Bounded in a Nutshell

I've recently upgraded from Jaguar to Panther. If you have no idea what I just said, then you need to get hip.

So far I'm very happy with the change. The tweaks to the UI are all improvements as far as I'm concerned. After I rebooted once or twice, the new version does indeed run faster and better than its predecessor. And the new version of the Finder is, I find, somewhat more intuitively laid out than the one that came before.

(Retooling the Finder after the fashion of the iTunes interface is an interesting move, btw. Especially in light of how a version of iTunes has been made available for Windows users. Just another piece of switch-bait, I guess...)

In other news, I really am planning on fixing up this site a bit. Expect slightly-better-than-mediocre things in the near future!

So there's a novel that I had to read back in my high school days called "Bonheur D'Occasion" by Gabrielle Roy. The English title is "The Tin Flute"; I don't recall that there was actually a tin flute in the story, but rather at one point a character makes a mental comparison between some event in her life and the feeling you have upon receiving a tin flute at Christmas. It seems like such a great idea beforehand, but then when you get it you're no longer sure why you wanted it.

And I, staring blankly at my computer screen, thought that it would make a nice morose comment on Christmas in general if I could hunt down that passage and blog it. Unfortunately, it seems that the Internet is not a good source for the full text of Canadian novels from the last fifty years, because a quick Goggling didn't turn it up. And I'm too lazy to try and hunt down the actual physical text of the English translation right now. But it's the thought that counts, right?

Contrary to what this might lead you to believe, I'm not actually feeling morose about Christmas or anything. I just thought that it would make an interesting comment.

I'm back in the Home Town for the Xmas week.

One consequence of this is that I'm limited to dial-up speeds.

This is problematic, since it seems that the entire internet is now not meant to run on dial-up speeds. As one simple example, the 'net interface to my Canadian bank seems to involve truly startling numbers of redirects, which -- I'd imagine -- are meant to be invisible.

Funny how visible they become when you're cruising along at 26400 bauds.

Update (Dec. 25, 2003): I've been informed by the friendly folks at cmdo that my use of the word "baud" is in fact incorrect; the proper term is "bits per second" or "bps". It should perhaps be noted that my first modem was an acoustic coupler that operated at 300 bauds (or bps, though "baud" was the term in use). So my mistake wasn't luserism so much as honest ignorance.

So to set the scene: I flew Out West for the American Thanksgiving this year, as is my wont. And riding around the highways of the city in question with the UCoS my host, we're quite naturally surfing the FM band. And -- naturally? -- about half the stations have injected Christmas songs into their rotation. By November 26th.

After the obligatory complaining about the slow, steady march of Christmas music earlier and earlier in the year, we got to discussing depressing Christmas songs. Now there's a fair few of these, as it happens: "I believe in Father Christmas" by Greg Lake, a cheery little ditty about disillusionment as one ages; "Father Christmas" by the Kinks, featuring a street-corner Santa being roughed up by some young punks; even "I'll Be Home For Christmas", which is a very sad song if you actually listen to it. But of course, my all-time favourite is "Fairytale of New York" by the Pogues with Christie McColl, which managed to incorporate elements of all three of these themes (broken dreams, rage against the System, loneliness) and is a pretty damn fine piece of music to boot.

I bring this up now because I'd mostly managed to avoid Christmas music from that day until this afternoon. Fortunately, the song that was playing on the radio at the deli was another fine piece of Christmas cheer: "2000 Miles" by the Pretenders. This is not to be confused with any songs by the Proclaimers that you might be thinking of; rather, it's another variation on the theme of separation during a time of notional togetherness.

One feature that's shared by both the Pretenders' and the Pogues' songs is that, while they're "Christmas songs" in some sense, they're not really about Christmas. Rather, they're using Christmas as a backdrop to make an entirely different point. As such, I find they make a welcome change from the syrupped-up new versions of older songs, and am actually a little regretful that they're tagged as Christmas songs and only make their appearance starting in late November.

OK, so this is something that's been bugging me for something like six years now, ever since the first time I had to program with Excel in a corporate setting.

Microsoft Excel is, in many ways, a nice program. It's a little bloated for my modest needs, but I can well see that most of what I regard as useless cruft has its place in the grand scheme of things. It's reasonably fast, and reasonably intuitive. Overall, Excel may well be my favourite MSoft product.

My principal criticism -- and the reason for today's rant -- has to do with the gap between formatting and functionality. Specifically: when I'm assembling my final grades of a semester, I usually have various numbers marked by colouring either the text or the background. These colours tell me things: so-and-so handed in Worksheet X five weeks late (and so gets a tremendous penalty), so-and-so's grade was adjusted for good and sufficient reasons, etc. Often, I would like the ability to treat these numbers differently from their conformist, black-on-white cousins.

The Excel IF function, however, does not seem to recognize tests based on the colour of a cell. Or if it does, it does so in an undocumented, hidden, and enigmatic manner that I've been unable to unriddle. And trying to bring VBA into it helps about as much as you might expect: not at all.

Of course, it's possible -- probable! -- that I've just been consistently dense for the past several years on this topic. That wouldn't surprise me very much, actually.

Well, OK, not really. It's exam time at my Urban Commuter Campus, and my classes have theirs moderately late in the schedule. Hence, I've posted an absurd number of office hours for the past couple of days, in just in case there are any visitors. The response has not been, let us say, overwhelming. But that's OK, because it's given me a chance to almost catch up on my grading from within the semester before I have to deal with the avalanche of finals that's teetering above me.

There's been a couple of interesting posts by Matt Yglesias about the mathematics of public policy. Now if I were a geek...

...well, OK, yes, I am a geek...

So fine. In the "static" model, let's first observe that all four "change" positions -- that is, the two ideal points, as well as the two adoptable policy points -- should be collinear. This is a modestly obvious consequence of the triangle inequality. Hence, Matt's diagram is a little deceptive, but that's OK, because it's just a picture, after all.

One of the comment-threads asked why the Lieberman circle was so much bigger than the Bush circle, and one response to that was that, well, this is just a low-dimensional projection. Now, the latter might well be true; I very much doubt that a realistic model of public policy is going to be a mere two dimensions. On the other hand, as long as we're only looking at two agents and a single status quo, then the two-dimensional situation tells us everything we need to know about the scenario. As long as we're assuming rationality (and why not?), then both players' optimal solutions are going to lie in the plane defined by those three points; anything else is worse, again thanks to the triangle inequality.

Now if you really want to geek out, consider this: an underlying assumption of Matt's model is that a Euclidean metric is an appropriate measure of satisfaction with a policy. Is this reasonable? If, as a commenter suggests, a Boolean model (where all questions are "yes" or "no") would be more appropriate, then we probably want to use a discrete metric instead. Or maybe something even weirder would work best. I'm not even sure how one would determine an appropriate topology from empirical fact. Yet one more thing to think about in lieu of grading papers...

Orcinus has been talking recently a lot regarding "eliminationist rhetoric" from the Right; it's apparently become fashionable, in certain circles, to threaten (or wistfully imagine) violence -- preferably fatal -- against one's political opponents. (Read: anyone who disagrees with your opinion on some particular matter.)

One of the standard branches of discussions, when such things come up, is that there's no real difference between the Left and the Right in this regard. That is, extreme political rhetoric from, say, Ann Coulter is roughly analogous to that of, say, Michael Moore. Now, there are arguments for and against this that one can make, but they're like most other blogosphere arguments and don't settle anything. I'd like to propose a little experiment.

The question at hand is this: Is there no real difference between Left and Right political rhetoric? Our null hypothesis is that, no, there is no difference. The alternative hypothesis is that the two flavours of rhetoric are distinctly different. The methodology is as follows:

• Find a couple dozen good examples of "political hate speech" from each side.
• Modify them to leave the meaning more or less intact, but with context removed. (So a statement like, "Republicans look funny" might get translated to, "Members of Party X look funny".)
• Post a list of these context-free statements someplace accessible.
• "Randomly" select a sample of interested observers; have them self-identify their own political viewpoints and awareness. Get them to mark each statement as being sourced from the Left or the Right.

The sampling could be problematic; I'm envisioning soliciting for volunteers, then choosing a sample from those who indicated their willingness to take part. Another problem is that any well-known attacks would probably be identifiable regardless of substitutions.

What would this tell us? Well, several things, potentially. If political rhetoric really is interchangeable -- with just the names of people and parties different -- then a participant would have an expected value of 50% right... no greater than chance. If there's a significantly higher correctness, then that would favour the alternative hypothesis. If would also be interesting to see how one's choices correlated with one's opinions.

I might see about pulling this off, actually, once I've talked it over with some statistician friends. Stay tuned...

Gah. Stupid cookie-based semi-stateless web applications.

My original post was more interesting, but here's the summary: sorry for the lack of updates, on the off-chance that someone's reading it. Expect a more active blog no later than the new year.