At the Aaron Swartz memorial in San Francisco (video)
, some interesting themes emerged.
The first is Aaron's passion for machine-readable public information. This principle is at the core of much that Aaron did, from enabling search engines to find public domain and CC-licensed content to downloading swaths of paywall-guarded documents so that the public can have access to its own information.
The second is the unbalanced power wielded by prosecutors. Aaron killed himself in part because he felt helpless when faced with a multimillion dollar federal trial featuring 13 felony counts. If Aaron couldn't face this, what hope have ordinary folks who aren't close friends of Harvard law professors, rights advocacy organizations, and expert witnesses, not to mention a chunk of cash from selling an Internet startup. Faced with expensive defense lawyers and the fearsome specter of the government's prosecutors, only 3% of cases make it to the trial which we're constitutionally promised.
The third, expressed by his girlfriend Taren Stinebrickner-Kauffman
and fellow document liberator Carl Malamud
was a call to the technologists and scholars and activists to become radicalized. Aaron did big things because he thought they mattered. Like Peter Singer, he stressed about the opportunity cost of not doing the most important thing in the world. His death has become, in part, a call for people in the free culture movement to step up and do more.
So here's an interesting challenge that combines all three: write a program that interprets and presents law. Though they predate computers by a few thousand years, laws are meant to be something like human-runnable source code. They're detailed, they're written explicitly, and they apply to everyone. And yet in many cases it takes someone with a graduate degree to understand what they say. People with graduate degrees are expensive and unevenly distributed.
Imagine we had a program which could turn laws and judicial opinions into machine-readable format. We could then write programs that took those laws and presented them in various ways, helping lay people understand both core details and subtle interactions. We could write other programs to organize this legal information into arguments given the evidence about a case.
Compared to people with graduate degrees doing stuff
, running computer programs is free. Someone without a lot of resources could understand what they're charged with, explore similar cases, and collaborate with friends on a defense. There'd still be a role
for lawyers to conduct the defense at trial and advise on the best way to convince a jury, but the time spent at trial is today dwarfed by the time and expense preparing for it. Let the humans do what they're good at&endash;convince humans of things&endash;and let the computers do what they do best&endash;tirelessly and cheaply examine lots of data and find useful patterns.
Like a patient who comes to a doctor after reading the medical literature and closely observing his body, a defendant who comes to a lawyer with a solid understanding of the relevant laws is in a much better position to face the plaintiffs and prosecutors who have the deck stacked in their favor. If we can make computers understand law, we can empower all citizens, regardless of income, to make fair use of the due process granted them by the constitution.
Building such a system wouldn't be easy. Human language is still hard for computers to understand. And legalese is even hard for humans to understand. There are all sorts of powerful people and organizations, private and governmental, with interests vested in law and courts being expensive and difficult to access. It's not easy, and that's why it should be done. A hard, ambitious, and meaningful project like this would capture the spirit that's been raised in Aaron Swartz's wake.