18 ms·
What a pain. Speaking as a former attorney, these situations piss me off. English prose really isn't equipped to handle long conditions, exceptions and such.
by newy 17y ago
What a pain. Speaking as a former attorney, these situations piss me off. English prose really isn't equipped to handle long conditions, exceptions and such. What's important in a contract is that it unambiguously captures the intent of both parties. Formulas, diagrams really should be used more often. We're not on typewriters anymore - Word can handle graphics. If you've ever seen a complex mathematical formula written out in legal prose, you'll know what I'm talking about. End rant.
- ars 17y agoSounds like it's time to write contacts in C :) I wonder how that would work in court?
- kevingadd 17y agoSomehow, when someone says 'unambiguous', C is the last language I can think of. I would sooner think of something more mathematically 'pure', or just reader-friendly, since the goal is expressing meaning, not actually running the contract... (I hope) I suspect getting it to fly would be the biggest challenge, though, since our legal system is very much bound up in tradition and precedents.
- ars 17y agoI actually was thinking that the goal is running the contact. Not the whole thing maybe, but any complex parts.
- mortenjorck 17y agoWhat's really needed is a language specifically written for the job. C was built for writing system software. English was built for writing poetry. Neither are ideal for precisely specifying the terms of a contract. Maybe if Lessig got together with a few linguists and information architects...
- JabavuAdams 17y agoI remember reading some sci-fi involving a treaty so complicated it had to be machine-verified. Too bad about that Gödel guy, though.
- cabalamat 17y agoSomething a bit like Prolog might do the job. There would also have to be terminal predicates coded in English. So theft (in English law defined as: "dishonestly appropriat[ing] property belonging to another with the intention of permanently depriving the other of it") might be: (define ([Person]X is-thief of: [Property]P belonging-to: [Person]Y) (and (X appropriates P) (X != Y) (X intends ((Time>now) implies not (Y has P))) (dishonestly))) (define dishonestly "this is purposely left undefined in the law") (define ([Person]X appropriates [Property]P) "X does some action causing him to have control over P") There could then be a compiler and other software to automatically reason about such statements.
- karzeem 17y agoGood points. As a layman, I wonder, do courts demand that all of a contract's substance be in prose, or is it okay for two parties to agree to a contract that's partly in, say, diagram form?
- grellas 17y agoThe terms of a contract can be evidenced by anything that shows a meeting of the minds, including the actions of the parties in implementing the contractual terms (so-called "course of performance"). There is no doubt that diagrams can be used in contracts and frequently are. Though a diagram would not stand alone, it would constitute strong or even conclusive evidence of the meaning of the prose part of a written contract (or the meaning of words exchanged by parties in the case of a verbal contract). Bottom line: a meeting of the minds always has reference to some context and all of that context can be relevant to determining meaning - if a diagram helps show the context in which parties are agreeing to something, it can and should be used to add clarity to a contract.
- saturnine 17y agoWhen dealing with contracts I've often reflected on the parallels between lawyers and programmers. Both attempt to craft their writing such that it's interpreted the intended way. Both try to anticipate changing environmental conditions and account for them. Both find their argot inscrutable to outsiders. And both can be torpedoed by a missing comma.
- cabalamat 17y agoMaybe there could be something akin to a programming language for contracts, and also for statute law.
- Daniel_Newby 17y agoWhy do lawyers so often insist on writing giant running sentences with unclear referents? Engineering practice is to chop sentences up into clear, concise chunks. A sentence like that one in a technical spec would have likely have failed review on the basis of complexity without bothering to understand it.