mathbabe

A few comments to the comments, in rough order:

//FoW comments in//

I agree with the spirit of what you wrote, but many of the details suggest misunderstandings of the actual law or exhibit a lack of awareness of legal scholarship. At the risk of sounding like I’m arguing poorly, I note upfront that I’m an attorney.
You say:

First of all, we would be adding opacity – of the mathematical modeling kind – to an already opaque system of law. It’s hard enough to read the legalese in a credit card contract without there also being a black box algorithm to make it impossible.

I actually don’t find this persuasive. Law’s opacity is of a different kind than the opacity of a mathematical model. In other words, there are plenty of contexts in which adding a “black box” somewhere in a legal system would not meaningfully detract from comprehension of the legal rule. As long as it’s clear what rule you have to follow, it shouldn’t matter that the law makes reference to the outcome of an algorithmic calculation (any more than it would matter if the law made reference to following a certain recipe as the method for determining whether something was “chocolate cake”). For example, certain courts have jurisdictional requirements that there be a certain amount in controversy. Since the rule is clear (“Have $X in play or take it to Small Claims”), there’s no reason not to put a black box around how the amount is calculated.

// I think the point is that “$X in play” is something that anyone can access and follow, but “and amount determined by a model developed by MIT graduates that you couldn’t understand” isn’t the same and would make the determination of venue largely incomprehensible to non-professionals//

You continue:

Right now lobbyists are used so heavily in part because they understand the complexity of their industries more than the lawmakers themselves. In other words, they actually add value in a certain way (besides in the monetary way). Adding black boxes would emphasize this asymmetric information problem, which is a terrible idea.

I’m not sure I understand this argument. What is it about adding algorithms to the process that is supposed to make it easier for lobbyists to unfairly manipulate information asymmetry? That is, it seems like you contradicted yourself. If lobbyists “add value” by “understanding the complexity of their industries more than the lawmakers themselves,” wouldn’t they only become more efficient at operationalizing that understanding if they could use algorithmic methods? Why is that supposed to be bad?

//I think Cathy meant that they add value to their clients, which in fact often results in subtracting value from everyone else in the world and thus is a bad thing rather than a good thing! Having watched the lobbying process in action, I’m inclined to agree with Cathy on this point.//

Further, you contend that:

So if it pertains to a law against shoplifting, say, you can’t have an embedded model which uses the proxy of “looking furtive and having bulges in your clothes.” You actually need to have proof that someone stole something.

But this not a good example of a context in which it even makes sense to talk about using algorithms in law. The question in a criminal larceny proceeding is whether the defendant took and carried away the personal property of another with intent to permanently deprive its owner of possession. The state has to prove each of those elements beyond a reasonable doubt to get a conviction. It makes no sense to talk about a conviction by proxy – the fact that a defendant looked around furtively and had bulges in his or her clothes is probative evidence, nothing more. There’s no opportunity for an algorithmic procedure here without fundamentally changing many, many other parts of how a criminal prosecution works.

//It’s not a good example, or maybe it’s a good example of a bad example. Let me try a slightly different example: the legislature of MA passes a new law labeled “attempted shoplifting”. The necessary elements of this crime are that a person engage in a pattern of eye-focus and tempo of limb-movement behavior which is flagged by an MIT algorithm. MIT was charged with the MA legislature with designing an algorithm that detects patterns of movement associated with shoplifting, and produced the algorithm in question as a result. Once the DA establishes (1) that the surveillance equipment was working properly, (2) that the defendant was properly recorded engaged in the sequence of eye and limb movements, and (3) that the computer software was loaded properly on the system, a guilty verdict and 6 mo-2yrs sentence is applied.

Cathy: what does the algo community think of this example? Is this the kind of thing they had in mind when they talk about algos in the law? Is this a dystopian twisting of what they intended when they talk about algos in the law?

Oblique: assuming I’m not creating a complete straw man, what’s your reaction to this new MA law both as a lawyer and as a matter of jurisprudence (policy)? Do you think the law stands to challenge on Constitutional ground (and if not, under what provision)? What unintended consequences when the MA legislature enacts this law?

Please limit all answers to 3 blue-book pages. A forced-curve will apply.//

You then go on to say:

And by the way, what counts as a black box is to some extent in the eye of the beholder. It wouldn’t be enough to have the source code available, since that’s only accessible to a very small subset of the population. Instead, anyone who is under the expectation of following a law should also be able to read and understand the law. That’s why the CFPB is trying to make credit card contracts be written in Plain English.

With respect, that’s not why that’s happening, and you seem to be confusing a credit card agreement with a statutory law. CFPB is trying to make credit card contracts be written in Plain English in the hopes that clearer disclosure of the awful, predatory terms of those agreements will discourage people from taking them. They simply don’t have enough power to do more, so they’re doing what they can.

//There is a fundamental distinction between statutory law (the relationship of the state to the individual) and the regulatory law the CFBP is engaged in (regulating the conduct of individual to individual. I think Cathy’s responding to the fact that the spirit of the CFPB proposals is similar to the fundamental need to have statutory law (the state) be published so that those subject to the law can know how to abide by it (no ‘secret laws’).//

Finally, you argue that:

Similarly, regulation law should be written in a way so that the employees of the regulator in question can understand it, and that means you shouldn’t have to have a Ph.D. in a quantitative field and know python.

Yet apparently you don’t need a law degree to have strong opinions about what law should look like!
This idea that “anyone who is under the expectation of following a law should also be able to read and understand the law” certainly sounds good, but I think it is ultimately misguided and a little naïve. It’s far, far more important that the law produce just results than that it be comprehensible on its face. For example, I could define all property crimes as “intentionally taking someone else’s belongings” and then just assign punishment by the cash value of the thing taken. This is relatively easy to understand. But it’s actually a worse system than a much more complicated set of definitions that recognize the numerous possible variations of such an act that really should be treated differently. Armed robbery vs. burglary. Petty larceny vs. large-scale embezzlement. Law’s complexity is much more a result of the factual complexity of the world than it is a result of the inability of its authors to speak clearly.

//I think there are some very deep questions embedded in this point. Oblique has a very good point about ‘necessary complexity’, although I wonder if there would be disagreement with the statement “having the law written in a way that is accessible is a very important goal and departing from that goal by having opaque or inaccessible law has significant negative consequences that should be kept at the forefront of mind by regulatory and legislative framers.”

“Law’s complexity is much more a result of the factual complexity of the world than it is a result of the inability of its authors to speak clearly.”—well, maybe it is and maybe it isn’t. Here I think Oblique is giving too much benefit of the doubt to regulatory and legislative framers, because, in my experience, there is also a shit-ton of ‘unnecessary complexity’ in statute and regulation for a whole host of reasons ranging from historical accident to plain old incompetence to overwhelmed staff to deliberate manipulation by vested third-party lobbying interests.//

FoW

LikeLike