Valid feelings, flawed foundations

[steuard]

The comments on my last post have helped me better understand my reaction to the mosque protests, so I thought I'd promote some of that to a followup post. (But this will be the last one, really!)

First, an important point that I hadn't thought about when writing my last post: whatever their ultimate cause, some people have very real feelings of pain and anger at the thought of a mosque anywhere near Ground Zero. Sure, the basis for those feelings is probably an irrational but very human generalization of negative feelings toward "Muslim terrorists" to "all Muslims" (much as I described before), but we all find emotions clouding our logic at times. Looking at it that way, some of my earlier comments were harsher than they ought to have been.

And in fact, I'd be more sympathetic to the folks objecting to this mosque if their emphasis were on their own feelings and their own limitations rather than on the "offensive" actions of the Muslims involved. I'd feel much more comfortable if everyone who objected simply said, "I'm sorry, I hate to admit this, but in this place I still have strong negative associations with the Muslim terrorists who caused me such pain. I know it's unfair, but I'm not ready to cope with a Muslim community center so close just yet." But that's not the majority of what I've heard.

What I've heard (as previously quoted) is that building a mosque there would be "offensive", as if the Muslims planning it were the ones responsible for the pain. They're not! It might be more sensitive of them to refrain from building there (particularly if the objections had mostly been made in the way I described above), but the underlying problem is not of their making. Holding rallies to tell the Muslims they're not wanted (to the extent of driving away fellow Christian protesters if they happen to be Arabs) isn't a way of saying, "This is about my feelings." And it certainly isn't a way of saying, "...and I know those feelings aren't fair and I'm working to get over them."

That is why all these protests and objections bother me so much (and, I think, why I don't accept the word "offensive" as remotely appropriate in this context). If there is a valid reason to ask that the mosque not be built near Ground Zero, it must be made clear by everyone that even making such a request is an imposition on the innocent Muslims who are entirely within their legal and moral rights planning the project. But I don't think that's the spirit behind these protests. (And backing that up, the NY Times just published an article about opposition to new mosques all over the nation, most in "far less hallowed locations" than the general vicinity of Ground Zero.)


As a final thought, it's important to remember that the unfair mistreatment of Muslims in our society causes pain and humiliation and harm, too. We have to balance two types of undeniably real pain: the feelings of those for whom Muslims evoke the specter of terror, and the feelings of those innocents who face daily suspicion because of it. I don't know the best way to handle that. It may or may not be right, but I tend to have more sympathy with those who are being unjustly vilified than with those whose (real!) feelings are based on a flawed generalization.

False associations like "the terrorists were Muslim, so all Muslims are terrorists" are responsible for some of the darkest aspects of human nature. I think it's healthiest for everyone if we as a society work to recognize and reject them. And that's why I find it so upsetting when public figures who make these statements are taken seriously by society and the media rather than being condemned.

Comments

[patrissimo.livejournal.com]

you have reached this false conclusion because your model is oversimplified.

First, an aside: you claim there is a 'huge correlation between "terrorist" and "Muslim"', which I find misleading. I think you're computing "huge" in a relative sense but then using it in an absolute sense. That is, a huge fraction of anti-US/anti-West terrorists are indeed Muslim, but only a small fraction of Muslims are terrorists...And that gets at the reason I suspect your claim is false. Identifying and stopping attempted attacks is not the only way to reduce terrorism. Over the long term, it is important to mitigate the factors that motivate terrorism in the first place.

I am going to be very meta-here and verge on ad hominem, but I am skeptical when people find roundabout causations (uncertain, complex, long-term) which point to the solution they are comfortable with and say these causations outweigh simple, direct, first-order effects which point to the solution they are uncomfortable with. It smells like someone trying to find an argument to prove what they want to be true. I do it too, and I hate it when people disagree with me with the type of argument I just gave, but I don't think that means it is wrong.

But I will engage with your argument's details as well.

More correct than huge association is huge relative risk. A factor of 100 (the Facebook thread) is a massive relative risk factor. Medical studies often find relative risks like 1.5 or 2.2 If you are building a statistical model for the CIA, and you find a feature which increased by a factor of 100 the chance that someone is a terrorist, that is an INCREDIBLY GOOD feature. It will almost certainly be the best feature in your model. Working with data sets at Google, you almost never find features that good. This is a huge statistical effect!

Give me that feature alone, and put me against an expert at analyzing someone's background who has a statistical model based on countries traveled to and country of origin and many other features, but who cannot use religion (and must remove religion correlates from his model - ie must control for religion and be evaluated on the controlled set), and I can fare well in a contest of "Pick the 0.1% of the population we are going to put law enforcement resources into studying and maximize the %age of terrorists who are in that 0.1%."

And this example shows why your " 1/10^3 and 1/10^5 on an intuitive level? (That is, in everyday life without explicitly doing math...I claim that on a practical level it is better in almost all cases to behave as if Muslims are no more likely to be terrorists than anyone else." is the wrong way to think about things, because this info doesn't get used in everyday life. Law enforcement has very few resources, they can only examine a few people. So what matters is: how can they pick the best people to examine, those most likely to be terrorists? It's not about increasing scrutiny on all Muslims - we can't afford that! It's about correctly picking the 0.1% of the population who the FBI runs more thorough background checks on, looks at the bank accounts of, etc. And in building that statistical model, the feature "Muslim" is going to be very very powerful.

This somewhat addresses your worry about targeting Muslims for heightened scrutiny, because I am not suggesting that we do this do a very large proportion of Muslims, so it won't piss off very many people. Many of those who are scrutinized will be Muslims, but few Muslims will be scrutinized.

I certainly agree it will help the Muslim world feel more threatened and mistrusted by America. And this has some negative effects. It also has some positive effects - like increasing the pressure on the Muslim world to self-police, making moderate Muslims ashamed of their radical brethren and more likely to speak out in the Muslim world - which is where potential terrorists are - against terrorism. I think potential terrorists are much more likely to become actual terrorists in a Muslim world with less self-policing than they are to become actual terrorists because the US searches Muslims more often and that offends them.

[steuard.livejournal.com]

This isn't really going to be an argued response, so it doesn't count as breaking my "this is it". Honest. :)

I am skeptical when people find roundabout causations (uncertain, complex, long-term) which point to the solution they are comfortable with and say these causations outweigh simple, direct, first-order effects which point to the solution they are uncomfortable with.

This is a Very Good Point(TM). I still suspect that I'm right, mind you, but there's an awful lot of psychology encouraging me to believe that whether rationally or not. Part of the problem is that I don't know how to quantitatively compare the expected harm of an increased probability of terrorism with the accumulating harm of discriminatory policies on Muslims that's happening every day. (And that's assuming that the net probability of terrorism actually does decrease as a result of greater scrutiny; you make a reasonable case for not going too far in the other direction.)

My final comment is that IF there's a way of increasing scrutiny on Muslims by law enforcement that doesn't substantially bleed over into increased hostility from the general population, and whose collective negative impact on the vast majority of Muslims who are innocent is clearly less than the expected benefit, then I'm fine with using religion in profiling data. But I'd want some sort of independent (and public!) study to verify those projections with high probability before each and every such use was approved. As you've pointed out, when I see a government institute this sort of policy I immediately start to worry whether there's a holocaust (or even just internment camps) waiting to happen.