Codex: A Legal Scripting Language for Ethereum
medium.com
medium.com
Based on the low two-digit number of hours I've spent reviewing documents and on the phone with my attorney so far, there are a lot of issues with the core idea of this (but IANAL).
> Where there are disputes, a blockchain can hold an objective record of events making dispute resolution relatively trivial
Good luck. That's sort of the point of contracts in the first place. To easily be able to say you agreed to something 8 years from now when a partnership/agreement has gone totally pear-shaped. And we see how well that's worked out.
> On didactic contracts: mapping out in detail every aspect of human behaviour
Good luck ad infinitum. That's just never going to happen. And this is why the first quoted passage is equally unrealistic. There will always be a nebulous area of a contract, the question is just how big that area is, how nebulous it is, and whether you're willing to accept that v. spend the additional resources to clarify it. Which with any negotiation increases the odds the other party will walk away.
> Since traditional didactic contracts are mostly unread, they do not in most cases involve active consent, which has moral hazard.
Every single word on every page of anything that needs to be signed has been read at least a dozen times by all attorneys and non-attorneys in this transaction. I'm not saying everyone reads every didactic contract but even busy people do not just take their attorney's word for it.
Obviously an acquisition is an incredibly complicated transaction, requiring an enormous amount of effort. The stakes are incredibly high.
There's plenty of low-stakes commercial endeavors that aren't so complicated, but could still use computerized dispute resolution.
Take a ride share. The people involved are looking to mitigate various kinds of risk. The driver wants to make sure you won't puke in his car, that he'll get paid, etc. etc. The rider wants to make sure that the driver has insurance, that he'll show up on time, and other kinds of guarantees.
So you can set a flexible system whereby should either party break the terms of the contract, the software provides a path to redress. Each of these guarantees can be modeled in the software, with systems for, say, uploading photos with timestamps if the customer actually does puke in the car.
The record, plus the collected history of adjudication results, provides a way for both parties to come to a conclusion about this particular dispute. Then both parties can decide whether to accept it and move on, or escalate.
If one chooses to escalate, then the record of the transaction can be pulled by an appointed arbitrator and then he can add the needed human touch to create a resolution. Again, if one party is not satisfied, then he can escalate to an actual lawsuit.
Over years of refinement, escalation, arbitration, the system can be tuned to such a degree that lawsuits in that arena become practically non-existent. Profits generated by the system can be re-allocated towards new market-making activities.
Imagine that if at the beginning of the ride, you had to lock off $100 in crypto, in case the driver claimed a dispute.
If you don't lock the money, the whole benefit from using cryptocurrency disappears really. You might just as well have it on the paper with outlined execution steps.
Sure, but most places that want this can just impose internal dispute resolution plus binding arbitration via contract, which achieves the same end without technology.
And imposing this cryptocurrency-based automated dispute resolution plus arbitration scheme requires all the same social and legal pieces, plus all parties having access to and trust in a common cryptocurrency-based automated dispute resolution system as the first step.
It just adds an extra required focus of shared trust compared to the status quo.
The line about didactic contracts being unread was related to the example of "terms and conditions". As you admitted yourself, even for high stakes contracts it is your attorneys who read and understand the contracts, not the people who must consent to be bound by the terms. It would be better to simplify legal matters completely by focussing on executables.
Even the clearest contract is worthless if one party is too weak to hold everyone to the contract's terms. "What, you want me to pay for the work you did for me? Come back with a lawyer and we'll talk."
That said, Ethereum looks really cool, and I'd be interested to see what kind of good they can do.
Ethereum solves that: as part of the contract you put the money into escrow, such that if the work is completed the money is automatically delivered to the other guy.
Personally I think the problem is on the other side: an automated contract is all well and good, but how can a computer ever make a judgement as to whether the work that was done was compliant with the terms?
It might be useful in situations where there is one, and only one, clear, explicit, strict and agreed interpretation of a contract.
But that doesn't cover most of contract law, where the point of the wording is often to disguise and/or hide by misdirection or omission legal implications that can be exploited for profit, or to bully the other party into believing they have to accept terms without challenging them.
That may be a profitable misuse of the law, but it certainly isn't the reason the profession of law exists. The point of contract law is to allow parties to cooperate in the pursuit of profitable courses of action. Power differentials exist, but laws get passed to prevent egregious misuses of those differentials. That's pretty much what contract law is, is refinements intended to protect smaller stakeholders.
> to bully the other party into believing they have to accept terms without challenging them.
Let me give you an example. I am currently in the process of switching companies. My boss suggested an arrangement, and when he presented it to the company, they asked me to sign a separation agreement. The terms of this agreement, in the document, are to remain confidential for a period of 2 years.
Now, there are a few aspects here that I have to consider. I do not want a legal requirement hanging over my head for years. Second, I do want the terms of the agreement and do not want to throw away the deal over it. Third, nobody wants to sue anybody, nobody is expecting the language of the agreement to ever see the light of day in court. The company has a specific goal in mind with the particular language they used on that contract.
This all gives me negotiating headroom. By understanding what the company is hoping to accomplish, (me not leading an employee charge for better treatment) I can push for better terms without blowing the whole deal. I was able to extract a concession down to one year.
But the reality is, I did all that, and didn't gain anything more than just practice honing my negotiation skills. I'm talking about the agreement now. Legally, I might be in the wrong, but as the company doesn't care unless they're faced with the precise situation that the language was inserted into the agreement to prevent, since I have no intention of doing that, I can pretty much do what I want.
You can look at their pushing the agreement on me as the company trying to bully me into accepting terms. There are situations where that might be the case. I refuse to do business with anyone where I feel the relationship might turn adversarial. If a relationship were to turn adversarial, with a more powerful entity trying to extract things from me, I'd exercise mobility and ingenuity to get out of it and then try to learn from the experience.
But generally, there are plenty of opportunities, especially if you live in the US, to where you don't have to swim with sharks if you don't want to.
Contracts offer a framework whereby people can collaborate on endeavors that have the risk of turning into adversarial conflicts, the hope is that by adhering to the terms, you'll keep the bears at bay. Doesn't always work, but if your eyes aren't bigger than your stomach it usually works out.
Etherium offers "Contracts 2.0", a way to have micro-interactions that wouldn't ordinarily happen because they're too legally-risky, the costs of the risk are far greater than the profit potential. The hope is not to provide iron-clad assurance that nobody will ever get sued, but to give everyone involved just a little more peace of mind. That could well be enough to allow interesting new commercial markets to take form.
I would love to ask clients to "sign here" on a blockchain-enforced contract before they decide not to pay just because they feel like they can, despite perfect delivery to their own satisfaction.
¹ Digitizing as conversion to digital but locally stored documents like they did with books doesn't do it with contracts, as these are more than information on paper, so the digitizing in this case had to address somehow also the contracts' functional part.
I was thinking more along the lines of having contract breaches being publicly visible, which would motivate the more sociopathic egotists of the world to keep up appearances of fairness. Making it harder to just move to the next victim because a reputation could be built up.
Edit: also bridging the gap between expensive lawsuits that end up not being worth it, or the temptation to publicly shame the bullies which is almost always tacky and never dispassionate.
What concerns me is the evaluation of the "do" - what is responsible for determining if something is "done"? How does contract dispute work within Ethereum? I can see Gnosis is doing predictive markets work, and I get that in theory it's possible, but are there mechanisms in order to evaluate whether a contract term has been met autonomously with Codex or other tools in the Ethereum world? If so, what are they?
I'd kind of assumed smart contracts were for things like financial trading where there are somewhat legitimate electronic oracles and parties are assumed to be competent, functional adults with an understanding of the risks. I can't really see it working out in the world of normal people.
This will be wonderful if it takes off. Imagine a company that issues "hour shares" based on work contributed. Ok, so Assembly[1,2] didn't work out too great, but the idea has a future.
[1] http://assemblymade.com/ [2] http://web.archive.org/web/20150321002401/https://assembly.c...
I've been thinking of something similar, basically a formal language to hold authorities accountable and consistent in their actions. Sort of an instruction set for government.
But, realistically speaking, trying to put an API on top of human behaviour is not easily done and it will most probably not lead to more happy humans.
So, how do you implement Pax's resolution if one party refuses to accept it?
The interesting clauses in leases are the ones that require tenants, for example, not to trash the place, and landlords to fix things in a timely fashion, not enter without 24 hours notice, etc. These disputes seem to require a human arbitrator to determine, for example, who's fault the damage to the apartment is.
Edits in italic: The idea that this will catch on in any non-trivial fashion is such flagrant nonsense that I don't know where to start. Note that I didn't see the author of this page saying that it will, but a lot of comments seem to indicate that this is the hope.
Most of law school is dedicated to being hammered over the head, repeatedly, with the fact that contracts constantly use this word:
REASONABLE
And many, many thick books written by a lot of very smart people have been dedicated solely to trying to define this word, which is hotly debated.
Quite simply, contracts are semantic documents that are used as reference for judgment based processes. Syntactic interpretation simply is never going to work, and that is all blockchain is going to get you. The existence of closing conditions to big deals (e.g. company sales) require human judgment - they are often things like "the reasonable likelihood that there is no need for further tax indemnification" or "there is reasonable likelihood that there will be no further suits filed against [x]". How in the hell is a computer ever going to make determinations about the reasonable likelihood of human actors that are sufficient to have as bases for contract conditions? And even if they can, is that what you want? I still want a human's finger on the trigger. And either way, until you have true AI interpreting contracts for you, this is a completely hopeless task. It so fundamentally misunderstands what attorneys do that it is just worthy of being dismissed out of hand.
Let's put it this way: saying that contracts can be 'coded' in this way is the exact same as thinking that all software developers do is write code.
I spend at least as much time writing contracts as I do trying to figure out what the heck it is that is actually going on, investigating, documenting requirements, satisfying stake-holders - it is not, unlike, in fact, writing software.
So, in essence, saying that contracts can be automated is literally the exact same statement that software development can be automated.
Show me how. I will make you wealthy beyond the wildest dreams of avarice.
NB: you may not read the TOS - but trust me, as someone who has written many, many website TOS's, people do read them. Important people, like states Attorneys General, representatives from acquiring companies and competitors, attorneys representing people who are trying to sue you and members of the general public who actually give a darn about their privacy. And, quite frankly, TOSs are a tiny fraction of what I do - and I have a very brisk software practice.
The vast majority of actual commercial contract negotiation is in documents that are heavily worked upon by both sides and reviewed many times by multiple people. If you think you can get even 20% of these people to become effective at writing markdown much less learning logical operators, well, I have a bridge to sell you.
E.g. An assurance contract a la kickstarter (for the most trivial case), options contracts, or multi-party escrow.
And that perhaps the scope of its capabilities would expand incrementally over time as people got better at thinking about contracts in this way.
An analogy might be to Squarespace/Wordpress for web developers. Sure, you can't make every website on there. But for a lot of simple cases you can, and those cases are expanding.
Putatively, I should be a prime candidate. I spend my nights and weekends (to the extent I have them) working in node and jquery, I cut my teeth on ANSI C two decades ago, I studied math and physics as an undergrad and now I'm a full time lawyer.
I am deeply, deeply hesitant to use this kind of agreement. It is not that the systems we have are ideal - they are definitely flawed - so my attitude is not that "if it ain't broke, don't fix it." The attitude is that of "this is likely to introduce way more risk and uncertainty into processes that are already risky and uncertain."
Let's put it yet one other way: if the point of this language is to make things clearer - it fails at the starting line, for the simple reason that 90%+ - probably 99% + - of people that have to deal with contracts cannot make heads or tails of a logical operator, and teaching them to do so with enough comfort that they can parse and evaluate such statements at anything near the speed necessary for commercial applicability is a sisyphean undertaking.
The advantages of Ethereum are in eliminating 3rd party trust and the costs that come with it. Right now if you want one of these simple financial services, there is both a direct financial cost associated with it and the implicit risk-expense of trusting someone else with your money. This is a drain on the global financial system that just doesn't need to exist anymore.
The idea that Ethereum or things like it will supplant the legal system is of course, absurd on its face. That will not happen. But what we may start to see, in my opinion, is the co-mingling of the two, and contracts being written with deterministic, executable components alongside more traditional legal verbiage.
Fundamentally there are some things that are amenable to mechanization and some things that are not. But right now both are encoded in human language. And the use of either one for the other's optimal purpose is perverse and unnecessarily expensive.
No wonder people hate the legal system. It's a system where humans can't come to consensus easily...
but you are right the future will be the status quo of contracts written in ambiguous english language where we pay someone 300/hr to argue over what the definition of the word "is" is.
$300 is the low end of the spectrum, dotcomrade.
And trust me, I assure you that there are very many exceedingly important circumstances where reasonable minds could disagree vehemently over the definition of "is" in a given context.
[Here is a recent example of an and/or typo potentially costing $3B.](http://nypost.com/2015/12/14/caesars-may-sink-because-of-all...) I don't recall how this all came out in the wash, but put it this way: precision matters.
Or, to put it another way: why do people who spend all day writing software code behave as if contracts can be written with any less precision? Dismissing arguments over the definition of "is" is the same as steamrolling over variable types in a strictly typed language. It is just indicative you don't have much respect for working in that language.
I'm afraid this is not as far-fetched as you may think.
Determining what is a "reasonable likelihood" of any event is basically the essence of Bayeasian analysis, a well-understood branch of mathematics that deals with states of belief. In fact, determining what is a reasonable state of belief, given some information about a certain situation is the whole point of Bayesian probabilities.
To give you an example, in the case of "there is reasonable likelihood that there will be no further suits filed against [x]" someone may well gather some evidence to build a model of the kind of people who are likely to file suits against entities similar to [x], then use that model to argue that person A is very likely to do so.
Personally speaking, I think this is a crap way to reason about human behaviour but then again, if you throw enough maths at an argument, eventually most people will back down, and let you have your way.
P.S. Sorry if this isn't clear above: the process of building a model and using it to estimate the likelihood of an event is fully automatable.
Yes and no. Software is written in programming languages that have a much stricter interpretation than legal language. In fact, programming languages' interpretation is _automated_ (as in compilers and interpreters). Also, programs execute in much better circumscribed environments than legal scripts.
>> Show me how. I will make you wealthy beyond the wildest dreams of avarice.
Are we talking about automating programming? Because there's quite a bit of literature on automatic programming, particularly from the eighties and nineties I believe- and you could argue that basically all of machine learning is just automating the hand-crafting of complex, rule-based software systems.
... which, as you're arguing yourself, is not that far from legal documents, eh?
I'd rather see a DSL embedded into an existing language with clear semantics.