Anyway, I’ve got to try to get yesterday’s main insight on the table before I lose track of it.
Ackerman: “Charlie Savage reports that a federal appeals court has reversed a lower court’s order to free a Yemeni national detained at Guantanamo on suspicion of being a member of al-Qaeda. Here’s his description of the reasoning employed:”
Judge Kessler examined each piece of evidence and found each insufficient to declare him part of Al Qaeda, arguing that flawed accusations cannot be assembled into a persuasive mosaic.
But Judge Randolph criticized that logic as “a fundamental mistake that infected the court’s entire analysis.” He argued that the evidence should be piled together as mutually corroborative because it is probable that a person with many suspicious indicators was part of Al Qaeda.
Spencer Ackerman comments:
From an intelligence perspective, that’s probably persuasive. From a law-enforcement perspective, that’s probable cause. (I guess. I’m not a lawyer.) But for evidence justifying indefinite detention without charge?
I share Ackerman’s doubts, though before last November and the “climategate” pseudo-scandal I probably would have gone the other way. (I suppose there’s no reason to hope for the present Supreme Court to improve on this. Rather, they will likely align as authoritarians with no thought for principle.)
The issue, fundamentally, is whether a “burden of proof” type decision vs. a “balance of evidence” type decision is indicated in any given case. As a person of Bayesian inclinations, I am generally inclined to a balance of evidence, and am frustrated by demands for “proof”, a beast which, it seems to me, exists either only in pure mathematics at best, or possibly not at all.
I think this question has a lot to do with how we respond to climate change and other sustainability questions. I think in matters of strategy, one uses a balance of evidence, but I now see that in matters of justice, one requires near-certainty.
Now, from a Bayesian perspective, twenty independent measures roughly indicating the same thing are collectively compelling even if no single one is. But we can’t use that standard in matters of justice, and the so-called “climategate” fiasco shows us why: prosecutors, law enforcers, or plaintiffs can trump up twenty misleading trails of evidence almost as easily as they can put together a single one. In a courtroom, at least there is a balance of effort on each of them. When the battle takes place on the battlefield of public opinion, though, it’s asymmetric warfare; the defense has far more work to do than the prosecution. However, even in the courtroom, it’s often the case that a committed prosecution has more resources to bring to bear, and the prosecution itself is not always ethical. When your opposition is a motivated person rather than indifferent Nature, the apparent “balance of evidence” can be totally misleading. That is why a burden of proof is necessary in a criminal court.
It may be hard for a scientist to understand the definition of “proof” that is needed in a courtroom; it’s certainly difficult for me to understand. But the standard of “beyond reasonable doubt” on any single thread of evidence, rather than “looks fishy” on a bunch of them, is there as a matter of justice.
As Einstein said, the Lord is subtle but He is not malicious. On the other hand, we have this word, “malicious”, so it must apply to somebody.
In doing science, we have, in Weiner’s sense
, an Augustinean devil as opponent; chaos, confusion, disorganization. As such, we can and should operate on the basis of a balance of evidence.
Unfortunately, we confront a culture which expects a Manichaean devil; the opposition is expected to be full of trickery and shabby false promises. Indeed, even our allies within the political culture can also be expected to be full of trickery and shabby false promises, trickery which the opposition will say reflects on ourselves. What the political culture demands is not constant adjustment to prevailing evidence but victory. Evidence is merely among the weaponry, and probably not key among the weapons.
The rational consequence for us scientists is to split the question:
- decide the science on the basis of a balance of evidence, and then
- contest policy as a matter of proof, arguing that the balance of evidence is as we state.
(As an aside, Anderegg, Prall et al directly addressed the latter question. Once we get to the latter sphere, we encounter as a matter of course brazen trickery of a sort that as scientists we are careful to avoid and reject within our own sphere. Which is why Anderegg, Prall et al was immediately and vociferously splattered with mud. Fortunately, the attack was so ludicrously overdrawn that the mud didn’t stick very well.)
We need to sort out the two types of debate, which for practical purposes (and to keep devils out of the picture) we can call frequentist and Bayesian. It’s totally obvious that what we are collectively doing on sustainability questions is excruciatingly far from the Bayesian optimum; that is, the rational expectation is that the result we are apparently headed for is far worse than the best we can do.
How to explain this to the public is stubbornly problematic. I am among those who believe that some of the more committed and sophisticated opposition is quite willfully and deliberately promoting a misunderstanding of the facts.
When they demand “proof” of this or that, they are applying an inappropriate model. We are stuck with a single-subject experiment; there is only one Earth (though there are many models that may or may not be suitable for purpose, they are not perfect) and so the pursuit of “proof” cannot be undertaken in the manner of a clinical trial.
When they engage us under the guise of pursuit of truth, they are exploiting a key weakness – that we are traditionally obliged to take all challenges seriously until proven otherwise. Unfortunately, that obligation needs to be rethought once we are dragged into the world where discourse is replaced by debate and balance of evidence is replaced by “victory”. We need to be aware of the tricks.
One of the key tricks of the opposition is a demand for a sort of “proof” that the problem itself cannot yield. As long as legislatures are dominated by lawyers and not scientists, this meta-issue itself requires understanding and careful handling.