= 27d567bc2d48d9f40e9ccc20ea370b15)
Lumifer comments on Polling Thread - Less Wrong Discussion
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (118)
I disagree. I think that judging models by the success of their forecasts in empirical reality is precisely how they should be evaluated.
Why? Doesn't that mean you'll end up accepting all manner of placebo and frequently misunderstand the reason that something works?
You're mis-paraphrasing "forecasts", where I just said "techniques" in general. I think the distinction between pre and post hoc is important here.
If they are actually forecasts (as in, the prediction was made after making the model) then it does lend some credence...but if the model was made post-hoc of seeing certain techniques work and observing certain trends, as is the case with most of what we are talking about, it's another matter entirely.
The fact that I want to judge models by their effectiveness in the real world does not imply that I have to be stupid about it.
The Red models make forecasts: they tell you what to do and what results to expect, all in the future.
That's usually called historical data on which the model is based (or fitted) :-)
But is it your contention that Red techniques (regardless of whether the underlying Red models are correct or not) actually work? You seem to have been saying that they don't.
We all try to be stupid about it, but all sorts of biases will begin coming into play with nebulous criteria like that.
A model based on historical data becomes more trustworthy once it makes a prediction in a novel scenario. If the Phlogiston model tells you to predict that fire goes out when airflow is restricted, it doesn't increase the model's impressiveness because Becher knew that fact before he made the model.
Yes, I was sort of saying that earlier when I was talking about generalities. But I am also in agreement with your previous statement, that they work for a subset of people who have certain goals.
That looks like a general argument against any kind of empirical testing.
You join the Red team, go out to a bar, meet a girl you've never seen before. That is a novel scenario -- the alternative is to accept that the Red techniques work because of biological imperatives hardwired into all human females, something I think you'd be loath to do.
And not to accuse you of rationalist sins, but do you think there is some motivated cognition going on?
Red Theory: Go to the bar wearing a suit and rolex because it signals wealth and status, find a girl and take her home. Make an advance. She'll act reluctant at first, but she's just testing your dominance so you should make another attempt later on. She'll rationalize it later in the morning - you have to provide plausible deniability, so she can pretend that "it just happened" and enjoy herself without feeling like a slut. She'll be attracted to your willingness to assert your desire.
Blue Theory: The display of wealth did catch her eye - she was raised in a traditional manner and wanted to settle down with someone who could provide, and she'd always had a thing for suits. However, once she got back to your place, the woman really didn't want to, but she felt awkward, was alone with a strange man, and was too tipsy to make good decisions. She just wanted to get the situation over with and go home. She regretted playing along with it and not giving a more clear denial the next morning. Her self esteem lowered as a result of the encounter.
The above scenario plays out repeatedly in clubs and bars everywhere. Same sequence of events, and same predictions about what will happen...but very different underlying models of what is happening. Why should Red get to be the privileged hypothesis about what's really going on here? Especially when Blue seems more in line with what people say they have experienced in such situations?
(Also, "a technique produces the results it has produced in the past" is not the same as "empirical testing of a hypothesis")
What exactly did she go "back to your place" for? What were her expectations at the moment she said "OK, let's go to your place"?
Really? That just casts women as helpless powerless victims. Sure you want to go in that direction?
So, what's the Blue hypothesis? Why did the woman go back to his place? Why did she consent? Which falsifiable (in the Popperian sense) assertion about what really happened can Blue make?
You know about selection bias, right?
Sort of, yes. I think a large segment of humanity (women and men) are hopelessly unable to assert their own preferences, and if an assertive person tells them to do something, they just sort of...go along with it. This is even more apparent when alcohol is added to the mix. You don't even need to go so far as the Milgram experiment - humans are shockingly (pun intended) compliant to much weaker forms of authority. The aversion to conflict can override a lot of other preferences. It's kind of disappointing, but that's just how humans are.
And that's before you add in complications relating to impulse control and people's short term vs. long term preferences.
However, it's not really victimhood I'm describing here - what I described isn't quite at the point where I'd call it a crime scene. I'd say it's primarily a mis-communication of preferences between two parties, fueled by hyper-assertiveness on one end and conflict-aversion on the other.This is the reason that feminists advocate much stronger forms of communication and consent.
Yes, but it would be foolhardy to ignore available evidence in favor of some hypothetical speculation about what the evidence that slips my notice might be. In any case, this seems like the sort of scenario where I'd advocate everyone err on the side of caution by making sure to get verbal consent in the absence of strong, insistent application of social pressure, and most importantly to take reluctance at face value.
Well, that's interesting. To hell with who gets to screw whom -- what do you think about social and political implications of this? This approach basically says that democracy cannot be anything but a sham, for example. It also heavily implies that people need a benevolent philosopher-king to rule over them.
If a "large segment" of people have, basically, reduced capabilities, what does that imply about their rights?
Oh, come on, the common use of the word "victim" nowadays (especially in the SJ circles) has nothing to do with crimes and legality in general.
Pretty bleak. See: Third Reich. But I don't see how knowing that humans are vulnerable to do things they don't like when authority tells them to means we should have more central authorities - there's some inferences you've made that are opaque to me. If anything, I would think it means we should be all the more suspicious of authority and vigilantly maintain egalitarianism. (Of course, if you're somehow guaranteeing that they be benevolent and competent philosophers to boot, that does seem pretty good...)
I don't think there is any need to adopt any other group's broadened usage of "victim" here?