Is There Something Wrong With "Profiling"?

[David Sklansky]

I just realized that this is a wonderful question. Because it brings to the fore two of the biggest mistakes people make when they are debating issues.

The answer to the question is "Of course there is". Any other answer is plainly ridiculous. Profiling is demeaning, unfair, and against what Americans stand for.

Yet it should be done. Equally obviously. As long as there is a good chance it will save a lot of lives.

But this post is not about profiling. And if people insist on hijacking this thread to talk about that subject (as they did, expressly against my wishes in my Glenn Beck tax thread) I might discipline them.

It just so happens that the profiling issue highlights the two big mistakes most people make when arguing.

1. Not acknowledging the strength of the opposing argument.

2. Not bringing in the mathematical subject of probability when it is appropriate.

The first mistake is seen all the time when abortions or stem cell research is brougt up. People don't seem to want to admit the the conclusion depends often not on their argument being right or that the opposition is wrong but rather that one good argument OUTWEIGHS the other good argument. Some people do see this, at least intuitively. Which is why a few anti abortionists are pro embyonic stem cell researcher. But most people strongly shy away from the tactic of admitting that their opponents arguments are strong but their's are stronger still. Is it because they don't trust their debating (or weighing) skills?

The other thing few debaters do, is properly bring in the probablistic aspects of an argument into that debate. For instance in the debates about caps on lawsuits or whether school buses should come equipped with seat belts, people rarely make a distinction between tragedies that are a thousand to one, a million to one, or a billion to one. But that is obviously wrong. If mandating school seat belts prevents a billion to one shot and costs multi millions, the money could save more lives elsewhere. Similarly if profiling reduces the chances of a tragedy from one in 20 million to one in 22 million one could easily argue that this tiny reduction doesn't justify inconveniencing or embarrasing a group of people. But reducing the chances from one in 10,000 to one in a million does.

Of course when arguments have a probability component to them, few people want to invoke them. Perhaps because they know that they (or their opponents)are so inept at it in spite of the subject's relative simplicity. Maybe one day I can change that.

[madnak]

The problem is that it's akin to showing weakness. You want to undermine the opponent's position as much as possible without allowing them to undermine yours. Sometimes its otherwise with friends, but not often on internet forums. Due to the anonymity, nobody can trust anyone else not to take advantage of such concessions so they're rare.

In other words, I think most people realize their opponents have strong arguments, but are reluctant to admit it for rhetorical reasons, they want to appear "strong."

[vhawk01]

[ QUOTE ]
The problem is that it's akin to showing weakness. You want to undermine the opponent's position as much as possible without allowing them to undermine yours. Sometimes its otherwise with friends, but not often on internet forums. Due to the anonymity, nobody can trust anyone else not to take advantage of such concessions so they're rare.

In other words, I think most people realize their opponents have strong arguments, but are reluctant to admit it for rhetorical reasons, they want to appear "strong."

[/ QUOTE ]

Because we cant trust that the other side will similarly concede that our arguments have merit. And the concession of any type of perceived weakness can make some types of poster essentially shut down and simply bask in perceived victory. Look at NotReady in the biblical errors thread: his concession that there is a good argument that there are some errors in the Bible is probably going to open himself up to a TON of criticism, some of it undue, and he is very unlikely to get the similar conception that a large chunk of the so-called biblical errors are either trivial or aren't really errors at all.

[David Sklansky]

I disagree with you two. If the goal is to persuade onlookers to the debate, it is not necessary that your opponent concede the strength of your arguments. Your willingness to do it (along with your explanation as to why those arguments aren't strong ENOUGH) will score points with the fencesitters.

[Speedlimits]

Stating that something is wrong, but then going onto say that it is necessary, shows a lack of understanding in basic philosophy. You maintain that profiling is wrong yet necessary. How can something be wrong if it is the only viable option given? Right and wrong imply a choice that is inherent within the concept. If something is necessary then there is no choice to be made, and it falls outside of moral guidelines (i.e right and wrong).

I agree with the rest of the post.

[chezlaw]

biggest mistake made by many is framing the debate to favour their strengths. Avoiding the hijack but the profiling debate is almost nothing to do with probabilities but about what we want society to be like and how much powere we are willing to cede to the profilers.

Acknowledging the strength of your argument - if the only goals are quantifiable like saving lives then it would be probabilistic (but they aren't so it isn't).


chez

[vhawk01]

[ QUOTE ]
I disagree with you two. If the goal is to persuade onlookers to the debate, it is not necessary that your opponent concede the strength of your arguments. Your willingness to do it (along with your explanation as to why those arguments aren't strong ENOUGH) will score points with the fencesitters.

[/ QUOTE ]

I dont know thats true at all. Election campaigns surely provide contradictory evidence.

And then there is the side-issue of what exactly each of us IS trying to accomplish on internet forums. I haven't come completely to grips with my own neuroses yet though.

[HeavilyArmed]

[ QUOTE ]
2. Not bringing in the mathematical subject of probability when it is appropriate.


[/ QUOTE ]

When apealing to the sensibilities of the general public this is never appropriate.

As I tell my genius son, the difference between you and an 80 IQ person is not too different from the difference between the 80 IQ person and a really smart dog.

[bunny]

[ QUOTE ]
I disagree with you two. If the goal is to persuade onlookers to the debate, it is not necessary that your opponent concede the strength of your arguments. Your willingness to do it (along with your explanation as to why those arguments aren't strong ENOUGH) will score points with the fencesitters.

[/ QUOTE ]
Not many of them - only the intelligent ones who can weigh the two arguments themselves.

If a dumb person was listening to two rather technical or complicated arguments. Person A says "my argument is very strong and the other guy's is weak" Person B says "person A's argument is strong but mine is stronger". The dumb person will just say - "both the experts agree that argument A is strong, they disagree whether argument B is strong. I'll go with argument A." It doesnt matter that not all dumb people will decide like this, more of them will decide to trust argument A than argument B, so it's a poor way to persuade dumb people.

I think you're definitely right about probability, people are too scared they will get it wrong, or that their audience just wont get it.

[Borodog]

David,

Is your middle name Obvious?

Joking aside, you are of course correct. The real problem is that so many people (I would say the majority of the population, since they went to public schools) do not understand probabilities or their importance at all. The news is filled with "man on the street" interviews with boobs who want policy X instated, regardless of the cost, "if it will save just one life!"

Demagogues of course play to this. They are usually advocating a policy position that favors a favored special interest group, but they are free to play the "If it saves just one life" card (and other, similar variants) because extremely few people understand the argument to be fallacious.

By the way, John Stossel has entire chapters of Give Me a Break devoted to this phenomenon.

[Iconoclastic]

Nothing wrong with profiling. Only wrong when the authorities profile certain groups and not others- like middle aged white males are not being rounded up en masse near schools even though pedophiles are nearly certain to fall into that demographic.

[Speedlimits]

[ QUOTE ]
[ QUOTE ]
2. Not bringing in the mathematical subject of probability when it is appropriate.


[/ QUOTE ]

When apealing to the sensibilities of the general public this is never appropriate.

As I tell my genius son, the difference between you and an 80 IQ person is not too different from the difference between the 80 IQ person and a really smart dog.

[/ QUOTE ]

What's your son's IQ?

[arahant]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
2. Not bringing in the mathematical subject of probability when it is appropriate.


[/ QUOTE ]

When apealing to the sensibilities of the general public this is never appropriate.

As I tell my genius son, the difference between you and an 80 IQ person is not too different from the difference between the 80 IQ person and a really smart dog.

[/ QUOTE ]

What's your son's IQ?

[/ QUOTE ]

Hmmm....
Son IQ = (Dog IQ, 160 - Dog IQ)

[arahant]

Is it really a hijack when you title the post 'is there something wrong with profiling'? It would be more interesting, ya know.

[revots33]

There's a reason why politicians never use probability when they debate issues. Most people don't understand it, and besides most people are motivated more by emotion than by reason (which is why attack ads and pandering to fear are so effective).

Take the school bus example. Most people will see children killed in a freak million-to-one accident on tv and demand something be done. There is very little thought to probability or costs vs. benefit.

In Long Island recently a drunk teenager sadly fell between the gap between a LIRR car and the platform and was killed. There will now likely be millions spent to correct this minor problem that could be spent in a much more useful manner elsewhere.

[andyfox]

"Of course when arguments have a probability component to them, few people want to invoke them. Perhaps because they know that they (or their opponents)are so inept at it in spite of the subject's relative simplicity."

I would suggest it's rather that they simply don't know the probabilities, or that revealing the probabilities would undermine their argument or jeopardize their agenda.

[soon2bepro]

I think a more important mistake than these 2 that people make when arguing, is something you already said elsewhere if I remember correctly. It's that many smart people who are right, think the truthfulness or logical correctness of the argument is important, when it is anything but.

Rhetoric, I'd say, is much more important. It doesn't matter what one says as long as it sounds right to the people watching.

And why is this? Because watchers are average people. And average people don't/can't apply logic properly. In any case, they're not looking to find out which statement has a higher probability of being true, they're just standing there watching and going with their strongest feeling. So usually arguing is all about making people feel good about your argument, and bad about your opponent's.

If anyone is interested, there's a recent comedy movie by Jason Reitman on the subject, Thank You for Smoking (2005), which I really enjoyed

[HeavilyArmed]

[ QUOTE ]
Hmmm....
Son IQ = (160 - Dog IQ)

[/ QUOTE ]

Just so.

[FortunaMaximus]

The better defined the probabilities of a specific set, the easier it is to find a solution to the set.

And if it's used in a way that makes certain that the purpose of the profiling is to cater to the individual needs, I don't see that it's a bad thing.

Vanilla psychology, well, isn't quite up to the task, so, yes, something along the lines of Asimov's psychohistory, even if it's fictional, has real merit in a benevolent global system.

But you have to start somewhere, and what the numbers do is just define another probability, which will fall into a category that gets well defined and has a much narrower margin of error with a larger data sample and more time.

It's constantly refining itself.

[MaxWeiss]

Profiling is a numbers issue and it is not wrong to use that numbers to make an immediate determination when no other data is present. I am friends with a black cop, and he said he hates to admit it, but profiling is REALLY accurate.

That being said, it shouldn't take much time to meet and sum up a person and alter your judgments once you take it to the individual level and have much more information on top of the basic profiling data.

[Mickey Brausch]

In my town there are 1000 inhabitants and among them are 100 blonde teenagers with acme. Out of the 20 people in the town jail at any time, on average, for committing a crime, 10 of them are blonde teenagers with acme.

Whenever a crime is committed, the police are paying particular attention to blonde teenagers with acme. Are they correct in doing so?

Whenever the police see anywhere in town a blonde teenager with acme, they search and question and ID him, way more than they do other citizens. Are they correct to do that?

Mickey Brausch

[luckyme]

[ QUOTE ]
In my town there are 1000 inhabitants and among them are 100 blonde teenagers with acme. Out of the 20 people in the town jail at any time, on average, for committing a crime, 10 of them are blonde teenagers with acme.

Whenever a crime is committed, the police are paying particular attention to blonde teenagers with acme. Are they correct in doing so?

Whenever the police see anywhere in town a blonde teenager with acme, they search and question and ID him, way more than they do other citizens. Are they correct to do that?

Mickey Brausch

[/ QUOTE ]

Hey, I used to live there! The police force was short staffed and had to really budget their resourses. There are 50 elderly ladies there who have never received a speeding ticket or jaywalked, model citizens.

Every time a crime was committed the police would question the 10 or so ladies who were in the vicinity. Iow, they dealt with the crime as if they had no knowledge of the statistics of people in the town.

Same problem in child disappearance cases. They treated everyone within 1 mile of the home and school as equal suspects, not putting any extra suspicion on the parents, family members, neighbors or prior offenders in the area.

oh, well, I've moved.

luckyme

[ALawPoker]

I love how everyone always ignores David's, rather frankly stated, purpose for the thread.

"But this post is not about profiling. And if people insist on hijacking this thread to talk about that subject I might discipline them." What was unclear about that?

[luckyme]

[ QUOTE ]
I love how everyone always ignores David's, rather frankly stated, purpose for the thread.


[/ QUOTE ]

hmmmm... my comments were about probablity, stated in the framework of profiling tha DS set up in the OP.

[ QUOTE ]
It just so happens that the profiling issue highlights the two big mistakes most people make when arguing.

[/ QUOTE ]

luckyme

[jogsxyz]

[ QUOTE ]
I love how everyone always ignores David's, rather frankly stated, purpose for the thread.

"But this post is not about profiling. And if people insist on hijacking this thread to talk about that subject I might discipline them." What was unclear about that?

[/ QUOTE ]

If this thread is not about profiling, DS should restate the misleading title.

[ QUOTE ]

It just so happens that the profiling issue highlights the two big mistakes most people make when arguing.

1. Not acknowledging the strength of the opposing argument.

2. Not bringing in the mathematical subject of probability when it is appropriate.



[/ QUOTE ]

Is this the subject of the thread or is it something else?

[ALawPoker]

[ QUOTE ]
If this thread is not about profiling, DS should restate the misleading title.

[/ QUOTE ]

"But this post is not about profiling." I think that addresses your 'if.'

DS seems to often title his threads to be inconsistent with the subject. Does it matter? The posts themselves make the intended subject fairly clear. I only mention it because I like the questions DS proses, and it's annoying to have to sift through everyone's hijacks about an issue that was only mentioned as an illustration of the point. But now I'm the one ranting, so I'll stop.


Luckyme, I wasn't so much addressing you in particular. I just used quick reply without thinking. Sorry.

[siegfriedandroy]

[ QUOTE ]
And if people insist on hijacking this thread to talk about that subject (as they did, expressly against my wishes in my Glenn Beck tax thread) I might discipline them.


[/ QUOTE ]

[Mickey Brausch]

[ QUOTE ]
[ QUOTE ]
And if people insist on hijacking this thread to talk about that subject (as they did, expressly against my wishes in my Glenn Beck tax thread) I might discipline them.


[/ QUOTE ]

[/ QUOTE ]My post was about the mistakes people make when attempting to be Bayesian and the excuses they proffer when they abuse Bayesian notions. The post used profiling only as an example. (BTW my profiling example was about "blonde teenagers with acme". Not "black Americans".)

Apart from this, you guys are spot on. /images/graemlins/smile.gif

[FortunaMaximus]

[ QUOTE ]
[ QUOTE ]
And if people insist on hijacking this thread to talk about that subject (as they did, expressly against my wishes in my Glenn Beck tax thread) I might discipline them.


[/ QUOTE ]

[/ QUOTE ]

LMAO.

[tolbiny]

[ QUOTE ]

2. Not bringing in the mathematical subject of probability when it is appropriate.

[/ QUOTE ]

How do you accurately assign probability to real life situations when comparing the harm that program A doea verses the benefits of program A? Using your example of racial profiling how much responsibility of the deaths of these three men (http://www.cnn.com/2006/US/11/25/nyc.shooting.ap/index.html) can you attribute to racial profiling? Is there a way to even reasonable calculate how often these events occur in a police force that encourages racial profiling V one that discourages V one that has a neutral stance. When the negative impact of a proposal cannot be measured as accurately as the positive impact, how do you proceed to assign probabilities accurately?

[ALawPoker]

[ QUOTE ]
How do you accurately assign probability to real life situations when comparing the harm that program A doea verses the benefits of program A?

[/ QUOTE ]

You don't need to be able to assign exact probabilities. I think the point is that the probability will generally be debatable, but one has to make an effort to estimate and consider it. You don't need to be exact, but any reasonable estimation carries more weight than the logical fallacy of "if program A results in more good than bad, then program A is certainly a good use of resources."

[tolbiny]

[ QUOTE ]

You don't need to be able to assign exact probabilities.

[/ QUOTE ]

The problem arises when you can get a reasonable measurable outcome on one side of hte equation and a non measurable outcomeon the other. Continuing with the example of Profiling- when Group A is trageted as more likely to commit crime B you can measure the rise in arrests and convictions for crime B, while not being causal can probably be strongly correlated to the efforts of profiling. On the other hand a large amount of the negatives of profiling are highly subjective. How do you measure the effects on a person who feels persecuted for his ethnic history? How do you measure the long term mental effects of a small group feeling singled out for persecution?

[ALawPoker]

[ QUOTE ]
On the other hand a large amount of the negatives of profiling are highly subjective. How do you measure the effects on a person who feels persecuted for his ethnic history? How do you measure the long term mental effects of a small group feeling singled out for persecution?

[/ QUOTE ]

Subjectively. That's where the debate comes in.