|log (1999/12/31 to 2000/01/06)|
Thursday, January 6, 2000
In the Big Rock Candy Mountain
I've also put up a page with my collected postings to alt.hackers for your possible amusement. Hey, it was a good day for navel gazing!
I was a bit hasty yesterday, calling John McCain an idiot. For a major Republican candidate to say that we could one day have a homosexual president is a good honest thing. His claim that gays in the military are "disruptive to unit cohesion" is probably based on his actual experience in the military. Which isn't to say he's not wrong! But he's not a total idiot; I'd love to see a McCain-Bradley race. Although I'd still vote Libertarian.
We have our first surprising entry in the referer log for this page: Welcome to the surfer who got here via a Google search on "erotic stories incent". Be warned, of course, that incent is illegal in most states! *8)
I finally came out of the maze of hallways into a large open room, high-ceilinged, the north wall a single tall sun-suffused window overlooking the city. It was, I guessed, a secretarial or clerical pool; a dozen desks were clustered together in the center, taking up nearly all the floor-space. The room smelled of coffee, perfume, donuts, a pink cashmere sweater draped over the back of a swivel chair, but all the desks were unoccupied.
(Four Stars)Bovine Inversus is writing a novel. I like it so far!
It occurs to me that, in the O-machine discussion yesterday, one obvious line of argument is that consciousness itself constitutes an Oracle that turns the Turing machines of our physical brains into O-machines (in yesterday's terminology, it would probably have to be a point-source oracle). I don't think I buy it, but it's worth thinking about. I wonder if that's the conclusion that the "hypercomputation" people are trying to get to?
3:30 am: Arg! Two days in a row! This time it's not CGI hacking, but obscure philosophical debates, that are keeping my mind spinning. More on that in a bit.
In the "fictional weblog" thread, Geegaw points to some interesting material about a diarist who did turn out to be fictional, for some value of "fictional", and some of the fallout from it. I had at least one reader write to say it sounded like a fun idea, and there should be no obvious sign that such a log was fictional. I'm sure there are people who feel strongly the other way; do you?
The new referer log (I generally spell "referer" with one too few "r"s, in order to Assert My Individuality) so far is showing the obvious log-lists and LinkWatcher (which seems to be back now), as well as a couple of postings I made today to the Weblogs board on eGroups. No huge surprises yet!
I haven't noticed this posted all over the place (although it was apparently on SlashDot), so here is a link to the generous Linux hacker who restored services to gazillions of passport.com users, by paying Microsoft's overdue $35 domain-registration fee. Worth a smile...
So, about this philosophical question that kept me awake back there before dawn [random boldfacing off, and warning: this is going to be very long, perhaps really too long for a log entry, but it's what I'm thinking about right now; I'm very interested in the question of whether brains can be implemented in Turing machines, at least partly because it bears directly on the problem of consciousness, which is something I hope to have solved Real Soon Now].
In the latest (January, 2000) Journal of Philosophy, there's a paper by Jack Copeland with the snappy title "Narrow versus Wide Mechanism: Including a Re-examination of Turing's Views on the Mind-Machine Issue." It's an interesting paper, although I think its claims are ultimately wrong, or at least not significant.
The very strong form of the Church-Turing thesis is something like:
A Turing machine can do any computation that can be done by any machine.
Copeland, and apparently a small host of people in the general philosophy-of-computation area, think that this is incorrect in important ways. To some extent I think this is a straw man; people do tend to say stuff like that, but if pressed they will qualify the statement in various ways.
For instance, everyone grants that if you hook up a hardware random-number generator to a Turing machine, you get something that can do something that no Turing machine can do. But it's not a very interesting something; the only new computation that I know of that such a machine can carry out is "give me a list of random numbers". Not too exciting! (I wonder if someone like Greg Chaitin knows of something wild and interesting that a true source of random numbers would let a machine do?)
Copeland thinks that there are a whole range of interesting machines that can do more than Turing machines. But the only examples he give are wildly unconvincing. Turing himself talked about enhanced Turing machines, in his Princeton PhD thesis, published as:
Turing, Alan M. 1939. "Systems of Logic Based on Ordinals." Proceedings of the London Mathematical Society, ser. 2, 45 (1939): 161-228.
The method Turing suggested is to enhance a Turing machine by attaching an "oracle" to it, producing an "O-machine". An oracle is a box which can, when requested by the Turing machine it serves, replace the contents of a given piece of the tape (that is, a piece of the machine's memory) with something else. The interesting oracles, of course, are those which replace a given tape-content x with a representation of f(x), where f is some function that isn't computable by a Turing machine.
So for instance if you had an oracle that somehow knew the answer to the Halting Problem (the paradigmatic non-TM-computable problem), you could make an O-machine that could solve the Halting Problem. But how would you do it?
The only potential physical embodiments of an oracle that Copeland mentions in his paper are (1) an infinite lookup table containing the correct output for every possible input, and (2) a device that can make an arbitrarily-accurate measurement of the electrical charge of an object, together with an object whose electrical charge happens to encode the (infinitely many) correct values of f. Both of these cases basically reduce to eliminating the usual requirement on a Turing machine, that the tape start with at most a finite number of non-blank cells. Copeland's point reduces to saying that a machine can do something that a Turing machine can't do, if only you can somehow get an infinite amount of information hardwired into the machine to start with.
How could a machine come to have access to an infinite amount of information? It could be directly stored (the infinite-lookup-table approach), but of course in the real world no infinite storage devices are available, and there is still the apparently unsolvable problem of how the infinite lookup table gets filled in with the correct values in the first place! It could be implicitly stored, as in the electric-charge example that Copeland gives, but no such device works in the actual universe, since we can't measure any continuous physical property to infinite (or even arbitrary) precision (and of course there's the hard problem about how you find the right object to measure the charge on!).
It could also be extracted from a "point source": some physical object smaller than a machine that, due to basic physical laws, emits the information. That's how a hardware random-number generator, ultimately, works: somewhere deep in its bowels there is a quantum randomness that it taps into. But, referring again to the actual universe, we know of no "point source" of any non-TM-computable information except for streams of random numbers.
I still need to read much more of the literature on the subject (if anyone knows of a Web copy of Turing's thesis, or of other relevant literature on "hypercomputation", I'd love a reference), but it seems that against the claim:
A Turing machine can do any computation that can be done by any machine.Copeland is advancing only the uninteresting:
A Turing machine cannot do some computations that can be done by a machine that cannot possibly be constructed in this universe, or any other universe remotely like it.
For the purposes of wondering, for instance, if human brains are (or are sufficiently simulatable by) Turing machines, this latter is completely uninteresting, since it does not count against
A Turing machine can do any computation that can be done by any machine that can be constructed in this universe.and humans brains are of course constructed in this universe all the time.
So I dunno! If Copeland knew of more convincing oracles, he presumably would have mentioned one in the paper. But if he doesn't know of any more convincing oracles, you'd think he'd be much less strident about how "false" the Church-Turing thesis is, since after all he's just pointing out a tiny and not very interesting nit: that machines can do non-TM-computable computations if you're allowed to load them up with an infinite amount of information in advance. Which, of course, you aren't...
2:30 am: yet another sleepless mind-spinning sort of night. Probably because I'm finally planning to actually code this next thing:
New Viridian rant
from Bruce Sterling.
As I said earlier today in another context,
"Does anyone else think it's kind of ironic that these "manifestos"
about how everything should be rich and amusing and attention-grabbing
and not-taking-itself-seriously are long and linear and dry and soporific
and not particularly funny?"
So call me a curmudgeon...
I had a dream last night (isn't it awful when people tell
you their dreams?) that I was Loki or Coyote, the
mischevious spirit of cleverness and pranks,
and I was running from the humans, who had
started using technology and cunning to try to catch me.
As I ran, I thought, "why are these humans so much more trouble
than the other animals?".
I realized it was because of my gift, because
I had given them some of my cunning, that they were
causing me such grief.
Eventually one got me in the shoulder with an arrow, and I
Sometime later in the dream, perhaps no longer playing Loki,
I was trying to explain nanotechnology to a primitive Goddess of
love and marriage.
She was catching on pretty fast.
I had a dream last night (isn't it awful when people tell you their dreams?) that I was Loki or Coyote, the mischevious spirit of cleverness and pranks, and I was running from the humans, who had started using technology and cunning to try to catch me. As I ran, I thought, "why are these humans so much more trouble than the other animals?". I realized it was because of my gift, because I had given them some of my cunning, that they were causing me such grief. Eventually one got me in the shoulder with an arrow, and I was captured.
Sometime later in the dream, perhaps no longer playing Loki, I was trying to explain nanotechnology to a primitive Goddess of love and marriage. She was catching on pretty fast.
We have an answer to the "when did years AD really start?" question from Saturday. A reader (from "ankeny1.ia.home.com"; c00l!) points us at http://www.newadvent.org/cathen/03738a.htm, and Dad (Hi, Dad!) reports hearing roughly the same thing on teevee. The same story is told for instance here. So we have Dionysius Exiguus starting the whole thing in around 525 or so (and getting the origin point wrong by an unknown number of years.) Now you know! No one ever thought "it's now the year 500 AD".
I've thought now and then about putting up a totally fictitious Weblog/journal, for a real-sounding person who would have a strange and artful and controversial life. Would that be an Evil Thing? How obvious ought I to make it that it was fiction? Could I just register it with LinkWatcher and all, and have the "by the way, this is all fiction" notice somewhere down in the fine print on a subsidiary page somewhere, under big JPEGs of the fictional diarist, his girlfriend, his boyfriend, his tatoos?
Where has LinkWatcher gone?
The Editing Room has a very funny Abridged Script of The Matrix. (I haven't seen any of the other movies he treats; I don't get out much.)
This week's cider: Woodpecker. Once again, a nice cider with an empty Website. (Cider Jack's site is still empty, and still has that dumb slogan.) Could there be some Sinister Meaning behind this seeming coincidence?
Y2K sightings: On Sunday, the grocery was pretty much normal, except there was very little bread or milk. Milk strikes me as an odd thing to hoard! There was plenty of skim (non-fat milk); M suggests that this is because the hoarders were mostly hoarding for their kids, and kids don't drink skim milk. Sounds plausible!
I've now seen my first computer-related Y2K glitch: the FTP server on the machine that hosts davidchess.com returns "19100" in the year field when you ask it for the date and time of a file with a date in 2000. Ooops! *8) I can picture the code:
printf("19%02d",years_since_1900);rather than say
printf("%d",1900+years_since_1900);Not a big problem, though; I even worked around it semi-elegantly by subclassing Net::FTP to give it a new mdtm() method that accounts for the bug.
Back to work, back to the Web! Well, this all feels very familiar. Did anything interesting happen while I wasn't paying attention? The oddest thing about coming back to the wired world in the Year Two Thousand is that everything is pretty much just the same. Where's my jet-pack, my robot servant, my TEOTWAWKI, my Soma? Hmph! *8)
Watched the pretty lights and the enormous crowds and the lack of disasters on the tube last night, cuddled up in bed with the woodstove going. The little boy fell asleep under his Star Wars book at about 10:30pm, but the little daughter was awake and bouncing off the walls the whole time. Just to make the event extra-memorable, I managed to kick the half-empty bottle of sparkling cider under the bed, where it wetted some of the dust and ancient lost items. That was fun! *8)
I eagerly await the first opportunity to say "that's so Twentieth-century!".
The computers seem to work. The laptop here did crash mysteriously once this morning (claiming to be "dangerously low on resources"), but since it's running Windows95 it does that now and then anyway. The playroom machine was thrashing wildly and running very slowly, but that turned out to be some kid's game that we installed recently which, as a side effect, installed the evil Broderbund Brodcast Background Agent, which is a known Bad Thing. I ripped that out of the registry (for about the third time), and it was back to normal.
So have we all re-made ourselves and our lives for the new millenium? It would be a good excuse to change just about anything one wanted to change; but I guess I'm pretty content with my part of the universe. I considered at least a new design for the log again, but I decided against it again. Maybe in 3000. Is there nothing about myself I can resolve to change, improve, now that the calendar's so different? Nothing I'm willing to admit in public, anyway! *8)
New Millenium? I saw Arthur C. Clarke on CNN (via an amusingly unreliable Internet connection from Sri Lanka) saying quite rationally that the third millenium, as a matter of simple fact, doesn't start until next year, but that it's perfectly reasonable to celebrate this year anyway because of the 2000-ness. My own pet viewpoint is that since the Whole Problem is caused by there being no Year Zero, what happens is that the first millenium is simply missing a year (1-999), and the second and third are therefore 1000-1999 and 2000-2999. So we can say Happy New Millenium with a clear conscience.
Of course, calling a 999-year period a "millenium" is a little iffy. But the whole thing is a little iffy, since no one really thinks that any particular thing (including the birth of Jesus) happened exactly at the beginning of the year-numbers anyway.
Happy New Year to everyone reading, or not reading, this and all their loved ones, heirs, and assignees. May we all have a content, exciting, fulfilling next thousand years, and finally figure out how to be nice to each other.