« Robot Warriors, Bunker Busting WMD’s and the Future of Peace | Main | Perhaps He Shouldn’t Refer To Himself As A “Bomb Thrower” »

March 21, 2005



"just I" should have been "I just," natch.


No, I am a hypocrite, I swear.

Winston Smith

"This kind of legal memo should get someone thrown out on their ear, but instead it got Bybee a lifetime seat as a federal judge."
Yes. Reaching these sorts of conclusions in the past would lead to professional sanctions, or at least having to repeat Con Law 101.
Now it gets you a federal judgeship or control of the Justice Department.

Winston Smith

submandave: "when this same "70% to 90%" value is used to imply that "70% to 90%" of the prisoners abused in Abu Ghraib were falsely imprisoned there is a distinct difference between being retained for questioning and being incarcerated."
Granted the populations differ (not all Iraq detainees were at Abu Ghraib, nor were Guantanamo detainees the ones the Red Cross says the military sources were referring to), still, what's the difference between being falsely imprisoned and being held for questioning for an average of three years at Guantanamo, where relatively few of the five or six hundred "held for questioning" have been charged with anything? I mean, unless we dispense with the presumption of innocence?


Winston & Dave The Red Cross Report which I linked before uses the phrase "persons deprived of liberty" which sounds more inclusive in that it would include anyone who couldn't leave. I think the article changed it to something less cumbersome. However, it is still a made up number.

Winston Smith

Hmm. Does that phrase apply to the stop-lossed National Guard members too?



I can argue against the purity of your ideals on statistics, or knowledge, and what I do know of statistics supports your claims and your standards... --- however --- I can't give you evidence but only an impression... what do we do about gross estimates in the world, especially in politics. Sometimes we need them. You can not always get scientificly valid evidence, and yet you can get evidence.

as a skeptic I understand a call that we suspend belief but in politics sometimes you have to act and you have to act with the information you have (like Rumsfeld and armour).

We don't know anything about North Korea, we don't know if Kim Jong-Il every married his wives, we don't know what he's doing day to day, why he's removing portraits or closes the border for a week or much of anything about anything, but we'll still have to make decisions, we'll still assign probabilities such as "how likely is he to use Nuclear bombs?" or "how likely it is he has them ready to go now?" and so I wonder what is the avenue to use such poor "statistics" or "estimates" which by scientific standards are not really statistics or real estimates.

Just questions, I don't have any answers on that myself except an idea that we have to use information we have even when it is not scientifically valid.

Back on topic, it's hard to get accurate statistics on Iraqi public opinion right now... but if we get a sense of it anyway, shouldn't we let that inform our decisions?



There is something to be said for estimates, or even bad polls. Sometimes bad data that is, non scientific data is all you have and so you have to act on it, this is true. But there are varying degrees of non scientific data. There is data that is sampled poorly or poorly asked questions or incomplete answers, this data while not great, can still be useful if one recognizes it's limitations.

And then there is guessing, which is from my reading of the red cross report what I believe their military sources did. I am sure if the Red Cross had actual statistics on detanees they would have used them so their inclusion of the estimates makes sense. Unfortunately humans are REALLY unscientific. And in this case the problem of Availablility might apply, this means that it is likely a military officer would remember more vividly instances where he released individuals who were wrongly arrested, and he might overestimate how often this happens by large amounts. (People do this with plane crashes) However that is not to say that he is necessarily wrong, but it definetly does not mean that he is right.

My concern here is not the scientificness of statistics though. This number when interpreted properly could be useful, lets say we used the word A Lot or more than half which are approximate phrases. This I have no problem with. But when people throw numbers around, other people take them as facts. When I say a lot a reader will probably get the idea that I don't really know, but if I say 70% it is a whole different ball of wax. My concern is that people will repeat bad statistics and disseminate wrong information. Sometimes a number is not better.

The reason statistics are problematic is because people forget that they are ALWAYS estimates. Some estimates are much better than others, but the actual numbers are unknowable. And so while I do think that we should allow statistics to inform our decisions and opinoins I think that everyone should think about those statistics before they use/repeat them. What did we learn from Dan Rather kids? Always question your sources.

So to answer your above question, yes, we should still use it, but we should keep in mind that it is an estimate and that their might be problems with it, and allow it to "inform" us in a limited manner, we should definetly use multiple sources for any statistics on iraqi opinion and compare them. And remain open to the idea that they might be horribly wrong.

For the record, I know that I'm preaching to the choir.


>When I say a lot a reader will probably get the idea that I don't really know, but if I say 70% it is a whole different ball of wax.

interesting comment and very well reasoned and reasonable in general.

I've quoted part that stands out for me. This is the part that is objectionable. Use of a number instead of a qualifier like "a lot" means you bring in the aura of quantification and as you say, this is like an implicit claim you know what you are talking about.

I guess that's where I draw the line as well. I'm not sure about this particular statistic, but maybe that's the point... we'd need a better characterization to know if it means "a large number of anecdotal reporting" or something better than that.

The comments to this entry are closed.