Learning Something New: Understanding Long Short Term Memory Networks

LSTMs are explicitly designed to avoid the long-term dependency problem. Remembering information for long periods of time is practically their default behavior, not something they struggle to learn!

Source: Understanding LSTM Networks — colah’s blog

I wonder what would happen if one trained an LSTM network with a couple million court opinions plus code and regulations. Would it be able to answer even a simple legal question?

Pax releases Codex, a legal scripting language for Ethereum

Pax is using Ethereum to build a peer to peer legal system. The best way to explain Ethereum is by contrasting it with Bitcoin. Bitcoin’s value comes from the fact that every transaction that happens is added to the record and can’t be changed. Each full node on the network holds a complete copy of the transaction history, which eliminates any possibility of double-spending. It turns out that this approach can also be used to create binding self-enforcing legal contracts between people, which can have use cases as wide ranging as employment, rent, deeds, property transfer, restitution, incorporation, subscriptions, billing, voting and dividend systems. Where there are disputes, a blockchain can hold an objective record of events making dispute resolution relatively trivial compared to traditional legal systems, both on the back-end (legislation) and the front-end (arbitration).

Codex is a legal scripting DSL (domain specific language) geared towards creating executable contracts via the Pax directory and API. Codex is a flexible and powerful way for clients to interact with Ethereum, and is underwritten by Solidity (one of Ethereum’s most popular high-level programming languages) and the Web3.js, which is a Javascript library which allows front-facing apps to connect to the Ethereum network.


As with Ethereum the focus of Codex is self executing contacts, but it should provide hints at expanding blockchain into other legal realms.

3rd Hackcess to Justice hackathon winners include app built on #Drupal8

After two days of brainstorming and collaboration in North Carolina’s capital city, lawyers, students and coders developed legal apps to aid farm workers, streamline legal aid cases and evaluate legal

The $500 third place prize went to Michael Silverman, a developer. His submission of a Legal Aid eligibility test is intended to provide a simple, easy to use, mobile friendly interface that allows individuals to learn if they are eligible for legal aid by answering a few questions. Silverman plans to continue to work on the app to be accessible through voice prompts which will be translatable to different languages.


Source: App to aid migrant workers wins 3rd Hackcess to Justice hackathon :: ABA Journal

That third place prize went to an app built on Drupal 8 as a custom module. That’s pretty cool. All of the projects worked on are on the Hackcess To Justice submission page.

Because profit is more important than innovation, high school testing still relies on the TI-83 graphing calculator.

But this is also a world where high school math students have to shell out $100 for the same TI-83 graphing calculator that their parents used twenty years ago (or one of its descendants, at least)—instead of using a free app that they could simply download to their phone. Why? Mic reports that the main reason is tradition. Texas Instruments has managed to get its calculators written into the standardized tests used by many schools. And inertia being what it is, it’s really hard to change something like that once it gets set down on paper.

Source: In an age of tablets and e-books, high school testing still relies on the TI-83 graphing calculator | TeleRead

This should serve as a reminder that it isn’t just the legal world that drags its feet when it comes to new and obviously better technology. The world is awash in examples of this sort of thing where a powerful incumbency holds back or outright blocks the adoption of new tech simply to preserve some profit margin. Ignoring, disregarding, or suppressing innovation in the name of maintaining profits especially in a near monopoly market is practically a rule of business.

Common Form aims to bring legal drafting and collaboration to the browser

/dev/lawyer Common Form http://writing.kemitchell.com/2015/02/09/Common-Form.html

First, it aims to make drafting generically, as one would draft a form or template, more efficient and reliable than drafting expediently for one deal and one deal only. This is possible because an out-sized part of drafting is glaringly menial paper shuffling, on the one hand, and haphazard grasping, on the other. Untold lawyers flail, even now, in haystacks of TNR-12, wondering what needle-point technical errors lurk within. The better off delegate that hunt to hired help so they can rack brains and files for that elusive, perfect such-and-such clause seen, written, or stolen some years back. The would-be early-adopter types among us pay out for clunky third-party tools, get bitten (again) and adjust practices or expectations accordingly, lose the faith.

Common Form aims to expunge these experiences from the practice of law, and to collapse the long cycle of incremental improvements to the state of the art by forms committees, CLE handout scrambles, and traditional publishing deals. Public goods in law ought to be cheaper and easier to make.

Second, Common Form aims to make verification and sharing of contract content free, reliable, and instantaneous. Many a lawyer has reached the end of a non-disclosure agreement only to realize that, yes, it is in fact the same agreement they have read and approved, from top to bottom, many times before. Conversely, many a bespoke drafting project has devolved to a second-rate knock-off of a standard form Not Invented Here, then been thrust into circulation nonetheless to justify process or bill, polluting the ecosystem. Clients pay dearly for such duplication, sometimes unjustly, and good lawyers find no joy in the taking or making. Everybody drinks it off, and the wheel keeps on turning.

The code is open source and available on GitHub at https://github.com/commonform. I’ll be taking a good look at this and will report back.

Sign legal documents with your finger, your iPad and Softsign