The details of the WEXIS systems aren’t public, but you can certainly read between the lines of stuff that Peter Jackson and others have published. One guess might be “SVM-based voting systems supporting human decisions”. Another might be “ask Dan Dabney” (grin).

As for near-term feasibility, depends, because the complexity of the problem varies a lot with subject area. Areas of law that are children of statute can hit 95% accuracy using Bayesian classifiers (I think we got close to that in some experiments about 5 years ago). Constitutional law (eg.) is a lot harder because most of the good discriminators also yield a lot of false positives. Citation networks, a la PreCyDent, would help a lot.

A second variable is, simply, how good it has to be. If we’re only talking about defining reasonable scopes for full-text search, it’s a lot easier. If we’re talking about simulating key numbers, probably never.

I do indeed need to go back and look at those efforts. One quick thought: I am imagining that this was an effort at a centralized service, and centralized services don’t fit the contours of the Net very well, especially when in terrain that houses pesky things like federalism, separation of powers, and judicial independence. Which is why it is stunning that we have learned so little from digital repositories in the sciences…. where distributed, federated efforts are the norm. The problem is not so much building a CALR service as it is building a standards-based CALR ecology, perhaps.

Thanks for your thoughtful response. One more comment: I’ve been re-reading recently all the AALL documents from the 1980s & 1990s respecting the unsuccessful attempt to create a free public CALR service centered on the Law Library of Congress. It might be well for all of us interested in free public access to law to revisit all that material and assess it, to determine whether any of that program can be salvaged in this new environment. Perhaps one feature of this new policy proposal that might differ from the old one is that it could involve multiple sources of material (e.g., the LIIs and university-based open legal digital libraries like Rutgers), not just a single new CALR system based at Law Library of Congress. I think the two key works are “Setting The Legal Information Agenda For The Year 2000 : Based On A Workshop . . .” / edited By Mary Kathleen Price, Margaret Maes Axtmann, http://www.worldcat.org/oclc/23940295&referer=brief_results ; and “The National Conference On Legal Information Issues : Selected Essays” / edited By Timothy L. Coggins National Conference on Legal Information Issues (1995 : Pittsburgh, Pa.), http://www.worldcat.org/oclc/34549094&referer=brief_results . A word of caution: the last attempt was very well organized, took up a huge amount of time and effort, and appears to have accomplished very little.

A lot of the problems with current public systems can be put down to a sort of echo-chamber or boy-in-a-bubble syndrome that stems from the use of Premium Service by policymakers and others in a position to influence the creation, development, and distribution of information to the public. In current biz-speak, you’d say that those in a position to control legal information policy aren’t eating their own dog food.

Now, like you, I would neither deny people the professional tools they need, nor overburden casual users with complexities they don’t want, nor try to kill off the private market in value-added legal information. The question, though, is how we manage to keep information providers in tune with the needs of a general public that is — so far as I have been able to observe — far more sophisticated and diverse in its use of legal information than the information providers believe.

And, of course, there is an imp within me that honestly believes that public transportation in the United States would improve by three quantum levels if the Washington Metro were shut down for a month, with signs posted at the entrances that say, “This is what everyone else has”. That same imp (busy little devil) makes me wonder what the reaction would have been within government had someone come along in 1987 and said, “the general public needs access to a worldwide system of interconnected textual databases on a wide range of political, social, and economic topics”. Probably something along the lines of, “we don’t have anything like that, we can’t afford to build it, and in any case it should be a Premium Service for the intelligence community”. But now here it is.

So, obviously, this is a matter of arriving at some public policy around the Basic Service. What should it be?