This attempt to compress some thoughts about data sets, statistics and doubt in legal was triggered by Warren Agin’s mention of “quantum” in his note announcing Dan Linna’s ascension to the chair of the American Bar Association’s Committee on Legal Analytics.

Without claiming even minimal ignorance of quantum theory, there seem to be some connections with situations of doubt that pervade law. Phrasing this as a law school exam question:

Imagine a few situations where there is great doubt about the meaning of “the contract” between two persons:

  1. While visiting some merchant online, a person clicks on a privacy notice “accept” button, without even glancing at the terms written entirely by (and also for) the other party.
  2. After a dispute arises under a written, negotiated contract agreement, each party looks for an applicable provision, but the written contract is silent on that point.
  3. There is an applicable provision and it says only, “the parties will act reasonably.”
  4. The “contract” is a business contract and the parties each have a form for the transaction. Neither party reads the other, but they specify the deal terms and reference both of the parties’ respective forms, sign both forms and staple them together, each unread by the other party.
  5. The whole signed contract lists the deal points and then reads, “The parties reference a set of 1,000,000 prior contract agreements (whose checksum is 93mfit8 ; the “ContractBase”). The parties agree that all aspects of their relationship will be conducted in accordance with the median (not the mean) of the provisions of the ContractBase for similar transactions and situations.”
  6. The same scenario as before, but the reference to the ContractBase specifically includes any subsequent contract that also references the ContractBase. (I.e., the ContractBase learns from later uses in the way that a statutory provision learns from judicial decisions.)

A. What is the difference in the elements of doubt in each one?

B. What is the likely practical “impact” of each of the above as compared to: i) the others (1-5) and ii) what the parties might have negotiated between them, in their actual circumstances, with practical regard to some different situations of actual parties, their sophistication, bargaining power and access to information and ability to pursue a remedy?

You have one hour to clarify all this!

Back to another aspect of doubt – an astonishing thing about quantum theory is that it requires us to assume that there are many, many pasts. Some phenomena are explicable only if all of the things that could have happened, did in fact happen. These divergent pasts operate only on very small scale, but are pervasive, underlying everything. Weirdly, the different pasts get collapsed when examined – collapsed at least for the reality that the observers inhabit from that point forward.

Law has a rough parallel – doubt about the past. When a dispute is to be resolved, there may be two different stories about what happened, or more. Some elements will be relatively clear from circumstances, data trails, witnesses or written messages and documents. But other elements will be unknowable (probably at least one critical element will be in dispute if the matter gets as far as a formal procedure). We presume that at this level there is only one truth, but there are two or more stories about the past, a certainty that neither of them is the whole truth, no capacity for the decision-maker to absorb a “whole” truth, and no practical way to know which story is closer on what points. The decision-maker has to chose one story or the other, or some mix/match/slice off-spring of the two stories. Of course, this collapse of the two stories has meaning only for the parties themselves, not for the judicial system as a whole; it does not bind the judges or anyone else. In practical management of a relationship, one will often find oneself having to speculate what a median (mean?) judge (or jury) would think of what I can prove versus what the other party may have an interest in asserting. For clarity and likely also for efficiency, will (Q!). i) it usually be important to have a large body of elements that are specified, or ii) we could get by if we could accurately depict the big messy ones such as loyalty, honesty, concern for the other?

The notion of multiple pasts also happens at group levels, notably expressed as different narratives about the past. The conflict of narratives can become quite heated when interests between groups collide or when the collision emerges from a prior stable state of suppression, accommodation or compromise.

All of this connects, I suppose, in statistics or complex systems. One of the great advances in scientific thought arose from radically expanding the notion of relevant facts and greatly softening the notion of causation. First in economics the horizon was expanded to explain market efficiency (perhaps a bit too confidently). The agents in the market didn’t have any unified thought, but collectively they behaved, at least in part, as if they did. Then in biology – life thinks without thoughts – but one has to assume a past extending billions of years composed up of very tiny increments of time and vast numbers of agents. There is a similar thing in AI – probability, game theory, utility function, etc. Thinking without “thoughts.” Just as “thinking” might be the thing that happens in the gaps among thoughts.