5 ms·

As I pointed out, what is "terrorism"? If you start by saying that "ISIS = terrorism" then there's no predictive knowledge in knowing someone's ethnicity, beca

by dalke 10y ago

As I pointed out, what is "terrorism"? If you start by saying that "ISIS = terrorism" then there's no predictive knowledge in knowing someone's ethnicity, because your definition is inherently geographically biased, and ethnicity is correlated to geography. As you say, it's a tautology, but you think that's meaningful, when it's actually a decision made by fiat. And wouldn't nationality be more predictive than ethnicity? Are those who attack abortion centers and abortion doctors terrorists? If so, is "Christianity" another factor? If not, why not? Is the Charleston church shooting an example of terrorism? If so, is "being white" another factor? When the US-backed Contras used murder, rape, and torture against the civilian population, was that terrorism? And if so, what were the key predictive factors we could use to help identify terrorists and terrorist supporters? Who gets to choose what is terrorism, who gets to choose the factors that go into the predictive model, and does that choice have a built-in bias which makes the model useless for screening purposes? And again, why is it the least powerful who are often designated the most suspicious?

coldtea 10y ago
>If you start by saying that "ISIS = terrorism" then there's no predictive knowledge in knowing someone's ethnicity, because your definition is inherently geographically biased, and ethnicity is correlated to geography. I fail to see how it doesn't have "predictive knowledge" under those assertions. The fact that my "definition is inherently geographically biased" and "ethnicity is correlated to geography" implies exactly the opposite -- that it's an obvious correlation.(The predictive power obviously coming from using that knowledge for excluding, or putting a less weight, to other ethnicities). >And wouldn't nationality be more predictive than ethnicity? Not necessarily. Some western ex-pat working in such a country, even if he got the nationality, would be less likely to share the same affiliations, backstory, religion and culture with those turning out to be terrorists. And inversely, culprits in Europe for example often have gotten the European nationality (or even have been born there in the first place), but most still share such (non-European) ethnicities. >Are those who attack abortion centers and abortion doctors terrorists? If so, is "Christianity" another factor? If not, why not? They could be said to be terrorists too. I didn't suggest anything about them though -- I said racial profiling is OK at airports and for catching the other kinds of terrorists (and for which such programs as TSA started anyway), not for each and every kind of loony. Airports are not abortion clinics, so they won't attract as many Christian zealots. >When the US-backed Contras used murder, rape, and torture against the civilian population, was that terrorism? Yes, but obviously not the kind that the US government tries to prevent now -- so outside the scope of this discussion. If people from certain rural areas tended to be Contras and one wanted to guess them out in such a scenario as with TSA, then racial profiling would be totally applicable there too.
- dalke 10y ago
  Sorry, you're right. I meant that you are using ethnicity as a proxy for geographic location. For something as broad as "Arab", it's not predictive - just like "Christian" is not predictive for figuring out who might kill an abortion doctor, or "Japanese ancestry" is not predictive for who will sabotage US military installations on the territory of Hawaii. (Hint: none of them; and the US has apologized for the internment camps.) It's predictive in the way that saying African Americans tend to get worse grades in school. The primary correlation is with poverty - African Americans tend to be poorer than average - not ethnicity. So, even if you think ethnicity is important, how do you know that other characteristics aren't even more important? Look at the VAM for schools to see how difficult it is to be predictive even when you have lots of data points. > "Not necessarily" That's approaching a movie plot scenario. Either you have evidence or you don't. Hand-waving counter-examples is not evidence. > "I said racial profiling is OK at airports and for catching the other kinds of terrorists (and for which such programs as TSA started anyway)" Racial profiling is illegal in the US. Therefore you have no US evidence that it would be effective. How would you even measure effectiveness ? Fewer hijackings every year? Fewer bombings? For that matter, why do you keep picking race as the key predictive feature? Why aren't you arguing for sex screening? Look at all the male terrorists. Why aren't you proposing that women should automatically be allowed to use the TSA Pre-Check lanes, so the TSA can put more resources on searching the men? Once that's done, then argue for screening by race, which we know is less effective. > "They could be said to be terrorists too. I didn't suggest anything about them though" I know you didn't. That's because you're picking the lazy solution for the easy problem that doesn't really exist. Think of it this way. There are 10M people of ethnic group X in a country. Of them are 10 who plan to carry out a terrorist action. If we put racial screening in place, we can catch 2 of them. On the other hand, there's a 1-in-a-million chance that the added frustration, humiliation, and resentment will cause someone to snap, and decide to carry out a terrorist action. Congratulations, you now have 10 new terrorists to replace the 2 you caught. Yes, this is also approaching a movie plot scenario. But since that's all you really have to back your views, then that's what I'll use too.
  - coldtea 10y ago
    >It's predictive in the way that saying African Americans tend to get worse grades in school. The primary correlation is with poverty - African Americans tend to be poorer than average - not ethnicity. So, even if you think ethnicity is important, how do you know that other characteristics aren't even more important? I wouldn't say that it's a primary/secondary correlation scenario with terrorism too. It's more a case of terrorism being a specific historical outlet for both domestic and international political grudges and issues that those people (in the middle east and the arab world) have had, and a common thread to those is ethnicity/religion. So it's not like we're studying some abstract phenomenon and making blind guesses and accusations: we know the history of the area, and of such groups there, we know their composition, and ideology etc -- and in what we know religion and ethnicity plays a very large factor. They are neither many Icelanders nor many Catholics in those organizations (there might be a few "converts" -- but still insignificant). So, it's more a case of applying what we know as weight factors. So, if we know that "of known terrorists 90% are from X ethnicity area, and 10% from various others", then we should certainly take that into account as weights. And sure, you'll have false positives any way -- nobody suggests this would be the be all end all method of finding a terrorist, just a weighted factor. >That's approaching a movie plot scenario. Either you have evidence or you don't. Hand-waving counter-examples is not evidence. The counterexample comes from what we know historically about such cases. There's a thing called life experience, following the news and historical information, understanding how people behave in general etc -- and using that to tell what's more likely. Not everything has to come from some exhaustive official statistical study and "evidence" (and some things are also dead obvious to even need such a thing). >For that matter, why do you keep picking race as the key predictive feature? Because that's the very topic under discussion in this subthread. >Why aren't you arguing for sex screening? Who said I'm not? I've already said that multiple factors should be used and weighted. >Why aren't you proposing that women should automatically be allowed to use the TSA Pre-Check lanes, so the TSA can put more resources on searching the men? I'm not proposing it because this is something I've never implied I believe should be done -- and thus just a bizarro thing made up to make my arguments look stupid (as if this exaggerated suggestion directly stems from what I previously said). I've used the term "weight" several times in this sub-thread. Did I ever give the impression that any of these weights (like "being an arab" or "being a woman") should be 1? Or that any weight should be 0 for that matter? So, yeah, if they want to check for suspects I'd proposed they factor in "being male" among other things -- not to imply they shouldn't check women at all, but that they should check them less intensively, unless they have other red flags too. The very notion of "profiling" means you use all available info about likely attributes -- not as a certainties, but as hints. In cases where ethnicity has historically tended to be skewed towards certain X or Y groups, that should totally be a factor too. Like age, sex, personal income, nationality, occupation, and several of other things. >If we put racial screening in place, we can catch 2 of them. On the other hand, there's a 1-in-a-million chance that the added frustration, humiliation, and resentment will cause someone to snap, and decide to carry out a terrorist action. Congratulations, you now have 10 new terrorists to replace the 2 you caught. Who is making "movie plots" now (ok, I see you address that later)? That's (the "frustration of a TSA check" or similar minor event) is not historically (from all known cases) what such people have in their background. >Yes, this is also approaching a movie plot scenario. But since that's all you really have to back your views, then that's what I'll use too. I'm really arguing either self-evident things or facts we can gather from history. I made no hypotheticals about individuals and their personal motives (which could be various -- just observations about people generally act and for reasons that hold for the majority, and can be backed by what we know from all publicly known cases).

by dalke 10y ago

coldtea 10y ago

>If you start by saying that "ISIS = terrorism" then there's no predictive knowledge in knowing someone's ethnicity, because your definition is inherently geographically biased, and ethnicity is correlated to geography. I fail to see how it doesn't have "predictive knowledge" under those assertions. The fact that my "definition is inherently geographically biased" and "ethnicity is correlated to geography" implies exactly the opposite -- that it's an obvious correlation.(The predictive power obviously coming from using that knowledge for excluding, or putting a less weight, to other ethnicities). >And wouldn't nationality be more predictive than ethnicity? Not necessarily. Some western ex-pat working in such a country, even if he got the nationality, would be less likely to share the same affiliations, backstory, religion and culture with those turning out to be terrorists. And inversely, culprits in Europe for example often have gotten the European nationality (or even have been born there in the first place), but most still share such (non-European) ethnicities. >Are those who attack abortion centers and abortion doctors terrorists? If so, is "Christianity" another factor? If not, why not? They could be said to be terrorists too. I didn't suggest anything about them though -- I said racial profiling is OK at airports and for catching the other kinds of terrorists (and for which such programs as TSA started anyway), not for each and every kind of loony. Airports are not abortion clinics, so they won't attract as many Christian zealots. >When the US-backed Contras used murder, rape, and torture against the civilian population, was that terrorism? Yes, but obviously not the kind that the US government tries to prevent now -- so outside the scope of this discussion. If people from certain rural areas tended to be Contras and one wanted to guess them out in such a scenario as with TSA, then racial profiling would be totally applicable there too.

dalke 10y ago
Sorry, you're right. I meant that you are using ethnicity as a proxy for geographic location. For something as broad as "Arab", it's not predictive - just like "Christian" is not predictive for figuring out who might kill an abortion doctor, or "Japanese ancestry" is not predictive for who will sabotage US military installations on the territory of Hawaii. (Hint: none of them; and the US has apologized for the internment camps.) It's predictive in the way that saying African Americans tend to get worse grades in school. The primary correlation is with poverty - African Americans tend to be poorer than average - not ethnicity. So, even if you think ethnicity is important, how do you know that other characteristics aren't even more important? Look at the VAM for schools to see how difficult it is to be predictive even when you have lots of data points. > "Not necessarily" That's approaching a movie plot scenario. Either you have evidence or you don't. Hand-waving counter-examples is not evidence. > "I said racial profiling is OK at airports and for catching the other kinds of terrorists (and for which such programs as TSA started anyway)" Racial profiling is illegal in the US. Therefore you have no US evidence that it would be effective. How would you even measure effectiveness ? Fewer hijackings every year? Fewer bombings? For that matter, why do you keep picking race as the key predictive feature? Why aren't you arguing for sex screening? Look at all the male terrorists. Why aren't you proposing that women should automatically be allowed to use the TSA Pre-Check lanes, so the TSA can put more resources on searching the men? Once that's done, then argue for screening by race, which we know is less effective. > "They could be said to be terrorists too. I didn't suggest anything about them though" I know you didn't. That's because you're picking the lazy solution for the easy problem that doesn't really exist. Think of it this way. There are 10M people of ethnic group X in a country. Of them are 10 who plan to carry out a terrorist action. If we put racial screening in place, we can catch 2 of them. On the other hand, there's a 1-in-a-million chance that the added frustration, humiliation, and resentment will cause someone to snap, and decide to carry out a terrorist action. Congratulations, you now have 10 new terrorists to replace the 2 you caught. Yes, this is also approaching a movie plot scenario. But since that's all you really have to back your views, then that's what I'll use too.
- coldtea 10y ago
  >It's predictive in the way that saying African Americans tend to get worse grades in school. The primary correlation is with poverty - African Americans tend to be poorer than average - not ethnicity. So, even if you think ethnicity is important, how do you know that other characteristics aren't even more important? I wouldn't say that it's a primary/secondary correlation scenario with terrorism too. It's more a case of terrorism being a specific historical outlet for both domestic and international political grudges and issues that those people (in the middle east and the arab world) have had, and a common thread to those is ethnicity/religion. So it's not like we're studying some abstract phenomenon and making blind guesses and accusations: we know the history of the area, and of such groups there, we know their composition, and ideology etc -- and in what we know religion and ethnicity plays a very large factor. They are neither many Icelanders nor many Catholics in those organizations (there might be a few "converts" -- but still insignificant). So, it's more a case of applying what we know as weight factors. So, if we know that "of known terrorists 90% are from X ethnicity area, and 10% from various others", then we should certainly take that into account as weights. And sure, you'll have false positives any way -- nobody suggests this would be the be all end all method of finding a terrorist, just a weighted factor. >That's approaching a movie plot scenario. Either you have evidence or you don't. Hand-waving counter-examples is not evidence. The counterexample comes from what we know historically about such cases. There's a thing called life experience, following the news and historical information, understanding how people behave in general etc -- and using that to tell what's more likely. Not everything has to come from some exhaustive official statistical study and "evidence" (and some things are also dead obvious to even need such a thing). >For that matter, why do you keep picking race as the key predictive feature? Because that's the very topic under discussion in this subthread. >Why aren't you arguing for sex screening? Who said I'm not? I've already said that multiple factors should be used and weighted. >Why aren't you proposing that women should automatically be allowed to use the TSA Pre-Check lanes, so the TSA can put more resources on searching the men? I'm not proposing it because this is something I've never implied I believe should be done -- and thus just a bizarro thing made up to make my arguments look stupid (as if this exaggerated suggestion directly stems from what I previously said). I've used the term "weight" several times in this sub-thread. Did I ever give the impression that any of these weights (like "being an arab" or "being a woman") should be 1? Or that any weight should be 0 for that matter? So, yeah, if they want to check for suspects I'd proposed they factor in "being male" among other things -- not to imply they shouldn't check women at all, but that they should check them less intensively, unless they have other red flags too. The very notion of "profiling" means you use all available info about likely attributes -- not as a certainties, but as hints. In cases where ethnicity has historically tended to be skewed towards certain X or Y groups, that should totally be a factor too. Like age, sex, personal income, nationality, occupation, and several of other things. >If we put racial screening in place, we can catch 2 of them. On the other hand, there's a 1-in-a-million chance that the added frustration, humiliation, and resentment will cause someone to snap, and decide to carry out a terrorist action. Congratulations, you now have 10 new terrorists to replace the 2 you caught. Who is making "movie plots" now (ok, I see you address that later)? That's (the "frustration of a TSA check" or similar minor event) is not historically (from all known cases) what such people have in their background. >Yes, this is also approaching a movie plot scenario. But since that's all you really have to back your views, then that's what I'll use too. I'm really arguing either self-evident things or facts we can gather from history. I made no hypotheticals about individuals and their personal motives (which could be various -- just observations about people generally act and for reasons that hold for the majority, and can be backed by what we know from all publicly known cases).