(no subject)

[noahgibbs]

While I'm looking at Dan Savage, this Q&A is from this week's Onion AV Club:

I am a freshman in college and a virgin. However, that is about to change. I have been going out with my girlfriend for several months now, and we think we're ready to have sex. I just have one question: What do I do with the condom afterwards? Do I throw it in the garbage in my room? Do I go to the bathroom and flush it down the toilet, even though this requires putting on a robe and walking down the hall with a dirty condom in my hand? Or do I throw it out the window and hope it doesn't hit someone?
Condom Rookie

The only time in your life when you're allowed to throw a used condom out the window, CR, is when you're in college. And don't worry about hitting a passerby: Anyone walking under the windows of a dorm knows they're running the risk of being hit on the head with a used condom.

Comments

[angelbob.livejournal.com]

And I'm all for a good mistress. So is this article, also from this week.

[capnkjb.livejournal.com]

The Onion is so very wrong.

Ok, back to writing about a man who deliberately pissed people off by asking questions about common-sense activities.

[capnkjb.livejournal.com]

... and by wrong, I mean amusing and funny and sick.

Just in case the meaning was misconstrued.


SEE! That's what this guy's talking about (kinda)!

[angelbob.livejournal.com]

So is this Socrates, or merely one of his spiritual descendants?

[capnkjb.livejournal.com]

Oh, if only. Socrates was amusing in .. well, in a very similar way.

This is Harold Garfinkel, proponant of ethnomethodology, a field of study within sociology. It has a lot to do with the basic assumptions people make in interaction - like, take this fictional conversation:

Dad: I picked up Tim. He's better on the bike these days now.
Mom: That's good, honey.
Dad: Yeah, he's also a sneaky one.
Mom: Oh, he faked an injury again?
Dad: Yup.

There are a lot of unspoken ideas in this exchange. Garfinkel would ask, "Better how?", "How is it good?", etc.

In one study, he would reply in this manner to people who asked him how he was:
How do you mean, how am I? Physically, mentally? How have I enjoyed my day with regards to what things? I don't understand.

People were, to say the least, ticked after about 2 seconds of this.

The gist of his argument has to do with the idea that people in society create and maintain a sense of order, and this sense of order is not always said. It is a way of understanding reality.

Quite frankly, it makes my head hurt.
(Hurt how?)(Oh, fuck off, enough with the questions!)(Why is it enough? What meaning-)

Right.

[angelbob.livejournal.com]

Fair enough. I've dealt with a few of these questions, usually from the point of view of how ridiculously difficult it is to get computers to figure any of this out.

The Cycorp folks have put many years into a model for this, with mixed results. Mainly they've spent ten or twelve years feeding the damn thing facts about the world so it can try to make assumptions. It doesn't work very well.

[capnkjb.livejournal.com]

It is nearly impossible to make a logical machine function in an illogical world.

This is especially true of machines. It is well nigh impossible to make a hypocritical machine (you can get a person to love the death penalty for 10 year-olds but hate abortion, but you can't get a computer to do the same without programming it that way and seeting it to ignore all else).

[queen-elvis.livejournal.com]

I know people who would argue with you about that. Though I can remember Pace coming home and telling me Cyc said Canada was a subset of Canada.

[angelbob.livejournal.com]

Easier than it sounds, it just takes a lot more (and thus more specific) rules.

That said, you're right. Cyc actually has something like six different rulesets because if you dump them all in one bucket, they conflict.

[angelbob.livejournal.com]

True, actually, if Canada is a set. Any set is a subset, but not a proper subset, of itself. Yay, math.

Cyc is really, really good for a rules-based system. It's also a fine demonstrations of the quite severe limitations of rules-based systems in practice.

Had they done it all probabilistically it would be harder to train but more robust. Maybe the next group with a vision and ten years will do that now that the rules-based version has been done.