Current
Earlier
Later
Archive
Home
Search
Mail
Stuff
Bigger! |
2001.02.11 : 2001.02.17
Saturday, February 17, 2001
Rant. As I lifted my computer case out of the trunk
last night, it seemed oddly light. A quick check revealed the worst - I
had forgotten to put my laptop into it at the office, and I had now come
home for a three day weekend without it.
(Helen suggests that I blame this on her as I was, no doubt, so enamored
at the thought of coming home to her that I forgot all about my computer.
So that's my story.)
In any event, this clearly wouldn't do. So this morning, I dragged myself
out of bed at the crack of nine (a single digit hour, and on a Saturday
fercrissake) to drive an hour up to the office, and an hour back.
And why? So I could do my email play
Quake write Plurp. Yeah, that's it. So you folks
better appreciate the sacrifice!
Plurp. Am I the only one who says do email? Surely no
one back in the old analog days ever said I'm going to sit down and
do mail. Odd.
Plurp. The modern world is a violent place.
A resident of the South Bronx
said that his shooting spree there on Friday was "essentially a self-defense
operation." Bernard Hoage, 24, an unemployed night watchman, attacked and
killed five men south of the A&P supermarket on Longwood Avenue, the
first attack of its kind in nearly two years.
The shooting, Mr. Hoage said at a
news conference, was prompted by "the way they looked at me and my girl.
(It was) pretty (darn) clear I had to whack these (guys) or they might
have done something. Hey - it was self-defense."
Pretty awful, isn't it? Actually, I just made that up. Compare and contrast
it, however, with this actual
CNN article.
The U.S. military said a
U.S.-led airstrike on Iraq Friday was "essentially a self-defense operation."
The attack, in conjunction with British fighter aircraft, hit five targets
south of Baghdad, the first strike of its kind in nearly two years.
The operation, Marine Lt. Gen. Gregory
Newbold said at a Pentagon news conference, was prompted by an "increased
threat to our aircraft and our crew. It reached the point that it was obvious
to our forces that they had to conduct the operation to safeguard those
pilots and the aircraft. In fact (it was) essentially a self-defense measure,"
he said.
Yak. Dubya, describing good relations between our North American
neighbors.
It's just commonsensical.
He's just destined to be a wonderful source of material for Plurp,
isn't he? It's like having a news conference with Yogi Berra every single
day.
Yow. Utterly way cool pic
of the Earth at night. (Dave)
Pop quiz:
-
Is this way cool or what?
-
Who is that one lone guy in hopelessly northern Canada and why doesn't
he move south?
-
What time of day was this picture taken?
Discuss.
Yo. Good news. Those folks at despair.com
who trademarked the "frowny face" are offering
a compromise to all of us who bemoaned their locking up this useful
symbol.
"Starting today, I have authorized
our company to begin selling legally-approved sad emoticons to the global
community via our very own website, under the tradename Frownies(tm).
As a result, anyone who wishes to continue to use this legally-trademarked
logo in their email will now be able to purchase a supply from www.despair.com
and use them with our blessing, provided that they do not violate the terms
of their end-user
licensing agreement."
Unfortunately, the only person in the company that is legally allowed to
manufacture Frownies(tm) at this point is Dane. So supplies may
be limited. (Ron)
MANUFACTURING WOES: Dyslexic
high-school dropout Dane Burke struggles to complete the manufacturing
process for a limited-edition, super-sized "Frowny™". He will fail.
Plop.
Jeffrey Zeldman ... co-founder
and group leader of the Web Standards Project, or WaSP, says he's
fed up with people who cruise the Internet using outdated browsers.
"It's like we're television programmers
who still use black and white because not everybody has a color television
set," he gripes.
An odd simile, in that no one ever produced TV programs in black
and white for that reason. Why? Because color TV broadcasts were backwards
compatible with black and white TVs. If they hadn't been, color TV
might never have taken off at all. (Witness the mess with HDTV, which seems
to require serious government strong-arming to get adopted.)
Note to stuck-up Web designers: Your customers are not your problem.
If your ever-so-fascinating new features can't even motivate people to
get a ,
maybe that says something about their value.
 Plurp.
The blue dog wanted
everyone to paint
themselves
blue.
Friday, February 16, 2001
Blab. A reader consumed with our love life writes:
<< Washington?
Microsoft pretty much owns that >>
WHAT?? What about that pretty lady
you seem to lust after????? Where did SHE grow up??
La Jolla. Why do you ask?
Blab. A reader with no definite opinions writes:
Does the sympathy
you don't have for virus righters extend to those who made the whole
thing so blasted easy? I refer, of course, to those masters of innovation
in Redmond, who have craftily replaced any semblance of security on your
PC with Folgers Crystals.
To extend some small amount of sympathy to system designers, we must point
out that viruses can still infect, and spread, on systems with good (but
traditional) security - passwords, access control, etc. etc. The problem
is that the virus typically executes with your access permissions,
so it can do anything you could do. And that's pretty broad.
Some system designers are pretty good about this, though. The Java security
model is a paragon of foresight and excellence in this regard. It very
carefully restricts what a foreign Java program can do on your system.
And that's a very, very good thing.
 Having
said that, we must counterpoint out that system desigers can also be dumber
than stumps. Right up there on our list is Microsoft's fetish for including
a Turing-complete language inside every dopey sub-product they ship and
providing absolutely no access controls for them. Another work of rare
genius is allowing any program written in any of those dangerous languages
to manipulate absolutely everything on your system, and most especially
your email system. These two bits of intense cleverness are responsible
for enabling the current crop of self-mailing (and very fast spreading)
viruses.
So could you folks who design systems please stop doing dumb stuff?
Thank you.
Plop. Why do they refer to the removal of testicles from cats
as fixing them? Is he fixed? Gee, I didn't know he was broken.
Plop. Speaking of broken, our own dear Mayor, Rudy Guiliani,
still thinks that a city mayor has some role as an arbiter of artistic
taste. Maybe we need to get him fixed.
I think what they did is disgusting, it's outrageous, said Moody
Rudy of a photograph
of the Last Supper showing that Jesus was a black woman.
Hey, Rudy - calm down! I mean, they have a photograph of it,
OK? How can you argue with that?
Yo. Is ours the last blog in the world to link to Grammatron?
Probably. By the time stuff makes its way to Time,
it's pretty stale. Hypertext fiction mumble mumble.
Yow. Jasper
Dog. I love dogs. And I love Lileks'
writing.
Yow. Failed
palindromes. Funny! And for the same reason as broken
jokes, I think. (lemonyellow)
Yow. Warnings
affixed to laboratory doors at MIT. Of limited but nonzero comedic
value if viewed in the appropriate frame of mind.
If you experience any dizziness
or lightheadedness while working in the room, leave immediately, call 100,
and report to the Medical Department to be evaluated. They will notify
Facilities (3-4948) who will respond and check the air quality in the room.
Yo. In that quaint old analog snailmail yesterday, I found one
of those postcard things addressed to Richard White. It said:
SINGLE?
We attract the type
of woman you're looking for.
We've done the selecting.
You'll do the choosing.
M. CHATFIELD LTD.
Represents the people
of style and substance.
Yes, in fact, that is amazingly pretentious. But that's not the
point. The point is that Richard White was my grandfather and, if he were
still alive, he would be something like a hundred years old. But he's not,
which, I imagine, makes him somewhat the worse for wear in the dating pool
these days.
And that makes us wonder. What type of woman are century-old dead people
looking for? And how does M. Chatfield select for them? Do they have databases
of the desires of the dead? And, presuming that they are a business and
not merely a couple of macabre buffoons displaying their bad taste in public,
how do they tell their investors that they will make money from dead people?
How do they even expect dead people to "do the choosing"? Do they have
a Ouija Board Division?
Sadly, since they chose that wonderfully nostalgic postcard medium as
their interaction protocol, we couldn't click on it. They did include a
URL,
however, so maybe you can figure it out.
Plurp. I think that brain function is the Big Mystery
of computer science, and that progress in understanding brain function
will be a key to synthetic intelligence.
There are trillions of examples of intelligent systems. They are called
"brains", and we haven't built any of them. They work great, across
orders of magnitude of size and complexity, capable of sustaining the lives
of insects and humans alike in a complex, changing world.
We understand some things about them. We understand really simple systems
like the neural systems of aplysia.
We understand a bit of the optical cortex in higher animals.
But there is so much more we don't understand. There are angels in the
architecture that we don't understand at all, even at the level of really
small brains! I'm not trying to minimize this problem. But I am trying
to suggest that a solution to the problem of synthetic cognition will involve
an intimate interplay between system neurophysiology and computer science,
the former to understand the only working examples of intelligent systems
we have, and the latter to help abstract and adapt it to new technology.
Human brains can do amazing things: poetry, physics, mathematics. And
I think AI got off on the wrong foot by being so enamored with these high-level
mental miracles. They are, after all, built on top of a vast cognitive
infrastructure - almost all of which is subconscious and not subject to
introspective observation.
We shouldn't start the journey to intelligent system by leaping for
the highest pinnacles. Instead, we should start simple, by understanding
the principles that make simpler systems - small brains - work. As we understand
the workings of simpler systems, we are likely to find principles that
we can generalize to help us understand more complex systems.
In my opinion, the understanding and construction of intelligent systems
is one of the great remaining goals of science. But, as a scientific endeavor,
it will require a thriving experimental approach just as much as it will
a thriving theoretical approach.
Can we build systems that have synthetic intelligence without any understanding
of brain function? Of neurophysiology? Of cognitive psychology? It is comforting
to computer scientists to think that we can. But I don't believe it.
Plurp.
The blue dog wasn't
broken. Nope. Definitely
not.
Thursday, February 15, 2001
Blab. Referring to our reference to some
moron in Denmark who confessed to writing the Anna computer virus,
a reader writes:
Let me be the first to be
admit amazement at your lack of knowledge of the Western European map.
This individual is from the NETHERLANDS and NOT from Denmark. Get
it straight! Or get a proof reader! Who's the moron??
You have our permission to "be the first to be admit amazement".
But - hey - Denmark, Schmenmark; all those Scandinavian countries look
alike to us. Have we ever told you about glue states?
Rant. Did you know that there are lots of states in the United
States? Several dozen, I think. But you know what? I can't figure out what
most of them do.
OK. A few are easy. California has the movie biz, all that dot com stuff,
and nice beaches. Washington? Microsoft pretty much owns that, and I guess
that means they can put it wherever they want. Florida? DisneyWorld. And
New York is, well, New York after all.
But the rest of them? No clue. Look at Montana. It's huge! And
what's it there for? As far as I can tell, its primary function is to keep
Canada from sliding down. Alabama? What does it do besides keep Mississippi
and Georgia apart? Nothing. And New Mexico is just the place where they
put the left over sand from everything else.
So let's get serious. California, Washington, Florida and New York -
these are real states. They're functional. They actually do
something. Those others? They're just structural, just keeping stuff from
moving around.
So as you, like me, fly back and forth between New York and California,
I want you to stop occasionally between games of Alice,
look out the window at that vast expanse below and think to yourself, glue
states.
Yo. You've all been busy making astral clones, of course.
But did you know you can reduce
EJB network traffic with them? Well, now you do. (Jim)
Plop. Did you know that the "frowny face" has been trademarked?
And not just in the form shown below. The typographic sequence :-( is also
included. So pay up! (Ron)
 At
a press conference, Despair's COO,
Dr. E.L.Kersten, announced
his intentions to sue "anyone and everyone who uses the so-called 'frowny'
emoticon, or our trademarked logo, in their written email correspondence.
Ever."
Yow. The British Virgin Islands is an interesting place. It seems
to have dodged much of the chaos and churn of modern society. Here,
for instance, we learn that one of their top concerns is that too few people
know where the pickle fork goes on a proper table setting.
"We are full-fledged in a
booming tourism industry and table setting could be considered the backbone
of the industry. It is therefore essential that our waiters are properly
trained."
Must be nice, eh? (Helen)
Plurp.
The blue dog turned out to
be a result of the unholy
union of astral
clones and pickle
forks.
Wednesday, February 14, 2001
Blab. A reader with a very unusual name writes:
Happy Valentine's Day, Cutie.
I heart U xoxoxox
We are indeed flattered, Mr. Xoxoxox, but we are already spoken for.
Blab. A well-read reader writes:
Falsifiability
is due to Popper, not Kuhn.
D'oh! Sometimes we are so silly. But we always have the best readers! A
good discussion of falsifiability is here.
And here is Karl Popper's Logic
of Scientific Discovery, in which he discusses (introduces?) falsifiability.
Blab. A reader concerned with our oblique reference to a Turing
test for infants exclaims:
If the MY REAL BABY doll
can't even breastfeed, then I don't think it's very realistic at all.
Plus, where's the Colic Circuit that
causes it to scream uncontrollably for hours even when it's been fed, changed,
rocked, bounced, cuddled, snuggled, sung to, and entertained?
Oh wait, they must be waiting to use
that feature in the Teenage Model. You know, the ones that they make high
school students tote around for a week to teach them how much care a baby
requires, so they shouldn't have (unprotected) sex. (A scheme which is
notoriously ineffective).
This begs the question: how many realistic
features can you pack into a "lifelike" baby doll before it becomes so
much of a pain in the ass to take care of that little girls just take out
the batteries and leave it in the bottom of the toy chest?
The more realistic a baby doll is,
the more care it requires. For the crude analog baby dolls, one of the
best features is that it can be set down, with no adverse consequences,
anywhere, for any length of time...
And as the mother of a wee lass who
is just beginning to take interest in baby dolls, I must say that the ability
to just hose off the doll (if plastic) or throw it in the laundry (if cloth-bodied)
is very important to me when I'm making buying decisions. I'm guessing
that this MY REAL BABY is too
delicate for such treatment.
-Your lactation freak fan
Who knew the Turing test would be so hard, eh?
(And frankly, we've often wished for an off switch on real children.)
Blab. A reader concerned with the exact nature of cannibalism
- and aren't we all? - writes:
According to m-w.com,
cannibalism is defined as the eating of human flesh (not an exact quote.)
One does not eat blood, one drinks it; thus drinking human blood is not
cannibalism. Whether naso-, dermo-, dentro-, or cerebrophagia qualify
depends on one's definition of flesh (and for this, m-w.com
provides several differing definitions), but I would suggest they are all
more likely to be cannibalism than vampirism. (I.e., I reject your
theory that the more specific perversion overrides the general.)
That would make you a theoriadenegarian.
But what if you boil the blood down into a thick sauce and serve it
over rice? Does that count as eating? How about if you add lots of corn
starch to it and let it cool down so it's chewy? Does it count if you freeze
it around a popsicle stick?
Language is so ... so ... malleable.
(BTW, we generally prefer Bartleby
because you can link to the individual definitions.)
Plurp. Uh, sorry about that cerebrophagia
link yesterday. We just typed the word into Google and plopped the
link into the little box. We hadn't actually read
it.
Honest. Sorry. Eeeeewe.
Plop. It seems that some
moron in Denmark has confessed to writing the Anna computer virus and
has been arrested.
We have no sympathy for these folks. They could not avoid knowing what
would happen.
These people are the roaches of cyberspace. They need to check in. And
not check out.
Plurp. Some folks at UCLA have created a Virtual
Los Angeles. How can they tell?
 Plop.
Happy Valentine's Day. He Who Shall Remain Nameless celebrated the occasion
at 5 AM by chewing on some irises in a vase in the living room, knocking
over the vase and drenching a table and rug, then throwing up on the floor.
A good time was had by all
We're taking him to the vet today to get some shots. Not that he needs
them; we just want him to associate misbehaving at 5 AM with getting stuck
by needles.
Plop. What's up with Bovine's
Web site anyhow? Lately, the last week or so, it crashes our browser at
least half of the time we go there. QNC! Not good. What rude thing is he
doing, and how can we get him to stop?
 Yow.
Very very many very very cool
pics of Eros courtesy of NASA. Look through the older pics, especially
this
one, which shows the whole asteroid - about 20 km from near end to
far. Now imagine that whole thing, tumbling end over end as you try to
land a spacecraft on it that has no landing gear. And it worked.
Amazing!
These guys really are rocket scientists.
Plop. The Washington Post has been infected with the same evil
meme as the New York Times. Both of their Web sites let you see current
stuff for free (well, NYT makes you "register"), and that's a good thing.
But both of them charge you money to see stuff in their archives.
To us, that means we can't link to them here, because the links are
certain to rot within a day or a couple of weeks, in the sense that our
loyal readers would have to (gasp!) pay money to follow the links.
And that just won't do.
So we're looking for a good daily newspaper, of the caliber of the Post
or the NYT that (a) has its daily editions on the Web, (b) has a searchable
archive on the Web and (c) doesn't charge for either.
Readers?
Plop. Omigosh! The New Yorker (you remember them - it
was one of those paper-based analog magazine things) has decided to go
... wait for it ... online.
Imagine! You can now read words and stuff on the Web. Talk about
leading edge! According to some random New Yorker spokesperson:
It features some content
available only online but leaves other articles solely to the print version.
Is this a brilliant business model or what? As Keanu Reeves would no doubt
say: Whoa!
Yow. Oh look. An early computer
model of daydreaming. Neat!
Rant. In my humble, uninformed, non-expert opinion, "traditional"
AI is unlikely to yield a system that encompasses the depth and breadth
of human intelligence.
AI has been going strong for some 40 years. I'll try to divide this
"traditional" AI into two approaches:
-
Algorithmic approaches based largely on formal methods intended to imitate
very-high-level brain functions such as planning, logical reasoning, and
linguistic parsing.
-
Constructive approaches based on modeling very-low-level brain hardware
such as neural networks and connectionism in general.
Now the question is: Is this enough? If we simply continue to do what we're
doing, will it result in systems that can do a significant part of what
humans can do? I think not. I think that we are a long way from understanding
how to build human-esque systems, and it's not just a lack of MIPS or the
notion that we haven't encoded enough facts.
There is no reason to believe that the very-high-level approaches will
succeed. In general, they take approaches completely different from what
the brain does. Their approach is formal, well based in mathematics but
without an accompanying proof of sufficiency. Yes, humans do use logic,
but not often in the scheme of things. Yes, we do parse speech formally,
but hardly ever. In fact, a great deal of what we do is not conscious at
all, and everything we do consciously is built on top of a vast
set of brain functions that are subconscious.
Let's examine some examples. Deep
Blue was remarkably successful. It beat the world's champion human
at chess. Did it demonstrate that brute force computing can do amazing
things? Absolutely. Did it get us any closer to the goal of synthetic intelligence?
I would argue not. It taught us no generalizable principles that allow
us to characterize or build intelligent systems. And making the computer
faster won't do that either.
Statistically based speech recognition has done remarkably well over
the past couple of decades, as demonstrated by a number of clever people
in IBM and elsewhere. But I would argue that it is unlikely to extend much
further by itself because it lacks the context for utterances that humans
have - what
Lenat calls "common
sense".
So let's look at Cyc.
Cyc is a project to enter, pretty much by hand, the millions of "facts"
that we regard as common sense: water is wet, titanium is a kind of metal,
babies grow into adults, cars can move.
I don't think Cyc will provide a general solution to common sense reasoning.
The first reason I say this is that the methodology for entering these
common sense facts is flawed. Humans don't remember a bunch of isolated
facts. We remember cohesively connected sets of experiences - things that
"makes sense" to us. In a probe
of Cyc's capabilities in 1994, this difference was extremely evident:
It was way too easy to run into the edge of Cyc's knowledge because
its database of common sense facts was sparse and ill-connected.
The second reason I think Cyc is problematic is that humans don't answer
common sense questions by using first-order predicate calculus. Answer
the following question: "Was Abraham Lincoln taller than you?" Which ever
way you answered, you probably did so by imagining Lincoln, noticing he
was tall in your picture, and imagining yourself next to him. Our common
sense reasoning is based on our experience, our connected, subjective experience.
Indeed, I would claim that Cyc's current direction of using predefined
"contexts" to limit search and predefined "conceptual dimensions" in its
representation is an indication that Cyc will not solve the general problem.
Connectionism is at attempt to address the problem from the other end.
I suspect that, at the lowest levels, intelligent system will have a connectionist
structure. But that's like saying you can build a cathedral with bricks;
it doesn't tell you how to get there from here. And connectionist systems
(neural networks, etc.), by themselves, run into scaling problems right
away. Hugo DeGaris will
not succeed in building a synthetic brain because (a) He injects no structure
into his neural jello and (b) The universe won't be around long enough
for him to emulate evolution.
Connectionism looks at the bricks and wishes they were a cathedral.
High-level algorithmic approaches assume that only the formal properties
of the cathedral are important. Both approaches miss that vast middle-ground.
We do have a class of systems that has remarkable abilities - cognition,
learning, adaptation, the development of entirely new skills. We call them
"brains". And they don't work anything like the systems that "traditional"
AI has built. We have every reason to believe that brains can do these
remarkable things, and no evidence at all that the systems of "traditional"
AI will ever be able to.
Plurp.
The blue dog was
a strict vegetarian and would
never even consider
cannibalism.
Tuesday, February 13, 2001
Blab. A reader honing the gentle art of combining
Plurp
topics suggests the following creative way to spend
the upcoming holiday.
"We spent this Valentine's
day naming the cat."
Fortunately, that suggestion is not intended specifically for us, or we
would have to spend that I Heart You day in divisive debate. And that would
be bad.
Blab. In response to our desire to find a bumper sticker that
says I'd Rather Be Driving,
a reader helpfully suggests yet another method for us to exchange our scarce
money for yet more stuff.
You can get custom bumper
stickers
here.
In quantities as small as one (1).
Of course the cost per sticker goes down quite a bit when you get bigger
quantities.
They have a bunch of existing designs
you can order there as well. I liked some of the ones in the Political
category - I might just have to buy an "I didn't vote for his daddy, either"
for myself.
I found this site last week, as I
was thinking of making up bumper stickers to vent my aggravation at people
who fling lit cigarette butts out their windows. I was thinking along the
lines of "Unless you're mooning someone, keep your butt in the car" or
"This road is not your ashtray", but I haven't yet found just the right
mix of humor and outrage.
The possibilities are endless.
Blab. A reader concerned with our management policies writes:
<< An
attorney for [the firm] said Weinberger was not the victim of discrimination
because all the firm's employees were called names and made to wear costumes
if they were late. >>
So, what do you do when your employees
are late?
It's a research lab! We're not
sure what late means in this context.
Yow. Are geeks
excited by Eros? You bet we are! Very extremely way cool.
Yow. Why
I Am A Bad Correspondent, by Neal Stephenson. Basically, "I'm busy;
go away." Better even than blaming it on the ancient Greeks. (Dave)
Yow. Wanna know how fast your connection to the Internet really
is? Use the Internet
Speed Thermometer. (Best when connected at > 56k; self-referentially
slow otherwise.) (Bill)
Yow. Have you played with your own virtual fish at the Virtual
Fishtank? Somewhat fun.
Plop. Here's our nominee for Most Oxymoronic (Or Just Plain Moronic)
Corporate Motto:
Poland
Spring: What It Means To Be From Maine
Readers are invited to nominate their favorites.
Yo. It seems that this ex-president Clinton fellow was going
to rent a big office near Carnegie Hall in Midtown Manhattan, not surprisingly
a fairly fancy part of town, for $300k / year. The Republican folks wagged
their fingers at him, saying that was a lot of dough for the taxpayers
to fork over. So now Clinton is rumored to have found a nice 8,000 sq.
ft. office (10x the size of our apartment) in South Harlem, a rather more
downscale part of town, for a mere $200k / year.
Just for fun, I think the Republicans should try wagging their fingers
a few more times and see if they can get Clinton to end up in a bombed-out
crack house in the South Bronx.
Hey, it could work.
Yak. Lunchtalk started out innocently enough. Careening through
a discussion of the new movie Hannibal, we wondered whether drinking
human blood should be considered cannibalism. We decided it should not,
on the theory that any more specific perversion with a specific name (in
this case, vampirism) overrides the more general category (cannibalism).
Then it got weird, as we wondered which of the following qualified.
Nasophagia
Dermophagia
Dentophagia
Cerebrophagia
Reader opinions (and extensions) are solicited. (Fortunately, only that
last one has any Google hits. Or maybe that's not so fortunate.)
Plop. Bill wondered how many times the word Spudnik might
appear on the Web. I guessed over 20,000. He guessed under 2,000. He
won. What's wrong with all you people anyway?
Yo. Ian, venerable, analog
guy that he is, attempts to discern the propriety of goddamnit vs.
goddammit
by referring to the venerable,
analog OED. How quaint!
Us? We use Google!
We speculate that the m version predominates over the n
version because of the American Puritanical aversion to admitting to the
use of curse words, the misspelled m version not being the actual
curse word, you see.
Rant. IMHO, the unconstrained Turing
test is not the right goal for AI to pursue.
My argument is very simple. It is not useful to try to reproduce all
of the quirks of human psychology and cognition, yet that is precisely
what the Turing test would have us do.
Imagine that we spent the next 20 years in a wildly successful crash
program to build something like Lt. Cmdr. Data, the android from Star Trek.
It would be a great achievement, but it would certainly not pass the Turing
test. Data cannot tell jokes in a humorous way because he doesn't "understand"
humor. In short, while he does many of the amazing things that humans do,
he does not reproduce all of the quirks of the human brain/mind.
If we wanted Data to pass the Turing test, we would spend a subsequent
50-100 years adding in all of the human quirks we could find. And we would
learn very little of use in the process.
When Turing proposed his "imitation
game" 50 years ago, it was a simple, easy to state idea that focused
attention on the problem of synthetic cognition. But it shows its age and
today, it is at best a distraction. This case has been made by others,
of course. Here is a good reference, with a telling "random word" test
that shows why only something that slavishly follows the human brain can
pass the Turing test:
Why the Turing
Test is AI's Biggest Blind Alley, by Blay Whitby
Plurp.
The blue dog was
primarily concerned
with digiphagia.
Monday, February 12, 2001
Blab. A reader concerned with our romantic life writes:
Hmmm...what plans does Dr.
Plurp have for Valentines Day? If you need any idea's Im sure I can come
up with something.
We seem already to be booked for that day. But readers are encouraged to
suggest, more abstractly, interesting or curious things that one could
do for Valentine's Day.
Yow. A Zen crossword puzzle, courtesy of the infinitely clever
Bovine
Inversus.
Plop. Sometimes it's just not hard to see
the truth.
When he arrived late to work
... [Laurent Weinberger] says he was handed [an Adolph Hilter] uniform
and told to wear it as punishment for his tardiness.
Weinberger, whose grandmother died
in the Nazi concentration camp at Auschwitz, refused. A few weeks later,
he claims, he received a demotion and a cut in pay.
Weinberger quit and is now suing on
the grounds of racial discrimination, anti-Semitic abuse and unfair dismissal.
Tullett & Tokyo Liberty, [Weinberger's
employer,] which is backed by some of America's leading brokers, admitted
... that Weinberger was called racial epithets such as "Yiddo" and "Jew
boy." ...
However, the firm denies any anti-Semitic
abuse. An attorney for [the firm] said Weinberger was not the victim of
discrimination because all the firm's employees were called names and made
to wear costumes if they were late.
Tullett & Tokyo Liberty is one of London's largest securities dealers.
Aren't you glad you don't work there? (Rebecca)
Yow. Wazzup?
vs. What
Are You Doing? Compare and contrast.
Plurp. In a rare weekend blog entry, Dave
has written what looks to be a rather complex review of House
of Leaves:
But of course The Navidson
Record doesn't actually exist (in our world or in Truant's) except
within this fragment of analysis. Time will tell, I guess, whether the
analyses and explications of our world's "House
of Leaves" become any more real than the (fictional) body of work surrounding
the (fictional) film.
This kind of thing fairly begs for someone to write House
of Leaves for Dummies. Dunnit?
Yo. The folks at iRobot
(love that
name!) are doing weird stuff.
To adults, MY
REAL BABY is an interactive, robotic, artificially-intelligent, emotionally-responsive
baby doll. To kids, MY REAL BABY is simply the most lifelike, responsive,
magical baby doll ever.
Sorta like Aibo, but less barking?
Yow. Bad
analogies, allegedly from the Style Invitational Report of the Washington
Post, July 23, 1995. My personal favorite:
Her vocabulary was as bad
as, like, whatever.
Plurp. It was the headline that got me.
Deal
Could Turn Phones Into PlayStations
And the image in my mind was that of a roomful of people, seated next to
desk phones, the cords pulled, straining as they muttered excitedly into
their respective handpieces, something like this:
Bloop ... bip, bip, bip ...
Bamp! No, dammit! BAMP!!
And yeah, I know that's not what they meant. But still.
Plurp.
To adults, this was nothing
like an interactive, robotic, artificially-intelligent,
emotionally-responsive blue dog.
Sunday, February 11, 2001
Blab. An avid reader with an early bedtime and an unhealthy
obsession with the hour of the day, certain avians and internal
organs writes:
Liver minus One hour, thirty
six minutes and counting, eagle one.
The reader is invited to ponder the meaning of the word daily.
Blab. A somewhat later missive from a reader who may be that
same reader seems to indicate a conclusion of some kind.
Eagle base to Eagle one,
abort mission.
We seem to have our own little paramilitary following. Aren't we special?
Plurp. Here's some guy
who's filed lots of Freedom of Information Act applications to find out
stuff from the government. Probably quite a conspiracy theorist whacko,
but the documents he obtained might be real.
Hey - it's the Web. Ya never know!
Plurp. Compliance
With This Publication Is Mandatory. And we don't want to hear any limp
excuses like Oh gosh I didn't know, or I forgot to read it,
or I didn't think this meant me. OK?
Plurp. Running through my head tonight, a mutation of a meme
by Mark Knoffler:
A little bit of down will
get you up,
A little bit of up will get you down.
A little bit of down will get you
up,
A little bit of up will get you down.
A little bit of down will get you
up,
A little bit of up will get you down.
Plurp. You know, between the chicken and the egg, the egg must
have come first. It is, after all, where the genetic material that formed
the chicken came from. Mutations that evolved previous species into modern
chickens occurred almost exclusively in the process of forming and fertilizing
eggs, not in the subsequent development process.
I'm glad we cleared that up.
Yow. Three guys going home from a nightclub were arrested for
a crime they witnessed and did not commit. Why? Because, writes one of
them, a Harvard law student, they were black. He wrote about the experience
in this killer Village Voice article entitled Walking
While Black. He rewrites the Bill of Rights to portray the way
The Law treats black people.
Go read
it.
Yow. Some cool stuff our group is doing at work - software
agents that can outbid humans in auctions.
Plurp.
The blue dog definitely
came after both the
chicken and the egg.
 |
