log (2005/04/01 to 2005/04/07)

older log
newer log


site news
talking place

RSS (summaries)
RSS (full posts)
  Atom (full posts)

What's playing?
Thursday, April 7, 2005  permanent URL for this entry

So arg. Our last entry was all about cognitive systems and the philosophical implications of functionalism; so this one naturally is going to be about stupid CSS bugs in stupid web browsers.

Just for variety.

For relaxation I finally started coding up a Broken Koans page to collect the various bits of Zen debris that've been generated here in the log, or that I've posted to alt dot zen, or whatever. And as usual starting a new web page I wanted to make it valid XHTML 1.1. (What's there at the moment is half-formed and unfinished; if you're reading it sometime in the future it may be all done and/or at least none of these remarks may apply.)

Well, everything was going along fine (well, almost fine: it's pretty annoying that IE doesn't know about the "hover" thing on anything but "a" elements), until I got to the first couple of bits of debris that are formatted as poems. No problem, thinks I, I know (after googling around for a good CSS manual to remind me) that that can be handled in a nice modern way by setting the "white-space" property to "pre". So I coded that up, and it worked fine in Opera.

I went over to IE to check, and the property was being ignored; the text was getting flowed together as usual and de-poemified.

It took me quite awhile to find a note on the Web saying that IE respects the "pre" value of this property only when "quirks mode" is off (which doesn't make much sense to me; are there old documents out there that assume for some unthinkable reason that the "pre" value doesn't work?). Well, fine, I want that mode to be off anyway, since I'm being all standards-compliant and stuff, and I don't want no quirks. So I look to see how to turn quirks off.

Quirks are off, says the Web, as long as there's a DOCTYPE for the page that isn't one of the special quirky ones. Well, but, I had a very nice DOCTYPE that specified XHTML 1.1, which isn't one of the quirky ones. So huh?

Turns out (at least the best theory I came up with after experimenting) that IE only recognizes that DOCTYPE if it's on the very first line of the page. But of course if we're XHTML, the very first line of the page is the "<?xml" line, or it won't validate. So heh.

I decided to break down and use quirks mode, by using a "pre" element (instead of a "pre" value for "white-space" on a "p" or "div" element). That broke in a different way! Turns out that in quirks mode IE gets font sizes wrong in various ways, and the only way I could find to get the stuff in the "pre" element to be the same size as the non-pre text would be to hard-code the font sizes, and I didn't want to do that.

So I backed off again. What I've done at the moment is changed from XHTML 1.1 to HTML 4.01 Strict. That doesn't require a "<?xml" line to validate, so I can give IE the first-line DOCTYPE that it insists on, and stay out of quirks mode, so the "pre" setting of "white-space" works. Feh.

'Course now the w3c validator isn't happy with me because the web server apparently isn't specifying a charset, and without the "<?xml" line it can't determine it from the content. So I'm off to see if I can tell the server how to send up a charset by fiddling with a dot htaccess file or something.

IE is annoying.

Although I suppose today's lesson is that that's really a pretty trivial thing to be annoyed about...   *8)

(Note post-uploading: turns out that IE also doesn't support the "fixed" value of "background-attachment" for non-body elements, so the very nice "semitransparent boxes sliding over a stationary background image" effect that I put together is only visible in real browsers, like Opera and Firefox, and not in IE. Sheesh.)

Tuesday, April 5, 2005  permanent URL for this entry

Probably in reaction to our musings about the celestial masters on Sunday, a reader writes:

Are you stoned?

Hey man, we're stoned on life. Crazy! Peace.

The little daughter was trying to find a book that she remembered reading awhile back, and she couldn't remember the title or author, but just some rather vague details of the plot. She searched the house, and called me at work, but to no avail. Just before I left work I posted the vague description to a relevant-looking newsgroup just on the offchance. By the time I got home we had an answer.

Nice to know Usenet still thrives!

I'm also seeing essentially no commercial spam in any of the (mostly Zen and Buddhism related) groups that I'm following via Google Groups. (Just a few fundie Christians spamming links to their "do you want to be saved?" sites.)

I find this odd; does anyone know what the story is? Are there really good spam cancelbots out there, or does Google Groups filter really well, or have the spammers just forgotten about spamming Usenet in their thirst for new horizons of weblog trackbacks and VoIP phones?

I wrote off to Brian Weatherson about the question of reasons for caring about future selves that we were pondering the other day. He very kindly wrote back with some names and pointers that I should follow up, but I'm too lazy right now. (So why am I even bothering to write this paragraph? I should just wait until I actually have something substantive to report.)

So this Churchland piece in the latest J Phil. I finished it, and I still think that he mises the mark.

His basic thesis is that it turns out that functionalism is wrong. Here, functionalism is the theory that what's interesting about "cognitive" things is the class of functions that they compute (and, because we're not nasty old behaviorists, we acknowledge that some of the inputs and outputs to the functions are the before and after internal states of the thing); and that since the function you compute is (in a significant sense) separate from how you compute it, what's interesting about "cognitive" things is a different question from the question of how human brains work.

This is wrong, Churchland says, because it turns out that what's really interesting about cognitive things isn't the functions they compute (it turns out, he says, that they compute functions that aren't even "remotely similar"), but rather the way that they compute it. Not "the way that they compute it" in the sense of the exact biochemistry of the human brain, but in the more general sense of a particular kind of nonequilibrium thermodynamics, and "vector-to-vector transformations" involving activation vectors in a particular kind of massively distributed system.

And finally he says that the error of functionalism, in thinking that the function was the interesting thing and that that was separable from the implementation of the function, was a bad thing for the field, because it kept people working on cognitive science and philosophy from thinking about how brains actually work.

I think Churchland is right that some interesting strides have been made in understanding how our brains work, and how that involves nonequilibrium thermodynamics and activation vectors in massively distributed systems and all. But I think he's wrong about most everything else he says in the piece.

First and simplest, he tries to refute "you can't reduce psychology to human brain neuroscience" by claiming that in fact you can (well, someday) reduce psychology to the science of nonequilibrium dynamics and activation vectors and so on. But the second doesn't refute the first at all; reducing psychology to the formal math of certain systems is far from reducing it to the neuroscience of a particular species. I really don't know what he was thinking there.

But he's also wrong about the "what's interesting about cognitive things" question. Which of these would be more interesting:

  • a creature that behaved in an intelligent-seeming way but did it using hardware that didn't involve vector-to-vector transforms of activation vectors in a massively distributed system, or
  • a creature that had hardware that did involve vector-to-vector transforms of activation vectors in a massively distributed system, but behaved incomprehensibly and apparently at random?

Okay, it's a trick question: they're both interesting. But which one would we call a cognitive creature? I think it's clear that the first is the cognitive one; the best evidence (the interesting evidence) for cognition is behavior, and behavior means what function you compute (and, with the obligatory nod to Searle, the fact that that function is attached to the real world in some useful way).

The creature that seems to have brain-like hardware but just sits in the corner emitting white-noise and turning upside-down now and then is interesting because we suspect it might be cognitive, and we'd like to talk to it, or cure it, or wake it up, or whatever.

But the creature that's talking to us and building castles out of lego is interesting because, hey, it's one of us; the fact that it's doing it with an incredibly fast single-processor CPU, or with some amazing hack that's never occurred to us before, doesn't mean it's not cognitive, it just means it has novel hardware.

So that's those two things. And on his final more or less historical claim, that functionalism's notion that psychology is autonomous with respect to neuroscience has "served to insulate the relevant research from exactly the empirical information that promised the most interesting and authoritative constraints on whatever models were put up for evaluation", I'm also dubious.

Now he's alot closer to the field than I am, but it hardly seems to me that people doing psychology or cognitive science or AI or anything else have been "insulated" from the people doing human neuroscience for the last forty years. My impression is that the two groups of fields are fascinated with each other, and have been for quite awhile. Even if we believe the functionalist position that psychology can't be reduced to human neuroscience, there are compelling pragmatic reasons to think about the hardware on which the one cognitive system we actually have in hand is in fact implemented; and my impression is that people have been doing that right along.

So I dunno. I'm not down in the thick of either field, so it's entirely possible that I'm missing the real subtlties or the concealed subtext of Churchland's piece; if any readers have deeper insights please send them along.

But anyway it's been fun thinking about it...   *8)

Sunday, April 3, 2005  permanent URL for this entry

So we sort of overlooked April Fool's Day around here (obRFCs: 4041, 4042), but since this whole week's log has the magic date in its very URL I felt I had to do something. I have therefore embedded a corker of an April Fool's joke in this very entry; see if you can find it! (In retrospect it's probably kind of obvious, but hey...)

The Teaching Company have again put out a couple of good free audio lectures: "How Are Popes Elected? Two Complimentary Lectures". I listened to the first one doing the grocery shopping just now, and it was interesting. Among other things, turns out the story about tapping the Pope on the forehead with the little silver hammer to make sure he's dead isn't true (or at least that's what Notre Dame wants us to think).

From Amptoons, the very cool Zoom Quilt. Now I'd like to be able to look sideways, please. *8)

Recommended as everyone's last readings on the media frenzy one slot back: the "funny" and "serious" links in this Jessamyn entry.

And from Reason, an amusing story about a French search engine in which the opinions of the elite will play the role that mere link-popularity does in Google. Although it's not clear how they're going to get the elite to sign up.

The flaws in the French plan are obvious. If popularity cannot arbitrate, what will? Mr Jeanneney wants a "committee of experts". He appears to be serious, though the supply of French-speaking experts, or experts speaking any language for that matter, would seem to be insufficient.

So in the latest J Phil there's a fascinating paper by Paul Churchland, in which he surveys the forty year history of functionalism, and describes why he thinks it turns out to have been wrong in pretty much every particular. I've read the first few pages, and my initial impression is that it rather misses the mark.

"Maybe Churchland's getting tired," I thought to myself, as I put the journal down and thought about maybe playing The Sims for awhile. And then I thought, "Maybe Churchland's getting tired? Here I am, knowing that I'm pretty good at like thinking thoughts and stuff, and that I theoretically enjoy advancing my own understanding, and even the world's understanding, of interesting stuff, and I'm about to start playing The Sims instead of reading the rest of this paper, and I'm thinking that maybe Churchland is getting tired? Ha!".

And then I thought about the celestial masters up in their crystalline control center looking down at the world and shaking their heads in sympathy and thinking, "Yeah, David's a little tired, all right. Such promise, and just sitting there playing The Sims. But after those last few lives, he deserves a rest in this one."

And that made me smile, but then I thought that maybe, after all, there aren't any celestial masters looking down at me and thinking that after all. And I looked at the Lord of the Rings calendar on the wall next to the computer where The Sims was loading, and Gandalf was looking out at me looking sort of worried, and I thought that maybe Gandalf there is just an actor, and if some wizard somewhere is looking worried about the fate of the world, maybe there's no narrator writing his worry down, and no eternal great conversation folding his worry into its developing body of wisdom, along with my tiredness and the color of the sky.

But then I thought that, after all, somewhere in the infinity of actual and possible and impossible worlds, somewhere in all the state-spaces in the past and present and future and off to the side in all the infinite number of conceivable and inconceivable directions, that narrator is inevitably hard at work, and that great conversation is going on. And the celestial masters in all their myriad forms are certainly looking down at me with sympathy.

And that was a nice thought.


earlier entries