> Most legal rules could be vastly improved by reducing them to principles that are combined in a framework of cost/benefit analysis.
Lawyers already do this. The world is still too complex to reduce to a single, central set of uniform principles. Instead you have lots of little pockets of temporary uniformity.
What happens is:
1. Lots of disconnected cases.
2. One day, a lawyer or judge notices a pattern.
3. A new principle is expounded in a case.
4. Other judges might pick it up and it becomes the common law. Or they don't, and it doesn't.
5. If it's widely accepted, the new principle begins to accrete exceptions, odd outliers, dusty corners and so on.
6. One day, a lawyer or judge notices a pattern ...
You need to stop thinking of law as a product and think of it as a process. It is not a static, manufactured good. It is a procedure for discovery.
>> Most legal rules could be vastly improved by reducing them to principles that are combined in a framework of cost/benefit analysis.
>Lawyers already do this.
I'm not really sure what you mean by this, but how lawyers think about things is not really relevant to legal outcomes. Also, none of the lawyers I know employ anything like the reasoning I have in mind in their work. Finally, if your claim is that the outcomes produced by our byzantine legal system is equivalent to the outcomes that would be produced by a principled approach, then I think you are mistaken.
> You need to stop thinking of law as a product and think of it as a process.
You need to stop assuming that you know what I think :). I don't think it should be static. The law acts to model aspects of the world that society feels should relevant to legal outcomes. The world is far too complicated for that model to be perfect, in my preferred system or yours. So, of course, we will occasionally notice aspects of our world that are not modeled and deserve to be, and there should be a procedure for doing as much.
There are many problem with the process we have in place. First, there aren't clean principles to begin with. The courts are tasked with ruling on the written law, which is highly imperfect. To the extent that the courts "discover" new principles out of the written law, they are effectively writing new law, which in theory isn't their job.
Second, the principles should come before the outcomes. In your flow chart, the outcomes are decided first in a large body of cases and then eventually the pattern is recognized. But the whole point of a principled approach is to improve the quality of outcomes, not just to simplify the description of the poor outcomes that would have been arrived at by judges thinking hard about it.
In general, I think you greatly overestimate the number of principles that would be necessary, in the right framework, to improve the legal system. But I recognize that the onus is on me to back that up and I'm not really in a position to propose a specific alternative system.
I think you are both right to a large extent, and the distinction is whether you are focused on the easy cases, the hard cases, or some point in between.
The repetitiveness in the common law is a massive feature and precedent does make the majority of cases quite predictable. And the outcomes of most cases are very principled - just not always with the hardest cases.
The limitation, though, is in knowing whether the facts push a case in one direction or another. Quality lawyering makes a huge difference there, and it is the lawyer's skill that enables her to figure out how to interpret and shape the facts.
As both of you have essentially noted, there are a potentially infinite number of fact patterns, but a finite set of legal rules, so any system must apply an existing law to a new set of facts from time to time. Whether that is viewed as applying an existing law or creating a new law is in the eye of the beholder, and legislators and judges disagree about that all the time. And whether that is a bug or a feature, is a deeper philosophical question that doesn't get at what an expert system can/will do.
I think the only way you'll get a truly effective expert system for the law is if the software is integrated with the people making the law - the legislatures and the judges. I'm doubtful that will happen. In part, that is because I am doubtful that computer systems will ever be able to decide cases in a way that seems independent, fair and humane to the ones being judged.
Short of that, my own startup (judicata.com) is working on mapping the legal genome so that we can identify the core legal rules, and how they are applied to different fact patterns. We think we can make big strides with legal technology in the coming years. It's not an Apollo level undertaking, but it's a pretty massive undertaking nonetheless. We're hoping that a systematic approach at understanding the law at a deep level of detail will allow us to make some of what has been discussed here a reality.
We have not published much about what we are doing, so I don't have an available link. I would be happy to discuss it on a one-to-one basis over email, though: itai at our domain.
> So, of course, we will occasionally notice aspects of our world that are not modeled and deserve to be, and there should be a procedure for doing as much.
That is exactly what the courts do.
> First, there aren't clean principles to begin with.
Yes, there are. The courts are forever elucidating them, combining them, perfecting them. The legislature periodically replaces them.
> Finally, if your claim is that the outcomes produced by our byzantine legal system is equivalent to the outcomes that would be produced by a principled approach, then I think you are mistaken.
> In general, I think you greatly overestimate the number of principles that would be necessary, in the right framework, to improve the legal system. But I recognize that the onus is on me to back that up and I'm not really in a position to propose a specific alternative system.
Well good news. This has been tried, it's called a "code" or "civil" legal system; as opposed to the byzantine "case" or "common law" system that is used in countries descended from Britain.
And it doesn't make the actual complexity of the real world go away. I'm struggling to convey this point. The law is not complex because of the way the law works. The law is complex because the world is complex.
That is an irreducible complexity that simply cannot be swept away by a single rational system. Even the civil law acquires warts, dusty corners etc and is constantly being altered, tweaked, fiddled with.
You don't realise it, but you have stumbled upon a very old strain of wishful thinking: society seems so irrational, surely we could do a better job if we started from first principles and nutted things out from there.
It has been tried. Many times. By many of the greatest minds that have ever lived.
>> So, of course, we will occasionally notice aspects of our world that are not modeled and deserve to be, and there should be a procedure for doing as much.
> That is exactly what the courts do.
Agreed, common law has this capability. I was challenging what seemed to be your assumption that a principled system wouldn't.
>> First, there aren't clean principles to begin with.
> Yes, there are.
You think the US Code of Laws is clean? The one that, according to Wikipedia, is around 200,000 pages long?
> I'm struggling to convey this point. The law is not complex because of the way the law works.
You've conveyed that opinion quite plainly.
> [Rational rules developed from first principles] has been tried. Many times. By many of the greatest minds that have ever lived.
What super rational system developed from first principles by the greatest minds of all time did you have in mind here? I'm well aware that there are legal systems that aren't based on common law.
Let me ask you this: do you think the staggering complexity of a modern OS is essential complexity? Because I don't. And yet there have been many attempts to reinvent from first principles that have failed to gain traction. Many of them, in a quest for simplicity, were overly simplistic in some regard. This isn't myth busters, though. A couple of examples don't translate to a proof.
Your position seems to be that we've stumbled upon a system of law here that introduces very little accidental complexity and that handles well the essential complexity of the real world. I don't really know how you can have that position if you've had exposure to legal reasoning. Have you studied constitutional law?
> What super rational system developed from first principles by the greatest minds of all time did you have in mind here?
I was referring more to the presumption that all complex systems, everywhere, can be replaced with a positively designed rational system. This is pretty hard to refute amongst the HN crowd because our entire bread and butter is creating positively designed rational systems. It's what we do.
As I said further down, in the limited case of laws, the civil laws have tried to replace a historical patchwork of custom, local rules and decrees by a single, positively designed, single rational system of law.
But this still frequently turns out to be inadequate. Injustices and oddities pile up around a glitch, which is modified in isolation; then more small changes occur, until a new pattern is recognised by the legislature and the code is rewritten.
Sound familiar? It should: in practice the civil law winds up being adaptive, just like the common law. At least the common law doesn't delude itself into thinking the whole thing could be nutted out from the very beginning.
Software engineers should know better. Just look at the archives of the IETF sometime. Here intelligent individuals discuss, design and promulgate the protocols that make the modern world pulse with activity -- the technology that makes it possible for us to retread an argument that is at least several hundred years old.
Yet RFCs are periodically deprecated and replaced with updates. Why? Because of unforeseen circumstances, errors, abuses and so forth. No matter how intelligent the designers, they were simply unable to foresee all the possible cases in advance. And the IETF, like the Common law, places intelligent, localised-to-problem adaptation of existing principles at the centre of its mechanism.
> Your position seems to be that we've stumbled upon a system of law here that introduces very little accidental complexity and that handles well the essential complexity of the real world.
I'm sorry if I gave that impression. The legal system can and does sometimes impose enormous drag. The the extent that laws become onerous and the legal system highly unpredictable, it's necessary to petition to legislature to reform them. Patent law is certainly one such topic.
But the complexity of legal reasoning is not accidental complexity, in the sense Brooks was talking about. It is essential complexity. It exists in the problem domain, which is all relations between human beings.
I agree that any good legal system should have the adaptive quality of common law. I'd agree that the major source of accidental complexity in the law is poorly written laws, but I also think that there is significant room for improving the framework in which laws are written, such that many rules (e.g. the rule against Hearsay) that have lots of explicit exceptions would be better served a simpler framework with a few principles from which the exceptions emerge.
Have you ever written enterprise software? Multiply that by everything.
> It has been tried. Many times. And they have all failed.
See in the most general cases the Cartesian rationalism of the French Revolution and its latter offspring the Russian Revolution. In the less ambitious case, the civil laws of Europe.
You seem to be confusing phenotypic complexity with actual complexity. Software engineering is the quintessential form of complexity reduction, code is never infinite in length. In machine learning you don't even need to come up with the rules, let the machine learn by itself.
Descartes loved reducing the "complex" world to math; Written law has been with humanity since times immemorial in fact being of the fundamentals of society, how is that failure?
You seem to imply that an automated judicial system would be immutable; i don't think anyone argues that laws should never change.
Lawyers already do this. The world is still too complex to reduce to a single, central set of uniform principles. Instead you have lots of little pockets of temporary uniformity.
What happens is:
1. Lots of disconnected cases.
2. One day, a lawyer or judge notices a pattern.
3. A new principle is expounded in a case.
4. Other judges might pick it up and it becomes the common law. Or they don't, and it doesn't.
5. If it's widely accepted, the new principle begins to accrete exceptions, odd outliers, dusty corners and so on.
6. One day, a lawyer or judge notices a pattern ...
You need to stop thinking of law as a product and think of it as a process. It is not a static, manufactured good. It is a procedure for discovery.