Hi! Help proposed!

[Edouard E]

Hi Wong Meng Weng,

I'm a french legal council (Business LAw/Contract Management), bilingual FR/US. Have been working in the software world. Know a little about Eclipse environment, javascript, HTML5, CSS... Highly interested in 'computational legal' (created VBA 'Access' databases and apps to automatize french contracts 20 y ago) Would love to help but don't know what to do!

Don't hesitate to contact me!

Ed

[Meng Wong]

On Sun, Oct 16, 2016 at 8:35 AM, Edouard E <emilio....@gmail.com> wrote:

I'm a french legal council (Business LAw/Contract Management), bilingual FR/US. Have been working in the software world. Know a little about Eclipse environment, javascript, HTML5, CSS... Highly interested in 'computational legal' (created VBA 'Access' databases and apps to automatize french contracts 20 y ago) Would love to help but don't know what to do!

 

Hi! Thanks for writing to volunteer!

Quite a few lawyers do write in to ask how they can help. I'll offer an answer, but other people may have other ideas, too…

It's possible that in a year or so we will be ready for volunteers to translate common contracts and important legislation into L4, our legal language.

The scenario would look something like this:

Some end-user needs to get some paperwork done – let's say an investment agreement.

(We call her an end-user; a lawyer might call her a "client", but she probably doesn't think of herself as a client because she's already decided not to use a traditional law firm for a variety of reasons known only to her.)

She probably starts with Google, and finds a sample L4 program that does 90% of what she wants. Maybe it's in the Legalese tutorial, maybe in some open library of L4 program samples.

The L4 program compiles to an English-language contract, in her choice of US or UK jurisdictions.

But she wants that contract to compile to French, for Quebec.

Suppose the contract happens to be an investment agreement. Investment agreements are governed by company law. Each country has different company law.

So some work needs to be done. We need people to

1. locate reference materials (model agreements roughly corresponding to the desired output); if no reference materials exist, interface with a law firm to commission reference works for the target jurisdiction and language.

2. read the relevant legislation, and encode it into L4, under a library named "Quebec Law". This way, L4 can compute parts of the contract as needed to satisfy regulatory constraints.

3. while referring to the reference materials, teach the compiler and natural language generation engine to produce output for "Quebec French".

4. confirm that the L4 contract compiles to a semantically identical form of the reference source. This step may require coverage-oriented development and testing across the permutation space of the contract domain.

How will this work get done? Probably some combination of volunteers and paid staff / contractors / bounty hunters; we will find economic mechanisms to make the work happen in an opensource-inspired way.

How can you prepare to do this work?

For coders fluent in English, we recommend you read Ken Adams to understand the programming style conventions of legacy pseudocode. This will take at least a month, probably two.

For lawyers planning to work in L4, we recommend you learn a declarative / functional language such as Haskell to learn how programmers and mathematicians think. This exposure may offer some insight into how formal languages afford new modes of thought and methods of expression which will appear in L4. Other languages like Ruby, Python, and Elixir are also good options. Along the way you should pick up some exposure to predicate logic, which will put you in a good position to learn modal logic, which is one of the tools Legalese uses to put law on a more formal footing. This will take at least a year, probably two.

We also have a list of other readings which offer useful background on how we see the world.

High on that list: Richard Susskind.

If you already know how to program and already have background in legal, perhaps the most useful thing you could do is just keep participating on this list as we develop the L4 language.

Any thoughts?

--
You received this message because you are subscribed to the Google Groups "Legalese talk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to talk+uns...@lists.legalese.com.
To post to this group, send email to ta...@lists.legalese.com.

[Edouard E]

Hi !

Thanx for your feedbacks and the great links to some further education that I'll check and try to understand as soon as possible ! Maybe then I would be able to bring some help in the computing process, while right now I can only look at silently and learn!

But with my little experience wit creating databases to compile and produce simple contracts out of end-user's criterias, I have been facing at that time a problem I could not easily and in a 'flowing manner' (meaning an automatic way) solve unless using some 'screen alert'. That problem (that you may have already addressed) is "Conflict of norms".

I mean, making a contract is to produce the concatenation of distributed legal norms into articles, like :

Article 1: Who are dealing?
Article 2: What are we dealing with?
Article 3: How long are we dealing?

and so on.

And then contractors, in the limit of local legal system, can implement many different articles to set every rights, actions, obligations, boundaries possible to their contracting process. This is where pop up some conflicting articles within the contract that may nullifiy all or parts of the contract. And I don't even talk about "sided clauses" that can also invalidate contracts.

And sofar, I don't know how to transcript this ability (to avoid conflicts or unsteadiness) into an A.I.

Have you already figure out that computing parameters?

[Meng WONG]

On Mon, Oct 24, 2016 at 7:04 AM, Edouard E <emilio....@gmail.com> wrote:

But with my little experience wit creating databases to compile and produce simple contracts out of end-user's criterias, I have been facing at that time a problem I could not easily and in a 'flowing manner' (meaning an automatic way) solve unless using some 'screen alert'. That problem (that you may have already addressed) is "Conflict of norms".

Thank you for sharing the problem that you ran into. It is very important that we should work on live problems at the frontier, and there is no better way to demarcate that frontier than to hear from pioneers like yourself about the problems that you encountered.

Based on our conversations with industry players, it sounds like we are definitely at that frontier.

On the industry front, people working in the field of contract automation have run into exactly the problem you describe: there is value in verification … which depends on formalization … but there is no standard approach to formalization yet!

On the academic front, there have been a number of papers published over the past 30 years, but again no standard approach to formalization has emerged as the leader.

So we are in a situation like databases before SQL: we need an E.F. Codd to come along and put the field on a proper mathematical footing, from which we can build a language, from which we can build applications, with which we can build an industry.

Back to your question – we have not solved this already, but it is definitely a problem that we are actively working to address!

If you look at page 39 of https://www.microsoft.com/en-us/research/wp-content/uploads/2008/05/tr-2008-61.pdf

you will see Formal-logical models of norms and institutions by Marek Sergot.

The goal that you describe: "this ability (to avoid conflicts or unsteadiness) into an A.I." is alluded to at

http://www.ai.rug.nl/~verheij/AI4J/papers/AI4J_paper_9_agarwal.pdf

Validity Check: Validity checking determines whether a contract satisfies all the constraints it must satisfy. For example, a marriage contract between an adult and a minor is invalid in most countries because the law of the countries does not allow such contracts. Constraints that need to be satisfied can be defined with CDL as well.

The next step is to use formal methods to test programmatic contracts:

http://publications.lib.chalmers.se/records/fulltext/234939/234939.pdf

And then build some startups upon those methods:

http://www.adjoint.io/

It is an exciting time to be working in this field!

[Edouard E]

Thanx for these good readings!

The fog is dissipating a little on my side! I have to dig that Idris language to see wether i can get the juice of it!

So we are in a situation like databases before SQL: we need an E.F. Codd to come along and put the field on a proper mathematical footing, from which we can build a language, from which we can build applications, with which we can build an industry.

It would be better for my comprehension if I could gain insight on a specific problem raised by the un-fit between existing computer language and automating contracts. I don't know if I'm clear... Do you have a paper about a programmer describing what automation of contracts requests in general and what language cannot do yet in this field ?

Reading your documentation and some more i've found online, it striked me - in fact - that a contract is somehow structured like a computer program. Therefore, i think it is only (maybe) a matter of approach...

Thanx for your feedback!

Ed

 

 

Validity Check: Validity checking determines whether a contract satisfies all the constraints it must satisfy. For example, a marriage contract between an adult and a minor is invalid in most countries because the law of the countries does not allow such contracts. Constraints that need to be satisfied can be defined with CDL as well.

The next step is to use formal methods to test programmatic contracts:

http://publications.lib.chalmers.se/records/fulltext/234939/234939.pdf

And then build some startups upon those methods:

http://www.adjoint.io/

It is an exciting time to be working in this field!