Wednesday, September 30, 2015

Abstract Human

I'm going to present a psychological model. I've never seen it in a professional publication (nor have I looked), I have no hard evidence for it, but I do believe it to be true.

According to the Machiavellian intelligence hypothesis, the main problem which drove human brain evolution was predicting and outmaneuvering other human brains. Unsurprisingly, this evolutionary process left us with specialized hardware for modelling other humans: mirror neurons. When we see someone else doing something, mirror neurons fire in our own heads to simulate the activity. When you "put yourself in someone else' shoes", imagine yourself as someone else, you are using your mirror neurons.

Let's create a more abstract model of the mirror neuron. We start with a black box representing the human brain. The box is quite complicated, and to this day we do not understand its internal functions. We do know that there are rather a lot of these boxes in the world, not identical but quite similar. The boxes constantly talk, compete, cooperate, scheme, fight, bicker, etc...

Each box is equipped with advanced planning capacity. A box can imagine hypothetical environments, and imagine what it would do in those environments. (The technical term for these what-if scenarios is "counterfactual scenario"; we have a very firm mathematical understanding of them). The box runs its usual programs within this counterfactual world, and the output is its own behavior in that environment. This is very helpful for the box to make plans. In humans, we sometimes call it "daydreaming" when one spends too much time in counterfactual mode.

A related piece of hardware allows for even more advanced planning: the box can simulate other boxes. Of course, other boxes are extremely complex, so they cannot be simulated directly... but because the boxes are so similar, they can be simulated directly on the hardware of any single box. A box simply goes into counterfactual mode, changes a few internal parameters to simulate another box, and then runs in the counterfactual world normally. The box keeps detailed lists of parameters to change in order to simulate each of the boxes in its social circle. These internal parameter change lists are the intuition underlying what we call "personality".

Now we get to the interesting part. Turns out, the box has an internal change list representing itself. Remember, all this hardware evolved primarily for modelling other boxes. When the box goes into counterfactual mode, a change list is applied automatically. Not having any changes at all is not an option. Some of those changes are overriding components normally attached directly to the physical world; they must be circumvented in order for the counterfactual processing to remain counterfactual. So, if the box wants to model itself, then it needs a change list like any other. This change list is the box' abstract social representation of itself. We might even call it the box' "identity".

Notice that the contents of the box' self-representing change list need not be accurate. It's a change list like any other, representing boxes as seen from the outside. The self representing change list is learned, just like the others, by observing behavior (primarily social interactions). Of course, the self-representing change list is used in virtually all planning, so its contents also affect behavior. The result is a complicated feedback interaction: self-identity informs behavior, and behavior informs self-identity. On top of that, self-identity also learns heavily from interactions with other boxes. If box A and box B have very different change lists for box A, then box A will behave according to its own list, but will simultaneously update its list throughout their interaction to account for box B's representation of box A. Oh, and A might sometimes change a parameter or two in its self-identity just to try it.

Ok, deep breath. Direct modelling of interactions is definitely going to be very, very complicated. Let's ignore that problem and consider another angle.

What if there are subpopulations with similar parameters? Then the box can simplify its change lists by keeping a single change list for the whole group of boxes with similar parameters. This single change list applies to the whole group; we might call it a "group identity". Of course, any box may belong to multiple groups. A change list for one box might look like "Apply change lists for groups X, Y, and Z, then apply all the following individual changes...". In practice, change lists consist mainly of group memberships. Special-case changes are less efficient, so we try to avoid them.

This means that a box' self-identity also consists mainly of group membership (although research shows most boxes are much more tolerant of special-case changes in their self-identity). And remember, the self-identity, like any other, is constantly learned. So a box can change its self-identity by changing its group membership, or even just pretending to change its group membership.

Notice that both group membership and group parameters are constantly learned. So, if box A suddenly starts wearing leather jackets, nearby boxes will update their change lists to increase parameters which tend to cause leather-jacket-wearing, including group memberships. In fact, A itself will update its own self-identity based on the new clothes. As many people have observed, trying on new clothes is trying on a new identity, and a change in clothing style will cause a change in behavior. Clever companies even take advantage of this in their ad campaigns; the Converse shoe-box company is especially good at it.

A Solution for Today's Legal Structures

A previous post presented the idea that much-maligned problems with today's legal system result because our systems were not designed to handle rapid complexity growth in our economy and society. The sheer volume of new regulation results from well-meaning bureaucrats struggling to keep up with economic and social developments. The lobbyist community has appeared to opportunistically "help" these bureaucrats understand new complexity in ways which align with sponsored interests. On the tort side, opportunistic lawyers comb the complexity for lawsuit opportunities, then search for clients who fit lucrative lawsuit niches. From the outside, laypeople see an ever-growing body of law opaque to non-experts.

How can this situation be improved?

Remember the core philosophy of common law: the primary objective of law is to be predictable. In the old days, pre-complexity-explosion, predictability could be achieved by precedent alone. As long as laws followed established precedent, the law would hold few surprises. But in the era of complexity, precedent is no longer sufficient for predictability. Humans have limited memory, limited time to fill that memory, and limited processing capacity. Once the volumes of precedent become large enough, no human can possibly hope to understand it... and we passed that point long ago. There is little point in law being predictable if it is not predictable to humans, and much of the benefit is lost once the law becomes opaque to laypeople.

On the other hand, I believe the core philosophy of common law is solid. The primary objective of law is to be predictable. So let's keep the jurisprudence, and consider how to adapt the system to an environment of rapid innovation and complexity growth.

Our goal is to find a system which produces law which is not only predictable, but predictable to laypeople. First and foremost, then, it must be simple. More precisely, it needs to be simple by human standards. Any number of nerds have suggested that we wouldn't need common law if legislatures passed laws perfectly specified in computer code, e.g. Java or C++. Hopefully half a century of AI research is sufficient to show that this is a spectacularly bad idea, but even setting that aside, it would certainly be opaque to laypeople. On the other hand, there will always be lots of unusual cases, especially in an environment of rapid complexity growth. We do need some mechanism for applying the rules to individual cases.

When a case is in a gray area, what do we want to happen? Same as in any other case. We want the outcome to be predictable. As much as possible, it should be exactly what a layperson with a grasp of the basic principles would expect. Think about that: the outcome should be whatever people expect the outcome to be.

Time for a digression. One of my favorite game theorists is Thomas Schelling. His book "The Strategy of Conflict" has been described as a guide on fighting dirty in game theory. One of Schelling's biggest ideas is a simple experiment: put two people in New York city and give them a large reward if they can meet. The two are strangers, with no way to communicate. Where do they go, and when? Schelling suggested noon at Grand Central Station, although experiments have shown that noon at either the empire state building or the statue of liberty are more popular choices. The important point is that people are able to successfully coordinate in this situation, without any communication at all. This type of problem was dubbed a "coordination problem", and popular solutions (e.g. statue of liberty at noon) are called Schelling points.

Going back to law, we see that the philosophy of common law casts law as a coordination problem. We want the law to be a Schelling point: the outcome of a lawsuit should be exactly what everyone expects. Today's common law uses precedence to achieve this. If the law follows previous precedent, then as long as everyone knows the full history, outcomes are predictable. The analogy in the New York experiment would be to give both participants a long list of all the places and times people had tried to meet. The problem is that the body of precedent has become far too large for even experts to know the full history. What we need is some new mechanism for coordination.

Let's stick with the New York city analogy for a minute. Previously, people met in New York by consulting huge volumes of records showing where previous people had met. Alas, these volumes have grown too large, and the records are too complex for an electronic search engine to help (meeting places depend on an endless multitude of special conditions, vary by hour and weather and number of window washers on the empire state building, etc). We want a new system for our hapless strangers to meet. What to do? One natural starting point is to build a very big, very obvious, very well-advertised monument in the middle of the city which says "MEET HERE" on all sides in giant letters visible from New Jersey.

That's a start, but there are problems. See, much of the complexity of the old system existed for a reason. Our giant monument is outdoors, which is great when it's sunny but not so good in rain or snow. Plus, the monument needs frequent cleaning, and no wants to be around for that. Not to mention the birds which nest there in spring... we need a more flexible system.

So we build several monuments. Some have outdoor seating, some indoor. There is a regular cleaning schedule. There is a monument in Brooklyn and another uptown, for more local meetings. But there aren't too many. Local residents can list all the relevant monuments off the top of their head, and advertising monument locations, features and cleaning schedules is one of the main jobs of the Mayor's office. Maps and schedules are readily available at regularly placed kiosks throughout the city. Precedence volumes are still available when necessary, but most people can figure out everything they need to know from the FAQ section of the pamphlets.

Now let's translate this analogy back to law. Our giant monument would correspond to some very simple but undesirable solution, like assigning everyone in the country a social rank and declaring the person of higher rank to be the winner in all disagreements. Great for predictability, but still a terrible idea. Predictability is a priority, but the laws still need to be reasonably good. So we build more monuments.

The monuments in our analogy correspond to the core principles of the legal system. The success of the whole system is measured by how well laypeople understand the core principles - the monuments - and how well laypeople can predict how the principles will apply to any particular case. Clearly, public relations and advertising is a key component of this system. The courts and regulators must constantly communicate with public. They need to set up regularly-spaced kiosks with maps and schedules and FAQs on the core legal principles. Their success will be measured by how well the public can predict case outcomes. That means regular studies run on laypeople asking them to predict how the law applies to various cases. Since the success of the bureaucracy will be measured by public understanding, the bureaucracy will be motivated to keep their core principles simple. Old monuments will be removed, and the total number of monuments will be limited.

There are still open issues. For example, how can we incentivize the laypeople in our studies to honestly report what they expect to happen rather than what they think should happen? On the public servant side, how are bureaucrats and judges incentivized to create laws which are both good and predictable, rather than just giving everyone a social rank? These are nontrivial issues, but they seem tractable. Let's leave them for later.

In summary, we want a legal system with a small, simple, actively maintained set of core principles. Public servants both apply the law to specific cases as judges and regulators, and actively spread information on the principles and their application to the population. Bureaucrats' performance is measured by studies on laypeople, where the best outcome is that the laypeople can perfectly predict how the law will apply to each case.

A Problem with Today's Legal Structures

Many people are surprised to learn that the primary job of a judge in the US is not to deliver fair, just judgments, but to deliver judgments consistent with precedent. This is the principle of common law: every court decision is itself law. Judges have the power to create new law in situations with limited precedent, but are bound by precedent when available.

The idea behind common law is that goodness is not the primary objective of law. Rather, the primary objective of law is to be predictable. If a law is sufficiently problematic, then the legislature has the power to change it. But if laws change frequently or are applied unpredictably, then people will be unable to plan around them. Society's day-to-day functionality depends on knowing that the law isn't going to apply in new and potentially inconvenient ways every week.

In Kosher Hot Dogs, I mentioned a major problem with common law in practice: a lack of central principles. Common law as practiced in the US mostly guarantees that people can continue doing what they're doing without unforeseen legal interference. But today's legal body, whether tort or regulatory, has become so unwieldy that non-experts cannot reliably predict how the law will apply to any new plans they make.

Despite the inevitable libertarian rhetoric, this is probably a new problem. The rapid growth of complexity in society has been one of the largest underlying social changes of the twentieth century. In 1900, a single person could understand every industry and their interactions, every significant political or religious group and their interactions, etc. The world was a simpler place. The end of communism was, as much as anything, a clear signal that central planning could no longer keep pace with economic complexity. It was precisely this complexity explosion which powered the early growth of information technology. IBM's first computer was built for the 1890 US census, when the population became so large that human calculators could not hope to complete the census totals before the 1900 census. The explosion of business complexity drove mainframe purchases up to the era of personal computers. All along, accelerating innovation has driven accelerating complexity growth, creating a feedback loop as more innovation is needed to handle the new complexity. With these patterns in mind it should hardly be surprising that our foundational legal principles are failing under this new complexity load. No one has ever designed a legal system for an era of runaway complexity.

What does this failure look like? Opacity. In regulation, it looks like a large bureaucracy creating tremendous amounts of new law. The Federal Register has a regular email containing a "table of contents" for the daily activities of the bureaucratic agencies. Today's table of contents runs for 11 pages, with each line linking to much longer documents from the individual agencies. That's one day's worth of new regulatory law. All this buys the US a well-deserved reputation for complex regulation, onerous for small businesses. Similar incentives drive similar results in tort law, where the US public is routinely surprised by outrageous lawsuit settlements. Companies are forced to protect themselves with farcical "terms of service" and "end user license agreements".

In the libertarian view of the story, all this results from the "ratchet effect". Every time some new sob story hits the press, courts and regulators create new law to prevent it happening again. Often whole new departments are created to handle the problem, accelerating the effect. Many of the laws and positions are poorly thought out, and none are ever removed. The result is a steadily increasing mountain of onerous law.

I don't think this story captures the full picture. Sure, we can point to specific cases which fit the ratchet narrative, but let's step back. England has operated under common law for literally a thousand years. The Muslim community has operated under common law even longer, and the Jews longer still. Much of the Torah and all of the Talmud is Jewish common law, consisting of the (often enjoyably snarky) rulings and commentary of rabbis. Yet for all those centuries, nobody seemed to think that runaway complexity was an issue. The Talmud certainly hasn't shrunk, but the growth of rabbinical law just wasn't a major problem. On the regulatory side, Napoleon famously created a meritocratic civil code similar to the bureaucracy of most first-world nations today. Less meritocratic bureaucracies have existed much longer throughout the world. Complexity didn't seem to be a major issue with these institutions until recently. Today's massive lobbyist infrastructure, for example, is an artifact of the last 50 years.

In our narrative here, the ratchet effect is real but not the main problem. The main problem is that our legal systems weren't designed to handle the runaway complexity of the modern world. As complexity grew beyond the grasp of civil servants, opportunists quickly appeared to take advantage. In tort law, these opportunists take the form of ambulance chasers and class action lawyers, combing the legal mess for lucrative lawsuits. In regulatory law, the lobbyist community "helps" civil servants "understand" the complexity in ways which align conveniently with sponsored interests. Meanwhile, well-meaning judges and bureaucrats try to keep up with demand, cranking out new law to handle the proliferation of new situations.

Unlike the ratchet narrative, our story attaches the ills of today's systems to complexity growth, especially over the past 50 years. Thus we would predict, for instance, that the lobbyist community emerged within that time range in response to new opportunities, and this is indeed the case.

Next post will discuss how to solve this problem.

Kosher Hot Dogs

If ever there were an industry that no one in their right mind would trust, it's hot dogs. First rule of hot dogs: you do not want to know what's in it. But in this unsavory industry, Hebrew National stands out. The slogan on the company's site summarizes their advantage well:
"When your hot dog's kosher, that's a hot dog you can trust."
And indeed, people do trust Hebrew National's hot dogs.

This is more remarkable than it might seem at first glance. Consider the problem from the perspective of a hot dog company. This company wants to carve out a niche: they will produce high-quality hot dogs, and sell them at a correspondingly higher price. Many people will happily pay extra to know that their hot dogs do not contain ground-up lucky charms, bits of fur, or the occasional lost dog. But how can the company communicate their quality to prospective consumers? How can they convince the public of the superior quality of their hot dogs? What claim could they make which an unscrupulous competitor could not copy?

This is the problem known in game theory as signalling: one party wants to communicate their superior quality to another party, but in order to do so they must send a signal which their unscrupulous competitors cannot easily copy. A certification body can solve this problem. Consumer Reports, for example, provides unbiased analysis of a wide range of products. Unfortunately, this solution is subject to attack in the real world by exploiting the limited information capacity of consumers. Any company can (and does) invent arbitrary metrics by which their product performs best. A less cynical interpretation is that each company stakes out a niche, claiming that their product is the best for X. If you want X, you buy that company's product. Just within hot dogs, we have Ballpark's "Angus", Applegate's "Organic", Nathan's "Bigger than the Bun", Oscar Meyer's "Selects", and several brands of "Premium Jumbo". Many of these brands have multiple lines of hot dog servicing different niches. Consumers with limited attention to devote will ignore most of these, and most are useless anyway. Thus the true signals of quality are drowned out by the noise of their competitors.

If a company is to charge extra for a truly superior product, then they need a more dramatic way to signal quality. Hebrew National does this by invoking kosher rules. Implicitly, the entire weight of the Jewish religion backs the quality of their product. The kosher rules force the company to produce high-quality hot dogs, and lets them communicate their high quality even in the noisy environment of the modern supermarket.

But what does that even mean? I don't actually know the kosher rules. I remember a few bits and pieces... no hooved animals, separate meat and dairy, something about which cuts of meat are acceptable... but I don't know most of it. Yet I'm willing to accept kosher standards as an assertion of quality, at least in the unpalatable hot dog industry.

I do know that kosher rules are generally intended to ensure food quality. They are a 3000 year old FDA regulatory equivalent. They are interpreted by an active rabbinical community, which makes sure that the word and spirit of the rules are properly applied to new foods and new food processing technology. The result is a regulatory framework which is roughly understood and highly trusted by laypeople, even though most of us do not have a detailed knowledge of the rules!  Just as important, I know that the whole community of people who observe, certify and maintain kosher rules is highly trustworthy. They consider themselves in service to God. Abusing the kosher rules or certification would be not just unethical, but a direct transgression against God. I can trust that the rules are applied consistently with the principles.

We have two key elements here. First, a regulatory framework based on simple principles (food that wouldn't be bad 3000 years ago, plus some purely ritual aspects). Second, a highly trustworthy community to work out the details of the rules in keeping with the principles. The result is a certification which laypeople interpret as high quality, even in a market cluttered with questionable claims of quality.

The utility of kosher rules as a regulatory body suggests that these properties could be useful for regulation more generally. As Hebrew National demonstrates, the regulation need not be state-mandated, though the kosher rules may offer some insight there too.

One obvious analog is Islamic banking. The most notable rules of Islamic banks are that they cannot charge interest and they cannot gamble. In practice, contracts are structured to achieve a similar effect to interest, but the consumer sees a number of benefits. Foreclosure is rare, and risk in general is kept low. Many risky investments are considered gambling, and are forbidden. An Islamic mutual fund could see enormous success in the Western world, offering a bank-like investment with positive return and a religious guarantee of low risk. Such a fund would certainly not have invested at all in subprime loans.

What about beyond religion? On paper, much US law follows similar principles. Legislative bodies lay out the original laws, judges make sure the details are in keeping with the original word and spirit. In practice, most US law follows other patterns. Tort law is mostly common law, with judges building on the work of previous judges without any original legislation and few central principles. Regulatory law is usually handled by bureaucrats rather than judges, again with few if any central principles and with less respect for precedence. Criminal law does follow the pattern reasonably well, with simple core ideas (don't kill, don't steal, don't assault, etc...) and details handled by judges.

The main shortcoming of US law, as compared to kosher rules, is the lack of organizing principles understandable by laypeople. In tort law, this means that people cannot reliably anticipate what might be subject to lawsuit without expertise in the subject, and so businesses are forced to take costly legal defensive measures. In regulatory law, people cannot reliably guess what regulations apply to their ideas without considerable research, creating major barriers to new businesses and innovations.

How can this situation be improved? Stay tuned...