[lisper]

Snideness is about HOW you say something, not WHAT you say. "What is the statistical significance of a YouTube video?" is a snide comment completely independent of the question of whether or not a YouTube video actually has statistical significance (whatever that could possibly mean).

BTW, your claimed math literacy is not much in evidence here. Of course opinions have no "statistical weight" (whatever that could possibly mean). Only data has statistical significance. So let's examine the data. So far in this discussion I've offered up three data points:

1. A video (hosted on YouTube, though I fail to see how that could possibly be relevant) of two individuals, one of whom claims to be a lawyer and another who claims to be a former police officer, both of whom say that it is unwise to talk to the police under any circumstances, and explain why in considerable detail.

2. The well-documented fact that the Innocence Project has to date exonerated 280 people.

3. The fact that Troy Davis was recently executed for a crime he almost certainly did not commit.

Are these data points statistically significant? I have no idea. I haven't done the math. But 1) neither have you and 2) the burden of proof is not on me. I am only arguing that it is not unreasonable to be wary of law enforcement. If someone wants to dispute that, the burden is on them to show that law enforcement is trustworthy. So far no one participating in this discussion has offered up EVEN A SINGLE DATA POINT in support of that position. So my three data points may or may not be statistically significant, but (and here comes a fine example of a snide comment) I dare say they have better prospects than your zero data points.

[ianterrell]

I shall largely ignore your feigned obtuseness (despite claiming not to know what things could possibly mean, you do seem to get the gist quite well) and get to the relevant section. As an aside, I note that I'll do my best to eradicate any trace of vagueness which the rhetorically inclined like yourself enjoy seizing upon—but that's most usually an unattainable goal.

> Are these data points statistically significant? I have no idea. I haven't done the math. But 1) neither have you and 2) the burden of proof is not on me.

Forgive me for not showing my work. Contrary to your claims that I offered zero data points, I did mention that there are over three quarters of a million law enforcement officers in the US. That would require a randomly selected sample in the several hundreds to determine with any degree of accuracy the character of the population.

You will likely counter that you're not arguing about the character of the population, in fact, you just wrote:

> I am only arguing that it is not unreasonable to be wary of law enforcement.

I must assume you're retracting your earlier statements, then. The ones that started with, "You assume that the police care if you are guilty or not. They don't." and continue on for several more sentences that speak to the character of the population as a whole, notably including an alleged desire to convict anyone regardless of guilt and strongarm confessions.

My point with the above is that if that were an informed opinion based upon data the holder would need to see a sampling of several hundred randomly selected officer/citizen interactions and see the majority of them end in a disregard for the rule of law, courtesy, etc. To my knowledge, that doesn't exist. Partially for difficulties I mentioned originally: nonuniformity of the population, selection bias in accounts, etc.

I shall perform an additional back of the envelope calculation. The Uniform Crime Report for 2010 shows over 13 million arrests that year. Even considering the total number of arrests appears to be declining year over year lately, I can assume relatively safely that in recent years (8-15 roughly) there have been over 100 million arrests. That is a large number of interactions between police and the citizenry.

While I don't have a number for convictions, I will assume it is also rather large. Wikipedia lists over 7 million people under correctional supervision, so that's at least a floor. Do I claim that the false conviction rate is 280/7 million? No, but still (and especially considering that 7 million is likely to be an exceptionally low estimate) my intuition tells me that a random sampling of convictions would find the vast majority to be not wrongfully convicted through law enforcement malevolence.

Your fear and anger toward the general population of law enforcement are not supported by the data. Note that that does not discredit the emotions themselves—they're worthwhile and valid. I even feel them in cases of police abuse, such as Troy Davis and, for instance, the unwarranted pepper spraying of the UC Davis students. I would only caution you against letting those emotions override your reason and thereby jumping to conclusions that the data does not support.

> If someone wants to dispute that, the burden is on them to show that law enforcement is trustworthy.

There's nothing wrong with the opinion that the burden of proof should fall on someone other than one's self. However, I might add that in practice the burden falls on the contrarian. And when one's opinion is against one of the established pillars of society—for better or worse, with no judgment implied—that person is nearly by definition the contrarian.

—EDIT: And just for fun, from a quick Google trying to get better numbers, I'll point you to a real study on the topic: http://www.amazon.com/Convicted-but-Innocent-Wrongful-Convic...

A description of it (http://researchnews.osu.edu/archive/ronhuff.htm) describes the arrived at number of wrongful convictions to be around 0.5%. More that half of those were caused by eyewitness misidentifications, leaving less than half to be caused by police malfeasance. The authors indicate that this is probably low, but: even doubled or tripled or quintupled, it seems that the data indicates the vast majority of law enforcement are not conspiring to convict everyone and anyone for a crime.

[lisper]

> I must assume you're retracting your earlier statements, then

No. The reason it is reasonable to be wary of law enforcement is because the police don't care if you're guilty. Let me be more precise about what I mean by that. It may well be that most individual police officers really do care if you're innocent. But on a systemic level they do not care. They are trained not to care. Their job (they are told) is not to determine guilt or innocence. That (they are told) is the prosecutor's job. (And the prosecutors usually pass the buck to the jury.) The police are not rewarded for letting innocent people go, they are rewarded for making arrests on probable cause. (And prosecutors are not rewarded for letting innocent people go either, they are rewarded for securing convictions.)

I did not mean to cast aspersions on individual law enforcement officers. I'm sure most of them are fine upstanding people. They work hard. They put their lives on the line. They deal with the scumbags of the world so the rest of us don't have to. I respect them and I'm grateful that they are there.

But the police (and prosecutors) work within a system that actively discourages them from thinking about whether you are actually innocent (except insofar as your actual innocence might prevent them from convicting you) and encourages them to think instead only in terms of whether they can collect enough evidence to prove your guilt beyond a reasonable doubt in the mind of a jury, or to present to you a sufficiently credible threat of being able to convince a jury so that you'll accept a plea bargain. Your actual guilt is (and by their training ought to be) irrelevant to them.

BTW, there's a defensible argument that the system ought to be this way, which is that determining actual guilt or innocence is really, really hard, and individuals are easily swayed by emotions and other irrational factors (in both directions BTW) and this system is designed so that no one individual's irrationality can affect the outcome too much. Flawed as it is, it is arguable that it's the best we can do given human nature.

Postscript (since I just saw your edit): a 0.5% false conviction rate is very high. There are about 8 million convicts in the U.S. At 0.5%, that would be 40,000 wrongly convicted. That's about 500 years worth of lightning strike casualties.

> the vast majority of law enforcement are not conspiring to convict everyone and anyone for a crime

Clearly. But a conspiracy is not the only way to produce bad outcomes.