTP bits, just bits — then anyone who can write code that sends bits can send those bits over the Net. A dumb network allows for indefinite growth.

There are some reasonable questions to ask, such as whether it’s possible to guarantee quality of service over an end-to-end network. E.g., if a doctor is performing surgery from a distance over the Net, you really want to make sure that his video feed doesn’t get dropped. This would seem to suggest that the routers along the way should assign higher priority to the doctor’s bits than to porn bits, which in turn suggests that there needs to be some way to tag bits as “doctors” or “porn.”

Then again, maybe not. Let’s suppose that a quality-of-service guarantee is not possible without breaking the end-to-end model. In this case, perhaps we should decide that the tradeoff isn’t worth making: that end-to-end is too valuable a principle to ditch just because of a few specific applications. I’d need to think more about this. But in general, I’m highly suspicious of any technology that gives the network more intelligence.

I saw an apparently unrelated article on The Volokh Conspiracy today, describing a new service called PlanetLab that apparently runs atop TCP/IP yet still filters content. It says that “the network itself will detect and crush rogue data packets before they get a chance to spread to your office or home.” The article is short on technical details, so I can only guess how it would tag data as “rogue.” I’m doubtful. And again, I’d like to see this intelligence pushed further out in the network. I don’t want my routers deciding what’s spam, or deciding what to do with something tagged as spam; I want to make that decision, or at least leave it up to my company.

There’s also the risk of moral hazard if the network tries to attack data on its own. We should be encouraging security from the ground up — starting at the CPU and moving up through the OS. If badly designed operating systems keep allowing viruses to spread, we should move away from those operating systems. If old programming languages make it easier to overflow buffers, we should find new programming languages. We could drastically cut down on the number of viruses and insecure systems by reducing our dependence on Microsoft Windows. If doing this would cut back on the “rogue packets,” and would have other benefits besides, then I don’t want my network covering up the real source of the problem.

As with most things, Adam has done a good job converting me to the cause of spam filters – Spamassassin and the like — over techniques like email obfuscation. Several weeks ago, I mentioned that obfuscating emails wasn’t such a bad idea, and that I wasn’t clear on why John Gilmore hated the idea so much. Discussions with Adam have brought me around to the idea that pushing spam filtration all the way to the user’s desktop — rather than requiring list admins to obfuscate their lists’ email addresses – is the right approach. It’s basically an extension of the end-to-end argument (the link’s not working right now, and there’s no Google cache of it available, but I know that it’s worked before; hopefully reed.com will be back up soon). End-to-end says that all the intelligence in the network ought to be pushed out to the user, and not vested at any intermediate point (like a router). Make the network as dumb as possible.

More relevantly, perhaps, user-end spam filtration is easy and cheap, and highly effective. If there’s an easy way for users to filter spam, and the user-end approach decentralizes the filtration power as far as possible, then we should definitely favor user-end filtration.

Adam has kept me informed of a great debate about this topic on the linux-elitists mailing list, particularly some very insightful commentary by Karsten Self. It’s definitely worth a read. Karsten ended up getting booted off a mailing list, because — as far as I can tell — the admins got firmly trounced during the debate. I think he’s worth rewarding with a slightly higher PageRank.

The people at Volokh were right about the Times’s coverage of Bush’s speech. What’s the point of this digression?

He also spoke at length about the need to halt sex tourism and the trafficking in people, including many young women, for the sex trade. The United States was cracking down on practices that Mr. Bush said were akin to slavery, and other countries should do the same. The cause is popular among some conservative Christians in the United States.

Can we address Bush’s claims on the merits, please? What’s this crap about its being popular among conservative Christians? Is it a good idea, or not? I’m inclined to believe that the bit about conservative Christians was the Times’s way of making Bush look even worse in liberals’ eyes. If you read Bush’s actual speech, the stuff about the sex trade seems pretty innocuous. He refers to a bill (the PROTECT Act) that has been — as far as I can tell – passed by the Senate. I can’t find the actual voting record, but it wouldn’t surprise me a bit if it was in fact “popular among” a great many senators, conservative or otherwise. That’s what a bill’s pasage would seem to imply.

Granted, Annan’s and Chirac’s comments did seem directly aimed at the United States and the neo-cons’ policy of pre-emption. They seemed harsh at times. But Annan’s words, at least, seem aimed at finding a compromise that simultaneously addresses the U.S.’s legitimate concerns about terrorism, and respects the rule of international law.

How is it possible to develop your own opinion about a subject when the Times tells you that “Mr. Bush’s somber address to the General Assembly drew a single 20-second round of applause at its conclusion”? Wouldn’t it be better for the world to know the content of what he said, rather than what the U.N. thought about it? Both are important angles for a newspaper article to take, but I’d suggest that the former is a lot more important for a democracy.

As I’ve said before, I think newspapers are primarily valuable as a way to convey raw information to us. We can make up our own minds, and listen to people who are more well-informed than mere journalists if we want to get someone else’s take on an event.

And no, I’ve not yet read what the Volokh Conspiracists had to say about the speeches. I merely saw that they recommended we read the original texts, went to read them, read the newspaper articles, and made up my own opinions. Now it’s time to see what some intelligent bloggers have to say about it.

I meant to post the other day about Frank Rich’s brilliant, dry, scathing attack on Mel Gibson (my cache), inspired by Gibson’s violent denunciation of Rich:

Then Gibson expressed his feelings about Rich. “I want to kill him,” he said. “I want his intestines on a stick . . .  . I want to kill his dog.” — The New Yorker, Sept. 15

Rich’s column is a great read, as are all the columns I’ve read by him in recent months. The place where media and mass culture intersect with politics is fertile terrain for Rich, and he never disappoints. I highly recommend everything he’s done.

In the second of today’s inadvertent Commenting On Commentators series, I’d like to point you to Roger Ebert’s review of Cold Creek Manor (my cache). It simultaneously explains why I avoid seeing a lot of movies, and why I love Ebert’s writing.

The latest EPIC Alert (a weekly newsletter from the Electronic Privacy Information Center) is really alarming. The slippery slope just gets more slippery.

David Weinberger makes a good point in a Wired article (my cache): it is unethical to build your regulatory systems such that there is no leeway. We don’t mechanically hand down a death sentence for a murder; we realize that there are mitigating circumstances that make some murders worse than others. The more we rely on technology for the enforcement of policy — whether that technology is Digital Rights Management to copy-protect CDs, or it’s content filtering on library computers — the more we move away from the real, flexible world we want into a world of automata. We have reason to be afraid.

I’ve been blessed today with a number of potential Quotes Of The Day, but I might have to give the nod to Ed Felten, famed Princeton professor of computer science and opponent of much Digital Rights Management, for this quote:

As a parent, I had to chuckle on hearing the American movie industry complain about the distribution of inappropriate sexual content to kids. But then again the whole room seemed at times to be an irony-free zone.

The context is that everyone at the hearing where Felten testified — on the subject of DRM — solemnized about the evils of porn on peer-to-peer networks. Among the solemn was the famed charlatan Jack Valenti of the Motion Picture Association of America. See, porn on P2P is the New Big Bad Thing. Back in 1995, I remember there was a big kerfuffle over porn on the Net, because of a baseless study by a then-undergrad at CMU. That article, and quite a lot of other hysteria, led to the now-infamous Communications Decency Act, which the Supreme Court roundly rejected in 1997. Now, because the recording industries are trying to kill anything that might possibly be used to carry unauthorized MP3s and movies, P2P is the New Vehicle For Porn.

Someone always wants to regulate the Net. Since quoting Larry Lessig never hurt anyone, here’s something that always seems relevant:

So here’s what I thought we knew: I thought we knew the Internet was going to shake things up, to change things. To mess things up. To change the way things have been. I thought we knew that, we lawyers, and we had committed ourselves to watching and waiting and letting things shake out. I thought we knew that. I thought we knew that if we stepped in now to regulate, we’d screw it up.

The Net is still evolving. If we let the network develop and see where we end up — if we leave everything open, drop the DRM, drop the insistence on strict control — I have no doubt that it’ll run off and create a world of unimaginable variety and freedom. Let the old media control it, however, and don’t be surprised if the Net looks like a television set that allows shopping.

Eugene Volokh’s paper on the slippery-slope argument (namely, “We can’t enact policy A, because if we do then B will be more likely, and B is really atrocious”) is a wonderful piece of writing. It tries to establish all the mechanisms by which the slippery-slope argument could be rational.

My two favorites — because I find them quite novel — involve lowered costs for implementation and enactment. Volokh uses gun registration as an example: why should anyone object to gun registration? Is claiming the slippery slope here — that gun registration might lead to gun seizure — rational? Surely in a democracy, we can choose to stop the process as far down the slope as we want, right? Volokh gives two plausible ways in which the slippery slope might work, quite independently of the people’s wishes. First, if people didn’t register their guns, the government would have a hard time seizing them; they’d have to send someone around to every house to check for a gun, which would use manpower that they don’t have. But second, without registration it would certainly be unconstitutional to send someone around to everyone’s house to search for a gun; that’s a “general search,” and the Fourth Amendment explicitly forbids it; that’s one of the main reasons that the Fourth Amendment exists. However, if people registered their guns, then the police would have probable cause to search the home of a registered gun owner, and could probably attain a search warrant. Adding registration makes an unconstitutional practice constitutional, and also makes it easier to enforce if it does become law.

Volokh’s paper is filled with insights like these. It’s quite low on legalese. I’d recommend it highly.

I also recommend Yochai Benkler’s “Free As The Air To Common Use” as a quite deep paper on “intellectual property.” The scare quotes are necessary in this case, because Benkler’s whole point is that our standard way of looking at information is flawed. If you view information as property, minus some special cases where it goes into the public domain, then Benkler says your background assumptions are off. He goes on to make a strong case that government laws which pull information out of the public domain and “enclose” it have violated the First Amendment. Highly recommended reading, though more legalistic than Volokh. It’s worthwhile at least for its restatement of the idea (central to the “legal realists,” if I understand correctly) that “intellectual property” – like all property — is a legal fiction created in order to further certain policy goals. When we recognize that there’s nothing “natural” about intellectual property, a lot of principles fall more easily.

Maybe it’s time for law school. Good thing I signed up for the LSAT just the other day.

If the point of the Internet is its massive decentralization, isn’t Verisign’s recent “hijacking” of the domain-name system (via Jeff; my cache) proof that we’ve failed? Indeed, isn’t the domain name system itself proof that a large part of the Internet’s functionality is centralized? Isn’t our reliance on the client-server model, whose relatively low-grade effects include “Slashdotting,” also a sign of such things? (Slashdotting isn’t just for Slashdot. Remember how hard it was to find a working news site on September 11, 2001? That’s Slashdotting hard at work.) Isn’t the Internet’s reliance on a backbones-and-regional-networks model also a danger? That hierarchical model of the Internet gives you events like Cable and Wireless’s decision to change its peering policy (my cache) and thereby cut off a good many Internet users from the network. If the phrase “single point of failure” means anything, it means “the ability of a single company to create islands in the network.”

What do we do about this? I’m not a network planner, so I don’t know if there’s any practical alternative to the backbone model. In the other cases — DNS, client/server and so on — I wonder whether a FreeNetish approach would work. A failsafe DNS would involve your directly connecting to a large number of machines to which you know a route, and using these machines as a bootstrap into finding routes to every other machine on the Net. DNS would no longer rely on a relatively small number of servers. Instead it would rely on all machines acting as routers. It would be slow — maybe deathly slow — but it would work.

As for the Verisign thing, I don’t know. Certainly massive public scrutiny turned them around quickly. In the future, I wonder whether another company’s cost-benefit analysis will make it give up such a disgraceful move so quickly.

From a New York Times Magazine interview (my cache) with the creator of an edition of the New Testament that is translated to be more appealing to teenaged girls, including “sidebars on fashion and romance”:

We use the New Century Version [of the Bible]. It translates the Bible thought for thought instead of word for word. The King James translation reads at a 12th-grade reading level. Most people in our country today do not read at that level. The New Century Version reads at a fifth-and-a-half-grade reading level, which is about the average where people can comprehend.

If that’s true, it profoundly depresses me.

I saw Sofia Coppola’s film Lost In Translation with my friend Bijoyini tonight. If you would prefer to be a blank slate before viewing it, please stop reading right here.

It’s a well-done movie on a lot of levels, not least of which is Bill Murray’s acting. He has the persistent hangdog look of a downtrodden man, which serves him well throughout the film. He’s the family man, away on a trip to Tokyo to promote some brand of whisky, and clearly trying to get away from a lumbering home life. He’s 30 years past his acting prime. He meets Scarlett Johansson, who is just impossibly attractive. (The opening shot of the film — lasting probably 20 seconds — is a still photo of Johansson’s butt. Some of us already have trouble sleeping, Sofia; would you cut that out?) They bond over their mutual loneliness. Romance, of a sort, ensues.

I thought that Murray’s and Johansson’s chemistry held the movie together; Bijoyini didn’t think it was such good chemistry, but then she was asleep for about 10 minutes of it during the crucial Murray And Johansson Fight In The Iraq War scene, during which we realize the truth of the old adage that there are no atheists in foxholes. I think Bijoyini would have had a much different impression of the film had she seen the war scene.

Bijoyini and I agreed, though, that Murray was pretty astounding as one of the two central characters in the film. His face can convey a great deal without doing very much, which is exactly what the role needed; he wants to fall in love with Johansson, but he has kids and a wife — the latter whom he no longer seems to love — back at home. He would seem to be on the verge of crying throughout the film if he had the energy left, but that energy seems to have left his character 20 years back. What’s left is a persistent dull ache that Murray conveys flawlessly. If you’ve ever closed the door after a date, and said goodbye to your date without kissing him or her, but really wished you had, and continued to curse yourself for the next 10 minutes — or just sat there sighing — then you will know where Murray’s character is coming from.

The cinematography and the directing, on the other hand, leave a fair bit to be desired. Often it serves a director’s needs to cut away from a long sequence of sad events, if only to give the audience some respite. Lost In Translation has a fair number of those, which is necessary when a lonely character like Murray’s is central to the film. But at other times, Coppola’s mise en scene and pacing don’t seem to advance the story at all, nor advance our understanding of the characters. The stop-and-go pacing only makes it harder for us to establish a relationship with these two people. We grow to care about them because of Murray’s and Johansson’s stellar acting, not because of Coppola’s scene arrangement.

Also, the decision to place the film in Japan is largely pointless, it seemed to me. It was an excuse to make repeated jokes about the Japanese difficulty pronouncing l’s andr’s in English words, and about how small the Japanese are. Oh, and about how bizarre their television programs are. If I understood the film correctly, Coppola placed it in Japan to ratchet up the feeling of isolation that the characters feel, thereby driving them into one another’s arms. Japan also seems to be the destination for washed-up lounge singers and movie stars, which makes everyone in the film seem a little more shabby. I guess the location served a purpose, but I wish Coppola hadn’t chosen to get so many cheap laughs.

Well worth seeing if only for the acting. Bill Murray continues to prove that he can do anything. Johansson is intelligent, and her character is watchful, in a way that few actresses her age can duplicate. And she’s only 19 years old; may god have mercy on my soul.

P.S.: I happen to agree completely with Ebert. He probably saw the same flaws in it that I did, but they mattered less to him.

I found one newspaper’s idea of the 100 best books ever (“as determined from a vote by 100 noted writers from 54 countries as released by the Norwegian Book Clubs”). I’ve cached it as well, for posterity. I’ve read 16 1/3 of the books on that list, if we count Inferno as 1/3 of The Divine Comedy. Looks like I have my work cut out for me.

Note that I’m especially pleased to see Midnight’s Children on that list. I thought it was a masterpiece, but I didn’t know it had such acclaim among the literati (or at least this subset of the literati).

For those times, late into the night, when no birds are chirping and you’re all alone with your conscience, and it feels like you’re the only one who cares, and that caring itself has become passé, and that people view indifference not just as a way of life, but as a necessary bodily function like breathing; and when you get that horrible clenching sensation in your stomach that maybe you made a mistake a long way back by starting to care; but you’re still not far enough along that you can tell everyone to go fuck themselves and retreat into your drink; on these occasions there is Bob Dylan, passionately and effortlessly singing “When The Ship Comes In” into your headphones.

Oh the time will come up
When the winds will stop
And the breeze will cease to be breathin’.
Like the stillness in the wind
‘Fore the hurricane begins,
The hour when the ship comes in.

Oh the seas will split
And the ship will hit
And the sands on the shoreline will be shaking.
Then the tide will sound
And the wind will pound
And the morning will be breaking.

Oh the fishes will laugh
As they swim out of the path
And the seagulls they’ll be smiling.
And the rocks on the sand
Will proudly stand,
The hour that the ship comes in.

And the words that are used
For to get the ship confused
Will not be understood as they’re spoken.
For the chains of the sea
Will have busted in the night
And be buried at the bottom of the ocean.

A song will lift
As the mainsail shifts
And the boat drifts on to the shoreline.
And the sun will respect
Every face on the deck,
The hour that the ship comes in.

Then the sands will roll
Out a carpet of gold
For your weary toes to be a-touchin’.
And the ship’s wise men
Will remind you once again
That the whole wide world is watchin’.

Oh the foes will rise
With the sleep still in their eyes
And they’ll jerk from their beds and think they’re dreamin’.
But they’ll pinch themselves and squeal
And they’ll know that it’s for real,
The hour when the ship comes in.

Then they’ll raise their hands,
Sayin’ we’ll meet all your demands,
But we’ll shout from the bow your days are numbered.
And like Pharaoh’s tribe,
They’ll be drownded in the tide,
And like Goliath, they’ll be conquered.

Our president now wants expanded subpoena powers (my cache), wants to make it harder for those accused of terrorism to be released on bail, and so on. The PATRIOT Act, in his view, doesn’t go far enough.

I’m really afraid of what will happen if he gets these expanded powers. My Congressional delegation is already strongly Democrat, so I don’t know what else I can do to get my voice heard. If anyone has any ideas about how to overcome a sense of powerlessness, let me know.

And now, another installment of our favorite game, Spot The Fallacy About Copyright Law! It’s our most popular game show, along with Spot The Fallacy About What Liberals Believe (currently hosted by Ann Coulter).

Anyone who doesn’t see precisely where Jim Maule misunderstands copyright is welcome to contact me, and I’ll gladly lay out the fallacies on this weblog. I’m never sure whether anyone cares about such issues; if you do, let me know.