Musings

We’ve all seen the gruesome pictures of the Iraqi Kurds, gassed in March 1988 in the town of Halabja near the Iranian border. Proof positive (as reiterated in the State of the Union address this Tuesday evening) that Saddam is an evil tyrant, who would “gas his own people.”

There’s no doubt that Saddam is an evil tyrant.

But, according to Stephen Pelletiere, who is very much in a position to know, it is overwhelmingly likely that it was the Iranians, not the Iraqis, who gassed the Kurds of Halabja.

And the President surely knows this (assuming he wasn’t too busy throwing spitballs during the relevant briefing).

So what was he talking about on Tuesday?

Update: Pelletiere’s conclusions have been hotly disputed in many quarters. HumanRightsWatch concluded that it was mustard gas and sarin, which the Iraqis did possess, which killed the inhabitants of Halabja. (The main pillar of Pelletiere’s argument is that hydrogen cyanide — supposedly part of the Iranian, but not Iraqi, arsenal at the time — was used in the attack.) And, even if Pelletiere is right about Halabja, it is almost certain that Saddam’s subsequent murderous campaign against the Kurds included the use of poison gas against other — less famous — targets.

This is an important point: the Administration is not arguing that we should go to war because Saddam killed 100,000 Kurds. The argument (which, when stated plainly, may sound a little callous) is that we should do so because he used WMDs (poison gas) to kill some fraction of them. So “details,” like what exactly happened in Halabja, matter.

So I’m not all that please at how my last post renders on the Mac (Mach-O Mozilla with the Mathematica Fonts). The “stretchy” characters (the integral sign and the overbar on the anti-D3 branes) … aren’t.

How does it look on other platforms?

Anyone using one of the MathML plugins for (gasp, shudder) Internet Explorer?

On another matter, you might have noticed that the RSS feed for Brad DeLong’s blog has disappeared. For the second time in a week, the RSS parser is choking, because his feed contains “undefined entities” (  this time; last time it was £). These are perfectly valid in XHTML, but must not appear in an RSS feed, for it to be valid XML. I think it’s fine to use their numerical equivalents (  and £, respectively). Or you can wrap the whole <description> as CDATA. Or maybe I need a yet-more-liberal parser.

Anyway, in a day or two, Brad will have posted enough new stuff for the offending article to slip off his feed, and then we should be good for another week.

Update: Actually, Brad’s problem comes down to his using faulty old templates to generate his RSS feed. The new MT Templates don’t have these problems.

(Warning! This post uses MathML. The equations will probably look like garbage unless you are viewing it in Mozilla, with the requisite fonts installed. Sorry, that’s just life.)

Kachru et al suggest a way to obtain classically-stable (and quantum-mechanically long-lived) 4D de Sitter solutions of string theory.

The starting point is the class of compactifications introduced by Giddings et al. There, the flux-induced superpotential,

fixes the string coupling and the complex structure of the 3-fold $M$, leaving the Kahler modulus, $\rho$ (we’ll assume only one) as a flat direction.

The next stage (a bit of handwaving, but not implausible) is to assume that nonperturbative effects induce a superpotential for $\rho$ which lifts this remaining flat direction. In a fairly robust fashion, one ends up with a supersymmetric vacuum in 4D anti-de Sitter space.

Now comes the tricky step. We imagine changing the fluxes (a discrete choice) so that the tadpole cancellation condition now requires the presence of one (or a small number of) $\overline{\text{D3}}$ brane(s). This breaks supersymmetry and induces a term in the potential which would normally lead to a runaway behaviour for $\rho$. Naively, I might guess that the coefficient of this term would be large, and that it would totally overwhelm the nonperturbative superpotential which generated a minimum for $\rho$ in the first place.

Well, not according to these guys. They claim that the coefficient can be small (so as not to totally destabilize the minimum) and, moreover, can be fine-tuned (again, we have only discrete choices) to produce a minimum with a small (tiny!) positive cosmological constant.

If you swallow all of this, it’s not to hard to believe the last step: namely, while this minimum is only metastable (we still have $V\to 0$ for $\rho \to \mathrm{\infty }$, after all), it can be incredibly long-lived — more than ${10}^{10}$ years.

Some obvious points:

• If any of this makes sense,, it should have a description in terms of 4D supergravity, presumably related to the solutions that I blogged about previously.
• Supersymmetry breaking in the real world is much larger than the scale of the cosmological constant. That’s always been a bit of a problem, but may be less so here. Usually, the problem is discussed assuming the supersymmetric vacuum is 4D flat space. Here, the supersymmetric minimum would more naturally be thought of as being anti-de Sitter. We have to lift the minimum of the potential by a lot (much more than its final value), thereby “explaining” why the supersymmetry breaking scale is so much larger than that of the cosmological constant.
• Of course, the height of the barrier is more directly related to the scale of supersymmetry breaking. So, as potential inhabitants of a false vacuum, we might, perhaps, have reason to be fearful of the next generation of accelerators.

Gary Markstein has the answer to the Iraq conundrum.

Dan Kabat was visiting today, and gave a talk about his work with Easther et al on M2-brane cosmology. I’ve commented on this before, so I thought I’d add what I further learned from my conversations with Dan.

One thing that confused me was their restriction a rectangular torus and to branes with wrappings parallel to the coordinate axes. Choosing a rectangular torus makes the computations much simpler (fewer parameters in the metric). The motivation for the latter is that “diagonally-wrapped” branes would not be compatible with maintaining a rectangular torus.

This is unfortunate, because the diagonally-wrapped guys are secretly rather important in the Brandenburger-Vafa type argument. It is only strings (branes) which are transverse which generically will intersect (and then, only if there are sufficiently few “large” dimensions). So the generic annihilation process is something like a (1,0) string and a (0,1) string annihilating into a (1,1) string.

Diagonally-wrapped strings (and hence, when we consider the back-reaction, non-rectangular tori) are inevitably going to occur. Brandenburger and Vafa never worried about the details of this. They merely noted that, even tranverse strings will generically never meet in more that 3 large spatial dimensions.

The other thing which vexed me in my original post was the non-semiclassical nature of the M2-brane, which made me wonder whether the naive picture of a gas of free branes could ever make sense (unlike a gas of free strings). If the branes are widely separated, then the “dressing” of the branes by those narrow throats simply represents the interaction of the branes with supergravity. Unlike the string case, there is never a “free gas” limit in which the coupling to supergravity can be made small (so I still don’t know how to do thermodynamics).

When the branes are not widely separated, of course, all bets are off. But, at least when some of the dimensions get large, we are “safe”, so long as we replace these “membrany” effects (to coin a phrase) with the coupling to supergravity.

Anyway, I’m a little clearer on what they’re trying to do so, on that score, Dan’s visit was a resounding success.

The activity of the MS-SQL worm slowed large parts of the internet to a crawl yesterday. Thanks, again, to Microsoft for enhancing my internet experience.

Even my banking experience felt the impact.

Locksmithing, plumbing, and a few other trades seem to persist in a guild-like mentality, where the “secrets” of the trade are passed on from Masters to Initiates. In the case of locksmiths, this is a signal case of what is elsewhere derided as “Security Through Obscurity.” When the “secret” leaks out, you are stunned to learn just how insecure the system really is.

A standard pin-tumber lock has P pins, each of which can be cut at H different heghts. That means H^P different combinations which, for modest values of H and P, could number in the millions. Since trying each combination involves cutting a blank and inserting it into the lock, this would seem to make pin-tumber lock invulnerable to brute force “keyspace search” attacks.

The situation changes dramatically when, in addition to the “change key”, which opens just this particular lock, there’s also a “master key” which opens all similar locks in the building. In this case, each pin has a second cut at some (unknown to you) height. As cryptographer Matt Blaze discovered, such systems (which all of us encounter in our day-to-day lives) are vulnerable to escalation of privileges (the owner of a change key being able to create a master key) through an elementary “Adaptive Oracle” attack.

The “Oracle” (which tells you when you’ve guessed right), in this case, is the lock that fits your change key. With P+1 key blanks (costing less that $2) and small bit of effort, you can create your own master key. The algorithm is so blindingly simple that you can probably guess it from just this description.

No? OK, here’s what you do.

Cut a blank to be identical to your change key, except at the location of the first pin, where you leave it uncut. Try it in the lock. If it doesn’t work, start trimming away until you find the height of the second cut. Since there are only H-1 heights to test, you will be done soon. Now take a second blank and repeat the procedure with the second pin. After using P blanks, you have learned the heights of the master cuts on all P pins. Use your last blank to cut yourself a master key.

There are many more details and variations in the paper. And, apparently, this has been known in some circles for a very long time. Now we all know.

Thanks to Ed Felten for the links.

Update: Of course, it’s obvious that you only need P, not P+1, blanks. But blanks are cheap, anyway.

I learned that, despite being Gecko-based, Chimera can’t render these pages as XML. So back it goes into receiving them as “text/html” (ie, no MathML support). And I filed a bug report with the Chimera team.

In case you’re keeping track, that means that the .htaccess file for this blog now looks like

RewriteEngine On
RewriteBase /~distler/blog/
RewriteRule ^$   index.html
RewriteCond  %{HTTP_USER_AGENT} Gecko
RewriteRule \.html$   - [T=application/xhtml+xml]
RewriteCond  %{HTTP_USER_AGENT} Chimera [OR]
RewriteCond  %{HTTP_USER_AGENT} Safari 
RewriteRule ^$|\.html$   - [T=text/html]

On the other hand, P&L Systems quietly released a new version of their Mesa spreadsheet, which fixes the printing bugs under Jaguar. Since I had previously ragged on them, I want to publicly say, “Thank You”.

So I’ve been blogging for a little over 3 months now. Perhaps it’s time to reflect on the experience. All in all, I have to way that it has been a tremendous waste of time. And I mean that in the best possible sense.

One thing that surprises me is that I have spent much more time futzing with the ‘plumbing’ of this blog than I expected to. Some people seem to change the layout of their blog every other week. The appearance of mine has changed very little. The most dramatic change I made in the superficial appearance was to rationalize the fonts (and sizes) to make the whole thing more readable.

On the other hand, I have thought a lot about some of the defects of the “traditional” blog layout and tried to rectify them here. For instance, consider the Blogroll. Traditionally, this is just a static list of links to other weblogs or sites. Ho hum, why should I click on that link? Or, perhaps, Oh cool, Aaron Bergman (another physicist-blogger) has a weblog, What’s he blogging about?

Now, of course, Aaron’s a bad example because his blog doesn’t have an RSS feed. But those which do (indicated in bold on my Blogroll) are syndicated, so you can see what they’re blogging about before clicking over there. That just seems a heckuva lot more useful than a static list of URL’s.

Another thing that’s basically irksome is the ephemeral nature of the weblog. Content that’s more than a week or two old drops off the main page, never to be read again. Yes there are links to the archives. But nobody clicks through to them. And yes, there’s a search engine (a mighty fine one, I might add), but unless you are looking for something specific, you’re unlikely to use it.

So I decided to add a list of Random Past Entries (changed once an hour) to the sidebar, in the hope that serendipity might bring to your attention some interesting post from the past. And, again, you shouldn’t have to click through to see if it really was interesting; an excerpt should be available with a mouse-over.

I thought that MathML was going to be easy. It wasn’t. But the next release of MovableType will bring creating posts with embedded MathML a lot closer to “easy” (by converting them on-the-fly from embedded itex). And hopefully the release of the Stix fonts will ameliorate the residual rendering problems, and make the whole process a lot more plug-‘n-play on the user side.

(Those of you who are Internet Explorer users may wonder what the heck I am talking about. The fancy sidebar stuff is only viewable in a Standards-compliant browser, like Netscape 7, Mozilla, or the next release of Apple’s Safari. MathML is natively supported only by the Gecko-based browsers, though I hear there are MathML plugins available for IE.)

The final thing that surprises me, looking back, is that it proves much harder than I thought it would be to say something interesting and relevant about physics in a few paragraphs, with few or no equations. I met a colleague at a symposium back in November, who said

Your typical post says, “I read this paper on the archives. It looks interesting.”

He was teasing (I think!). But it does raise the issue of the general shallowness of the blogosphere. Just because the topic happens to be physics doesn’t guarantee any greater level of profundity.

All in all, it’s a heckuva lot easier to write about something other than physics. Which, despite my better efforts, seems to be what I’ve been doing much of the time.

Which reminds me of an anecdote. Many years ago, Sacha Polyakov approached me at a cocktail party. He was very apprehensive at the prospect of teaching his first-ever undergraduate course a Princeton. “At least,” he said after we’d discussed it for a while, “it’s only Nonlinear Dynamics, and not Quantum Field Theory.” “Oh,” I said (to one of the great men of modern QFT), “why is that?” “If it were Quantum Field Theory, I’d feel responsible for the subject.”

I laughed so hard, I nearly p…

Oh, never mind, read it yourself.

Do I ever get to blog about anything else?

A new version of Kung-Log is out, “rewritten from the ground up in pure Cocoa.” Lots of yummy new feature, like syntax-colouring, a customizable “HTML Tags” menu, …

Dave Hyatt’s got Safari’s :hover code working, so the popup RSS Feeds in my Blogroll will work in the next version of Safari, too. They’ve squashed an impressive number of CSS bugs. And they have the rudiments of an XML parser going. Who knows! Safari may even support MathML someday!

It’s always good to have a niche. If you’re Mel Ulrich, that niche is “operatic baritone with shaved head, who looks great with his shirt off”.

I saw him last year as Stanley Kowalski in Streetcar Named Desire, and tonight as Joseph de Rocher in Dead Man Walking. He’s a fine singer and a good actor, but with his particular mix of qualifications, he is … umh … without competition.

As many of you know, the reason I started this blog was because I thought the weblogs would prove to be an excellent vehicle for “informal” physics discussions.

Being able to type equations (preferably in TeX) and have them display inline in your browser is an essential part of that. My first attempt to put MathML in this blog didn’t work so well. Perhaps you think I entirely abandoned the notion. Not at all.

Some of you might also have noticed my dogged obsession with bringing this blog up to full XHTML 1.1 compliance which must, I admit, seem pretty quixotic.

Anyway, there was method to my madness. With a little mod_rewrite trick in the .htaccess file of this blog,

RewriteEngine On
RewriteBase /~distler/blog/
RewriteRule ^$   index.html    
RewriteCond  %{HTTP_USER_AGENT} Gecko
RewriteRule \.html$ - [T=text/xml]
RewriteCond  %{HTTP_USER_AGENT} Safari
RewriteRule \.html$ - [T=text/html]

I can serve up XML to Gecko-based browsers which grock MathML (but only when rendering XML files), while sending “plain old” HTML to other browsers. It’s actually the same file; just the MIME-Type is altered, depending on what browser does the asking.

This only works if my blog is 100% valid XHTML. XML parsers puke at the slightest error and refuse to render the page, unlike your basic HTML parser, which will try to render something out of even the most broken HTML.

I think I’ve succeeded. There were surprising bits of invalid (X)HTML hidden away in obscure corners of MovableType, as this thread delineates. I think I’ve caught everything now, but if you’re running a Gecko-based browser and something here returns an XML parser-error, let me know.

Assuming you have the fonts installed, this means that (MathML)

will look like (screenshot)

in a Gecko-based browser.

But, of course, that’s only the first step. In the forthcoming version of MovableType, Ben and Mena have announced the introduction of a Text-Filtering API. I’ve already mentioned some of the benefits, but for present purposes, the biggest payoff is that we can write posts with embedded itex (a dialect of LaTeX) and have MovableType automatically run them through itex2mml to convert them to embedded MathML.

That’s not quite how I created this post. Instead I ran it through the online itex2mml converter. But you get the idea…

The beginning of the semester leaves me too busy to post much in the way of serious stuff, but let me commend to your attention Aaron Bergman’s summary of a recent paper by Visser et al.

A good post by CalPundit on the “debate” over Global Warming. It scares me when leftists argue that Science is just a matter of social consensus. It scares me even more when right-wingers argue the same.

In syndicating the RSS feeds of my BlogRoll, I discovered that I had alarmingly increased the size of the main page of this blog.

A little examination revealed the cause: Lawrence Lessig, Sébastien Paquet and Zimran Ahmed decided to include the full content of each of their posts in the <description> field of their RSS feed.

Hey, guys! That’s what the (very optional) <content:encoded> field is for. The <description> field is for a short summary or excerpt of your post.

The extra baggage adds 23KB to the size of my main page. Significant, but not a killer. I’d be happier somehow if they published more, uhm, succinct <description>’s. But, if you really want to, you can read the full content of their blogs right here…

Anyway, here’s a screenshot of what I am really striving for. A much more useful version of the ubiquitous BlogRoll, isn’t it?

Update: Come on, Distler, don’t be a doofus! You can’t rely on people to do the right thing. Use a filter instead.

I didn’t like the “RSS Feeds” at the bottom of the Sidebar. Too ugly. So I got rid of them, in favour of something much, much nicer. Move your mouse over the “Links” section in the Sidebar, and you’ll see what I mean.

Or, at least, you will if you are using a Gecko-based browser, like Mozilla. From the sounds of it, KHTML-based browsers like Safari will support this part of the CSS2 Specification real soon now too.

My apologies to those (you know who you are) whose browsers aren’t Standards-Compliant.

P.S.: OK, OK. I suppose this could be done using Javascript. If anyone has a simple (no browser sniffing!) suggestion for how to do this, I’d be happy to entertain it. But it’s hard to beat 10 lines of CSS code styling nested <ul>’s (the latter generated using the mt-rssfeed plugin) for simplicity. Thanks to Eric Meyer for showing the way.

Is there no end to the mischief engendered by this stupid, stupid law?

As Ars Technica reports, Lexmark has gone to court, arguing that third party Toner Cartridges (!) contravene the DMCA:

In a 17-page complaint filed on Dec. 30, 2002, the company claims the Smartek chip mimics the authentication sequence used by Lexmark chips and unlawfully tricks the printer into accepting an aftermarket cartridge. That “circumvents the technological measure that controls access to the Toner Loading Program and the Printer Engine Program,” the complaint says.

A hearing on the matter is scheduled for today. This is dangerous, dangerous territory. The thinking behind this is mad: putting a chip on anything makes it digital media.

… and hence protected by the DMCA. To which I say, “you ain’t seen nothin’ yet!”

Update: Ed Felten has added his comments on the legal issues involved. I think he’s a little too dismissive of the “circumvention” argument. If reverse-engineering a “cryptographic secret handshake” isn’t “circumvention” under the terms of the DMCA, it’s hard to imagine what is (DeCSS anyone?). And I’m not sure I buy his webserver analogy. Isn’t breaking the copy-protection on a piece of software, allowing you to execute that software, a violation of the DMCA? Or do you actually have to be able to see the source code before it’s a DMCA violation?

Apple’s Safari is out, at least in beta. To the suprise of many, including me, it uses the KHTML rendering engine (the one behind the Konqueror browser) , rather than Gecko (which powers Mozilla, Netscape7, Chimera, Phoenix,…).

It’s a nifty-looking browser, incredibly lightweight (a 3+ MB download), and fast as heck (I thought Chimera and Mach-O Mozilla were fast!).

Unfortunately, the KHTML engine does have a few CSS bugs. Mark Pilgrim has a review and a tracker for these CSS bugs. A couple affect the rendering of this blog, which is why it currently looks like crap in Safari (and, I suppose, Konqueror, too).

Fortunately, Dave Hyatt’s blog gives every indication that the Safari Team are working fast to squash the bugs.

Methinks I hear a crackling sound as the Browser Wars heat up again.

There’s a great scene in an old Laurel & Hardy movie, in which Stan and Ollie are hired as gardeners. Their first task is to trim a hedge and, to save time, they decide to start at opposite ends. Predictably, when they meet in the middle, one has trimmed his side 6 inches lower than the other. So they try again, and this time the other side is lower. The scene ends with Stan running over the former hedge with a lawnmower.

I was reminded of this by Winterspeak’s passionate defence of cutting taxes on dividends. First they cut taxes on capital gains, violating the principle of tax-neutrality. Now, to compensate, they propose cutting taxes on dividends (even lower). You see where this is going …

Except the analogy with Laurel & Hardy isn’t perfect. See, in addition to issuing stock, there’s another way corporations can raise money, namely by borrowing. Taxing interest on corporate bonds at a different rate from the tax on dividends or capital gains is every bit as much of a distortionary violation of tax-neutrality.

You haven’t heard the Bush Administration advocate slashing taxes on interest from corporate bonds…yet. Don’t worry, you will. The analogy with Laurel & Hardy was flawed. Perhaps the Three Stooges would be more apt.

I hadn’t noticed when their paper appeared last May, but Trigiante’s talk at a Workshop in Leuven caught my eye.

They’ve found stable de Sitter solutions to gauged N=2 supergravity. The gauging is always by a product group (SO(2,1)×SO(2) or SO(2,1)×SO(3)), where one of the factors is noncompact and the other factor admits a Fayet-Iliopoulos term.

It would be quite interesting to embed these gauged supergravity theories in string theory. For compact gaugings, the gauged supergravity theory turns out to be a “consistent truncation” of some ten-dimensional string background. The gauge group ends up being the isometry group of the internal space.

I’ve never had much of a conceptual handle on the noncompact gaugings of extended supergravity theories, but if one could find these solutions as consistent truncations of some string background, that would be very exciting.

The blogosphere is all a-titter about the latest arch-conservative, self-described Orthodox Jewish, UCLA undergrad blogger, Ben Shapiro.

I have to say I found something a little puzzling. It wasn’t the writing which, as expected, was unremarkable (sort of Rush Limbaugh meets Doogie Howser, with a touch of Meir Kahane thrown in). No, what I was puzzled by was, given his prominently-proclaimed commitment to Orthodox Judaism,

1. Where’s the yarmulke?
2. Does he not evince even the slightest respect for the prohibition on lashon harah?

(Photo courtesy of Tom Tomorrow.)

My hat’s off to Donald Rumsfeld!

The Washington Post and the Manchester Guardian are reporting that, as Reagan’s “Special Envoy to Iraq” during the Iran-Iraq War, Donald Rumsfeld helped pave the way for Saddam to acquire materials for chemical and biological weapons, aware that these weapons were being used “almost daily” in the war with Iran .

In his position, I’d have a hard time keeping a straight face while blustering about Saddam’s possession of chemical and biological weapons. How could I ever maintain the proper tone of righteous indignation, knowing that I had been instrumental in supplying him with those weapons in the first place?

But Rummy … Rummy pulls it off with aplomb! Were it not for these Freedom of Information Act-induced articles, I would never have suspected a thing. He’s that good!

Note to Hollywood: Give the man a role in the next John LeCarré thriller. He looks the part and man can he act!

I don’t think I explained, in my previous post, why Ben and Mena’s announcement is so exciting. The point isn’t that it’s boring to type HTML tags. The point is maintainability.

For example, consider my recent upgrade of this blog from XHTML 1.0-Transitional to XHTML 1.1. Among other things, that meant that <blockquote>Some Text</blockquote> was no longer valid XHTML. Instead, one needs to write (say) <blockquote><p>Some Text</p></blockquote>.

In the forthcoming text-filtering architecture, I would not have to muck with my blog entries at all. I would just have to edit some filter template somewhere to insert the extra <p>...</p> in the output. Under the current system, I had to go in and edit my blog entries by hand. Luckily, since I haven’t been blogging for very long, there was not that much to do. But if I had several years worth of content to modify …

Anyway, the contents of this blog may, over the years, grow intellectually-stale. But it will always be maintainable with the right text-filters.

Ben and Mena Trott have announced that the next version of MT will have a pluggable text-filtering architecture.

Currently, the text filtering options are: 1) convert line breaks to <br /> [and double line breaks to </p><p>] or 2) do nothing.

This sorta sucks, because if you want to include structured text (eg, <blockquote>’s, etc.), you need to turn off the convert line breaks option and insert all of the HTML tags yourself. The whole idea of a CMS like MovableType is to separate the “content” from the structural HTML markup (just as CSS lets you separate the structural HTML markup from the visual formatting).

The screen shot of the web entry interface looks cool. I hope this is supported in the XML-RPC API, so that blogging tools like Kung-Log can access the new features.

Woohoo!

First post of 2003. No real content, just wishing you, dear reader, a very Happy New Year.