Stephen Laniel’s Unspecified Bunker

I’m reading Thomas Kuhn’s book The Copernican Revolution: Planetary Astronomy in the Development of Western Thought right now. It’s really very good, helping me to remember that the things we believe are true now weren’t always so. It’s also helpful to remember that pre-Copernican astronomy was, in fact, quite accurate: navigators needed to know where to expect various stars to be at any time of the year, and the two-sphere model of the universe (the earth a fixed, unmoving sphere in the middle, and the stars stuck to the outer shell of a large, far-away sphere) worked exceedingly well at that for a long time. I’ve not yet reached the point in the book where the two-sphere model actually gets cast into doubt, but I assume it has something to do with the retrograde positions of the planets.

Kuhn makes a point of differentiating between the data (the points of light that we see in the night sky), and the model (e.g., that we’re looking at an outer sphere). What strikes me, though, is that sometimes model becomes data. It is, for instance, no longer in doubt what the moon is made of, or how it orbits: we’ve actually set our feet on it. And given how many planets we’ve visited, their positions at any given moment in time are also not in doubt. We also have no reason to believe that the earth is flat: plenty of people have gone around the world. Certain models are simply wrong now, and certain models have gone on to become raw data. Philosophical doubt will only take you so far: once astronauts came back from the moon with samples of rock, it would take the most absurd skeptic to doubt the moon’s position and composition. (We could, I suppose, dither over the reliability of testimony. At the very least, that will disappear when we have tourists going to the moon.)

Kuhn also writes, on the subject of how hard it is to believe that the earth actually moves (rotates):

The idea that the earth moves seems initially equally absurd. Our senses tell us all we know of motion, and they indicate no motion for the earth. Until it is reëducated, common sense tells us that, if the earth is in motion, then the air, clouds, birds, and other objects not attached to the earth must be left behind. A man jumping would descend to earth far from the point where his leap began, for the earth would move beneath him while he was in the air. Rocks and trees, cows and men must be hurled from a rotating earth as a stone flies from a rotating sling. Since none of these effects is seen, the earth is at rest. Observation and reason have combined to prove it.

But by analogy, at least, it doesn’t seem hard to plant the seeds of doubt that a moving earth is possible. If I stand on a rolling platform that someone in front of me pulls, and throw a ball in the air or jump up and down, the ball will land back in my hand and I’ll land safely back on the platform. So why should it be hard to believe that objects moving on the earth behave the same way?

Under Linux, we get used to having the device node /dev/cdrom pointing to our CD-ROM drive. Then programs like CD rippers will know where to look to find new CDs.

Well, for some reason Ubuntu Hoary got rid of /dev/cdrom. The solution is simple, but why did they mess up this one little thing? If anyone can find a good explanation for the error, let me know.

David Wells and Curt Schilling are both on the disabled list. Closer Keith Foulke is really blowing it recently. And now three members of the Sox have been suspended — including manager Terry Francona (“Franconer” if you’re Sox color commentator Jerry Remy), pitcher Bronson Arroyo (who’s our star pitcher in the absence of Schilling and Wells), and right fielder Trot Nixon. I think we’re kind of fucked for the next few weeks.

I’ve just installed the Markdown plugin for Blosxom, and I’m curious to see it at work.

First of all, it’s supposed to convert double newlines to new-paragraph marks correctly, and not convert single newlines to new paras. Let’s see if it does that.

Then it’s supposed to convert asterisks and underlines to italics (specifically the <em> tag), and double-asterisks and double underscores to bold.

I think it does other crazy stuff, but those are the shortcuts that I’ll be using a lot.

If you happen to have a blog on this site — I’m looking at you here, Josh and Sarah — this will probably come in handy for you.

Speaking of which, if you look at Carrier’s blog, you’ll notice that she dumped in a bunch of links without paragraph marks between them. Somehow I’d like Markdown to know that the single linebreak between each link should be turned into a <br/>.

Or perhaps I should just install Markdown on her blog as well, and ask her to use double newlines between links.

 . . . Sweet! Check out how this post looks before Blosxom converts it to HTML. It does what you’d expect it to do, for instance, when you include text inside of backquotes to indicate that the backquoted text is code — it converts any HTML characters in there to their escaped versions. So for instance <br/> becomes <br/>. That’s just neat.

It also handles blockquotes intelligently: just precede the lines containing blockquotes with the standard email quote character >:

Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

The plugin has a great philosophy: your Markdowned text should look just fine even before it gets turned into HTML. In particular, except for the hyperlink stuff, Markdowned text looks just like an email. Smart. Very, very smart.

Thanks for playing along while I think out loud.

P.S.: Oh yeah. Because of one little irritant in Markdown that is super-useful 99% of the time, every post now has a new-paragraph mark between the post title and the first line of the post itself. Hence everyone viewing this site through an RSS reader will now see that every one of my posts has been updated. Sorry about that.

P.P.S: At least for non-RSS readers of this blog, the extra space between the post’s title and body is now gone. This bit of CSS does the trick:

.storybody > p:first-child { margin-top:0; padding-top:0; }

Here .storybody is the block within which the posts themselves are wrapped, and .storybody > p:first-child is the first “p” child-element of the <div class="storybody">. I’m not sure if all browsers support :first-child; IE6 probably doesn’t support it (it’s lousy in its treatment of CSS pseudo-classes generally), but maybe the IE7 Javascript code will fix that.

I’m going to make a big generalization here, just to toss it out and see if anyone can poke holes in it. My claim is this: one of the reasons liberals are losing the PR battle recently (which is to say, at least since Gingrich) is that we’re too committed to the idea that if people just learn the facts, they’ll come to our side.

So we’ve spent a great deal of time showing — see?! The evidence is right there in front of your face! — that there were no WMDs in Iraq, that the procedure laid out in Florida led unequivocally to the conclusion that Terri Schiavo would have wanted to die, that Bush’s Social Security plans don’t add up, and so forth. Liberal bloggers, in particular, spend endless quantities of time debunking facts coming out of the right wing’s spin machine.

But none of that matters in the least. The truth is not an especially good criterion for political success; I wonder whether liberals have learned this yet. Three guarantors of political success that Republicans understand phenomenally well are 1) manipulating the media, 2) starting the debate on their own terms, and 3) claiming moral standing ahead of factual standing.

They don’t always win with these techniques (the Schiavo thing seems to have backfired on them, and the DeLay battle doesn’t seem to have a moral high ground on which to stand), but they work pretty damn well. Americans, for better or worse, didn’t care much about the demonstrable fact that the war in Iraq was a war of choice fought on weak pretenses. What seemed to resonate with them was that “Hussein is a bad guy. He’s tortured and killed his own people. He needs to be stopped.” The debate began on those terms, President Bush held to that line, and the responses of liberals — “But wait! Do we want to send our troops off to die for something that we don’t really need to do?” — didn’t work in the least. For one thing, the liberal response was just that: a response. For another, our response never redefined the terms in our favor; we never had the power to say, “Let’s stop talking about it in terms of moral propriety; let’s instead talk about the uses and abuses of the U.S. military.” We never shifted the ground to ourselves; instead, all of our responses were of the “Yes, but . . . ” form.

Likewise in the Swift Boat mess, I’m pretty convinced that Kerry — and much of the modern Democratic party — thought that Americans could be convinced by a simple presentation of the facts. The simple fact that Kerry got a Purple Heart should have been quite enough to end the debate — if Americans were interested in facts. And we are, mostly. But since there’s a lot of news spinning around all the time, Americans can’t be expected to catch every single nuance and check out Josh Marshall every time they hear a new bit of conservative spin. So when it comes time to decide how to vote, I imagine a lot of Americans reasoned as follows: “Bush is strong in the war on terror. Is Kerry? Hard to say, but I seem to remember reading something about how he was a coward during Vietnam.” The media further this by running headlines that say things like “Kerry denies cowardice in Vietnam.” That’s an absolutely accurate portrayal, in all likelihood, of what Kerry actually said, and it plays perfectly into Republican hands.

And yet Kerry delayed his response for a very long time, I suspect because he had faith that the American public would see the sham for what it was.

The “flip-flopping” bit during the Kerry campaign was an even better example, because the Republicans’ central claim was really hard to demolish on the basis of the facts — unlike Iraq or Swift Boat. Was Kerry a flip-flopper? I submit that the question is terrifically hard to answer conclusively unless you do at least a little reading. Until you’ve done that reading — and maybe even after you’ve finished it — you have the nagging and oh-so-piquant term “flip-flopper” flip-flopping around in the back of your head.

At the end of it, Republicans did the impossible: they took a candidate who really was pretty solid on paper, and knocked both his legs — decorated war hero, distinguished senator — out from under him.

It’s probably a commonplace by now that the truth has nothing to do with politics, but I guess it took me until now to see that liberals’ obstinacy in the face of this fact is what’s killing them.

P.S.: It occurs to me that the debate over creationism versus evolution is another liberal catastrophe a-brewin’. Again, we think that this is a sheer battle of facts: nearly every scientist admits that evolution is a fact, so those of us on this side think, “If we just talk with them enough, they’ll see the light.” But the evolution debate isn’t about the facts — it’s a pure political play. So it may in fact be the case that more arguing isn’t going to work — that it’s simply the wrong strategy if we want to win. And we do want to win, don’t we?

P.P.S.: The examples keep coming up: how about the debate over whether humans are causing the greenhouse effect, and whether we could reap substantial benefits to the long-term health of humanity by modest policy changes? (That’s how the debate ought to be framed, but it never is — it’s Big Business v. Tree Huggers.) We like to think that it’s a debate on a level playing field between two honest adversaries. But it’s not.

I’d hate to suggest that the strategy for dealing with liars is to be as dirty and deceitful as they are. But I wonder  . . . 

Dan Savage this week features advice from older women that they wish they had known at age 15. Among them is this, my favorite (for reasons that will soon become clear):

The outgoing, macho-acting, good-looking guys you’re attracted to will treat you like crap. The quiet, nerdy, smart, and bookish guys you are not attracted to will treat you like gold. But you may have to seek and draw them out, as they are usually shy.

Yes, goddamnit. Yes.

Steve Kemp submits random web pages to Firefox and watches what happens. Every now and again he gets it to crash. It’s called “fuzzing” — jargon that I just learned from Friend Seth the other day — and apparently it’s common. It makes sense: see how well the program can handle abnormal data.

I imagine the trick is figuring out which data are most likely to cause problems — or, somewhat relatedly, figuring out what a “random” web page is. If the software you’re testing is a black box, I suspect that submitting random pages is your best bet. But if you know the innards, and know something of the design, you can probably home in on problem areas more quickly.

We need more systematic tests like this against major pieces of open-source software. When I worked at Groove, they’d run through a ton of “smoke tests” before certifying a new build, to guarantee that it hadn’t moved backwards in quality. The QA team would run it through an even stronger set of tests using dozens of different environments (different proxies, different versions of Windows, etc.). The open-source community needs this.

I assume something like this goes on before, say, new versions of Firefox are released. Right?

I’m reading Ebert’s funny review of Blade: Trinity, and I know I’m overthinking it, but bear with me: Ebert writes,

Vampires in this movie look about as easy to kill as the ghouls in “Dawn of the Dead.” They have a way of suddenly fizzing up into electric sparks and then collapsing in a pile of ash. One of the weapons used against them by the Night Stalkers is a light-saber device that is, and I’m sure I have this right, “half as hot as the sun.” Switch on one of those babies and you’d zap not only the vampires but British Columbia and large parts of Alberta and Washington state.

I think the idea is that if a point source were that hot, it would destroy everything around it and then some. So I wonder: is there a good simple formula for the drop-off in heat as a function of distance from a point source? I’d have to assume that heat falls off proportional to the cube of the distance, but I want to make sure.

P.S.: I’m aware of Newton’s Law of Cooling, which says that the rate of change of the temperature is proportional to the difference between the current temperature and the object’s surroundings. The solution to the NLoC differential equation is exponential with respect to time, so maybe that’s part of my answer  . . . 

Friendster shows me a new list of People I Might Be Interested In every time I sign in, and today it showed me a bondage fetishist who’s into Linux. I  . . .  um  . . .  whoa. Awesome.

I happened to be curious why the middle digit in area codes always used to be a 1 or a 0, so I googled a little and found the Wikipedia entry on the North American Numbering Plan. At one point the article says,

The North American Numbering Plan Administration (NANPA) is now overseen by the private company NeuStar Inc., who will face the task of adding at least one or two digits to the system within the next 25 years, likely before 2030. During that time, all public and private phone systems on the continent will have to be upgraded and reprogrammed (or even replaced) to recognize the new dialing rules.

Of course the problem’s still going to be around, but I wonder whether Voice over IP will change things. The space of IP addresses is much larger than the space of phone numbers, so if we could get people to use IP we’d be set. Really, I think numbers are a bad idea to begin with to get in touch with people: my email address is a much easier thing to remember, and it should be the central way to connect to me: if your computer happens to connect to stevereads.com on the email port, you can send me email; if it connects on the AIM port, you can send me an IM; if it connects on the VoIP port, you can call me. It seems like it’d be better for everyone if we dealt with the email-address abstraction instead of a phone number.

Whatever you might think of the Abu Ghraib scandal, I hope you’ll agree with me that Rush Limbaugh’s treatment of the 1-year anniversary of the scandal is base, vulgar, demeaning, trivial and deeply offensive.

Of all the arguments in favor of Linux, I don’t think that “simplicity” is going to get us very far. I’m reading a SecurityFocus article on the topic right now, and it seems to make a lot of the errors — and point out the shortcomings — of all the simplicity arguments that anyone ever bothers to make.

Part of the author’s claim is that command-line apps are simpler. Sure they are. But they also will never draw a mass audience to Linux. So if we must insist on simplicity at the expense of GUIs, we can also write off mass-market acceptability.

Even if every Linux application is built from command-line apps, that just seems to force the complexity a level higher. You can build every app out of command-line tools, with a spray-on UI on top, but the final application will still be complex. You’ve gained nothing by bolting together many small command-line applications with pipes; this seems to me no less complex than bolting together many small function calls with pipes. In both cases, the idea is that small pieces are easier to debug, but those small pieces could just as well be clean functions within a single program.

I’d like to see the argument that when we’re comparing equally market-friendly apps — say, OpenOffice versus MS Office — Linux still wins. I don’t see any reason why that should be the case, but I’m willing to buy it.

Friend Seth has been arguing with me lately that Linux will almost certainly be less secure than Windows, given the complexity of the problem that both Windows and Linux are trying to solve. For all my Linux partisanship, I’m fully open to this possibility. If Linux is more secure than Windows, I don’t believe that “simplicity” is the reason. It may be more secure because there are more people looking out for bugs and reporting them. It may be that the Linux community dives on bugs faster, and is more open about its problems than Microsoft; Bruce Schneier says that “Unfortunately, Microsoft treats security problems as public relations problems, and they’ll do whatever they can do to get the most PR.” For whatever reason, the Linux community treats bugs differently. I’m not sure if this is a durable part of the community, but it’s at least the current way that the community operates.

“Simplicity,” though, seems like a weak leg to stand on.

A question for any folks who know Linux internals fairly well: does the Linux kernel support ACLs and atomic file operations natively? It looks like Mac OS Server 10.4 is going to support ACLs out of the box, and I’ve wondered for a while whether the classic Linux permissions list (read, write, execute and sticky for owner, group, and world) is too narrow. And it galls me that Apple is wearing its ACL-superiority over Linux (“a capability unprecedented on any UNIX- or Linux-based platform”) on its sleeve.

As for atomic file operations, my sense is that there’s no Linux system call for atomic writes. This can’t be possible, can it?

I’m really not psychologically prepared for the torrential rain we’re having here today. I went out this morning in Dorky Compromise Gear — namely thermal socks and new sandals — to get coffee with a friend at 10 a.m. at La Luna in Central Square. Not only was the bus half an hour late, but it was cold and drizzling the entire time I was outside. I’ve been nesting ever since. I think I’ll actually be wrapped up in a down comforter drinking cocoa as soon as I get some work done. Feh.

I got in trouble with a former employer because the employer knew fuckall about technology, and noticed that I was using SSH to connect to my home machine to check my email. SSH is encrypted, and it allows communication in both directions over the pipe, so the employer was afraid that I was sending confidential documents to a remote machine. They fired me a few days later, despite my pleas that I was doing nothing wrong. In retrospect, I understand their institutional conservatism. I still think it’s based on a misunderstanding of technology, but it doesn’t matter anymore.

Anyway, now I see that a Debian developer has pointed to Corkscrew, a program to tunnel TCP connections through an HTTPS proxy. I’ll need to play with this, but it looks like I could use this to establish an SSH connection without my employer noticing that I’m establishing an SSH connection; it would just look like an ordinary web-browser connection. I’m not sure what happens on the remote server’s end; presumably the server would have to have a program listening for HTTPS connections and monitoring their contents; if there’s something in the TCP header, say, it’ll decide whether this HTTPS connection should go to the local web server or to the local SSH server. Presumably Corkscrew is a layer between the portmapper (or whatever the proper term is for the part of the OS that assigns ports to listeners) and the applications.

A similar thought occurred to me recently when I read an article about Vonage’s suit against an ISP. Vonage’s claim was that the ISP looked for packets destined for Vonage’s port and threw them out. The FCC settled with the ISP; they won’t be tossing out Vonage’s packets anymore.

This seems like it’ll lead to an arms race unless the government takes this sort of action all the time. But the arms race seems like it’ll always end in Vonage’s favor. It goes like this:

1. ISP throws out packets destined for Vonage’s port;
2. Vonage switches to port 80 (or some other common port), which the ISP simply cannot block. Vonage then installs something like Corkscrew on client machines: it examines incoming packets and decides whether they go to Vonage’s client or to an ordinary port-80 listener like a web server.
3. The ISP starts examining the contents of packets, just like Vonage’s “meta-listener” has been doing. It throws out packets whose signature matches the Vonage signature.
4. Vonage starts encrypting packets. All that the ISP can see after the encryption is the TCP header and an encrypted payload.

There doesn’t seem to be much that the ISP could do after that last step, though I’d be curious if anyone can think of an attack.

Cringely wrote about one variant a while back: instead of actively hunting down its opponents, the ISP increases the priority of certain preferred packets — say, packets destined for its own (the ISP’s own) Voice-over-IP client. There’s nothing wrong with doing this, but it immediately adds latency to Vonage’s VoIP client.

It seems to me that the only good responses here are either market-based (I stop using my ISP if the quality of Vonage communications sucks) or governmental (the feds require ISPs to treat all packets equally — or better yet, require ISPs to be separate from manufacturers of hardware and software). I wonder how this will play out over the next couple years.

Schilling’s honesty on the subject of the recent bench-clearing Sox-Devil Rays brawl is refreshing:

On the WEEI program, Schilling said, “When you’re playing a team with a manager who somehow forgot how the game is played, there’s problems. This should have been over a little bit ago. Lou’s trying to make his team be a bunch of tough guys, and the telling sign is when the players on that team are saying, ‘This is why we lose a hundred games a year, because this idiot makes us do stuff like this.’ They were saying this on the field.

“ . . . I don’t think it was a mystery to any of us why we were out there,” he said. “Guys on the field on their team said the same thing. It’s a situation they started and they escalated.”

I don’t know whether what he says is right — Piniella, for one, disagrees with him — but that kind of bluntness is way too rare.

I happened to be in the frame of mind to think about the philosophy of science today, and I had a little time on my hands, so I swung over to the Boston Public Library to check out some books on the subject.

I basically got off on a tear inside my own head whilst at a coffeeshop today, I guess because I have a hard time taking skepticism about the truth of scientific results seriously. I’ve had this kind of mid-head eruption many a time, but the most recent outburst came after Josh Wretzel and I got in a text-message discussion (of all inappropriate media) about Thomas Kuhn. I texted him, “If Kuhn was right, then why does it seem like certain ideas — such as that the earth revolves around the Sun — are here for good?” Josh replied, “Aristotelian [hence marking the first or maybe second time that ‘Aristotelian’ has appeared in a text message] physics also seemed beyond refutation for a time.” It went back and forth. Eventually I suggested that we both read Deborah Mayo’s Error and the Growth of Experimental Knowledge.

Below the fold, 1,285 words about why I would very much like to stop thinking about the philosophy of science, but can’t.

(more…)

Way to go, Princeton!

It seems that a few years back the Frist family (yep, that Frist family) donated a big chunk of change to the University to build what was later named the “Frist Campus Center.” And starting this morning, students and even faculty members will be standing outside the building ‘filibustering’ Frist — reading aloud from phone books, bios of the seven nominees, the bible, textbooks and probably a bunch of other stuff since they’re going to go for 12 hours straight today. And they’ll probably keep going tomorrow.

Here’s some more information on what they’re doing from one of the filibuster’s organizers.

It’s good to know that at least a small portion of America cares about this stuff as much as I do.

I’d be interested in modeling the most and least depressed cities as a function of their wealth, weather, and so forth. I posit a strong positive correlation between the average temperature in the town and the happiness of its citizens.

Duh.

(Via Jaldhar Vyas, syndicated in Planet Debian.)

The New York Times has a praise-filled article about Mariners hitter Ichiro Suzuki. It’s a fine article, but I’d like to point out a particular weakness in baseball statistics: they’re too often based on counts, and except for people in the Bill James school they’re never relativized. Take, for instance, this quote:

His quick pace could also put him in another select group next month. Suzuki needs only 49 hits for 1,000 in his major league career. He has played 653 games.

The fastest player to reach 1,000 was Chuck Klein, in his 683rd game in 1933. The second fastest was Paul Waner, in his 711th game in 1930.

The pace of games necessary to hit 1,000 is a fairly worthless statistic. It seems pretty clear that the relevant number here — unless our goal is to invent new records — is the number of at-bats necessary to hit 1,000. And the number of at-bats you have depends on how good your team is: a good team will go through more of its batters before the pitcher gets three outs. So if Ichiro were hitting for a good team (which he’s not, so the argument doesn’t hold here), we’d have to adjust his at-bats-to-attain-1000-hits statistic downward. Likewise, ie he’s pitching for a bad team, we may want to adjust it upward.

Secondly, it’s rare to see the standard deviation in baseball statistics, when the number of standard deviations from the appropriate mean is really what we should be going after. In this case, for instance, maybe hitters in the 1930’s tended to get a lot more hits; my understanding is that for a time (maybe not as late as the ‘30’s — I’d have to reread some bits on baseball history) the game was more about the battle between batters and fielders than between batters and pitchers; batters were much more likely to get hits then than now. So we should compare how long it took every batter to reach 1,000 hits with the mean among all batters in a given span of time (say, the ‘30’s), and normalize to the standard deviation; maybe Ichiro is a 1.2-standard-deviation hitter today, whereas George Sisler was a 2-standard-deviation hitter in the ‘30’s.

Once you start down that road, it’s important to start normalizing against other quantities. The first is home-field advantage: maybe Sisler’s home field was especially good for batters, and Ichiro’s isn’t. You end up trying to model a lot of these quantities, and end up with a more reliable way of comparing players across the generations.

I tend to think that most such comparisons are pretty worthless. I find comparisons among contemporary players much more enlightening and valuable, because the point is to do something interesting with the statistics — namely, decide how much to pay a player and which players to put on a team.

Running down that line of thought, Bill James has a fairly convincing argument that the only relevant quantity is how many runs a player produces (or in the pitcher’s case, prevents). That’s a fairly hard task, because someone like Ichiro might produce a lot of runs on a team like the Red Sox and not very many on the Mariners. So it seems like an individual player’s run production will inevitably be a function of the team he’s on. But batting average and total hits don’t seem especially interesting.