Westlaw Next

Ashlee Vance, Jan. 24, 2010: NY Times Legal Sites Plan Revamps at Rivals Undercut Price Sees the new Westlaw Next as part of the longtime rivalry with Lexis and a new reaction against competition from Google and Microsoft.

Jill Schachner Chanen, ABA Journal Exclusive Inside the New Westlaw, Lexis & Bloomberg Platforms pretty much sees the new development the same way, as a reaction against Google and Bloomberg eating at the Westlaw - Lexis "duopoly." Schachner Chanen does a bit more in-depth reporting, but merely saying that Westlaw Next is "googlizing" legal research misses a great deal of what this product brings to legal researchers.

There have been several reviews, as above, of Westlaw Next (WLN, sometimes called Project Cobalt), the new version of Westlaw that will be released to the law firm market on February 1, 2010. My understanding is that law schools will be given passwords then for faculty and staff, but not students. The last I heard, they had not decided about law school pricing -- whether they would charge the law schools anything extra for the upgrade. The pricing is supposed to be as a “modest premium addition” to the original Westlaw, since it’s designed to make researchers more efficient and effective. That is, it’s partly designed to make good researchers more efficient and effective and to make poor researchers effective. It’s set up to automate behind the scenes what good researchers do, so that even poor researchers can have some pretty good results.

As a long-time teacher of legal research skills, I have to say it gives me some really mixed feelings. I worked hard to gain the skills I have and value and honor them and those I meet who likewise have honed their legal research chops. And yet, there is no reason why computing power should not be making many of these tasks easier for us all. This is what the dream of the future is all about. I suspect there will still be levels of skill. Reference Attorney Greg Garrick said his advice to formulate the best query would be “What would your ideal case say? If you can’t articulate that, go back and think a bit longer.” Since that is pretty much what you need to do now for a good Natural Language query, I think we will still see levels of skill.

The new Westlaw Next (WLN) has been in development for more than five years, and it shows. It is a very slick and well done piece of software. The programmers thought carefully about how researchers work and interface with Westlaw, and did huge amounts of research at all levels of use. They looked at useability; what causes confusion in users, and what makes sense to them. They did in-depth analysis of users’ real research logs, recreating the searches and looking for opportunities to improve the search, retrieval and ranking. They analyzed eye-tracking of large numbers of users, and found that bright, flashy designs actually drove the eyes away. Focus groups and design reviews, performance testing where real associates were hired assigned research tasks and paid as though it were a case for a regular client, all went into testing this product. They added all kinds of cool tools to increase user efficiency. They have folders to store the research results with drag and drop functions. The results can be stored for one year, shared with others in a group who have Westlaw passwords, searched with Westlaw power, and subdivided into other folders within the main folder, exported, and so on.

The final analysis is that users come to Westlaw for legal research for two primary types of tasks: 1) known document retrieval and 2) exploratory issue-based research. The first needs to be made as simple as possible. The second is the real heart of the matter. Westlaw researchers did a masterful job of breaking down the process of online research and then analyzing where their product either added extra difficulties or could improve.

Online legal research:
Step 1: Choose the database {WLN eliminates this!}
Step 2: Enter the search {Had to decide Boolean or Natural Language WLN eliminates choice}
Step 3: Scan the results
Step 4: Evaluate the results
Step 5; Browse results with Westlaw’s linking tools (KeyCite, Results Plus, Key Numbers etc)

*Repeat the above steps over and over in an interative process until you are certain you have found all relevant documents OR you have run out of time or money.

Choice of Database
You can still choose a database if you want to, but it really makes no sense to do so. WLN automatically searches all the Westlaw databases and locates any relevant documents in any databases for you. I recently stumbled across an unknown Massachusetts real estate database that had relevant information for a research problem, but I had not known of it because it is not listed in the Massachusetts secondary sources tab. It only shows up in the Real Estate Practice tab, under Massachusetts. This WLN method would have found the document without my stumbling upon it. You do not pay for the search; you pay for the access to the databases. So, if you search all the databases in Westlaw, and locate information in databases beyond your contract, you will have to decide if it is worth enough to access it. Fortunately, the display offers a good amount of text around the keywords for each document retrieved.

Search Format
This is another big change. You no longer need to select if you are searching in Terms and Connectors (Boolean) or Natural Language search style. The program can detect (in all cases except quotes around phrases or a single term search) whether the searcher intended to enter a Boolean or Natural Language search. If you are using connectors (/s & +2), the program runs a Boolean search for you automatically. Otherwise, the search runs as a Natural Language search, but adds more power.

WLN does not rely simply on keyword searching or relevance ranking, though those things are there. Where before, users had to guess the precise language of the document they wanted to retrieve, and (in Boolean searching) needed to know the syntax of they Westlaw system, the Westlaw Next system attempts to use the vast information of the West Publishing system and the vast hive mind of the Westlaw users to bolster each search. WLN uses the indexes, including “see also” entries to add extra terms to searches, uses treatises and other secondary sources to help develop searches from simple entries to create a better retrieval, all behind the scenes, without the user needing to know the special terminology or doing the “background reading” that librarians have so long counseled. But the system is further backed up with the West attorneys who test the system and tweak it constantly. They are not creating a taxonomy on the fly; they are letting the user lean back on the indexes, the treatises and other secondary sources, without realizing it. The system also uses the West key number system and KeyCite to bolster the indexing and help create a better search. Perhaps a concrete example or two from the Reference attorneys who tested this will help illustrate

Allison Stover reported on the test of multiple states' statutes on driving while intoxicated. Different states call it different things, and have different abbreviations or titles for the statutes: DUI, DWI, OUI. They called this the mis-matched term example. The indexes of the West statutes included "see also" references that assisted with formulating the search, and the search on WLN turned up the correct statute for each state as the top result for each state, while a natural language search was not as effective.

Reference attorney Greg Garrick discussed testing the problems of searching outside a familiar jurisdiction where the researcher is unfamiliar with the terminology, a special problem with statutory research, where legislative drafting often is non-obvious. His example was searching for the Minnesota freedom of information act, which has unusual terminology. Garrick also discussed the problems of searching where an issue can be described in many different ways. The Monell case, where a municipality is liable of the civil rights violations by its employee, could also be described as "imputed liability" or "respondeat superior," or "vicarious liability." Another example he gave was "Gift before death" and "Gift Causa Morta." WLN is an aid in locating all the documents that discuss the issue from these various terms. Garrick also mentioned spelling errors (forum non convenions (sic)).

WLN also uses massive customer usage data, in the way that Google uses it. Stripped of personal information, the system trackes the Print, Copy with Reference or with Cite and the KeyCite interactions of millions of users. Many users may look at irrelevant documents, but very few will print or copy with reference or with cite unless the document is very relevant to their search. So the system will add the search terms to such documents as (I suppose) metadata tags. These words may be reinforced by other users using similar searches and coming to the same documents, or linking to these documents through KeyCite. This creates other tags, and reinforces the correlations. Poor researchers wash out statistically (I am reassured), and good researchers reinforce the correlations, building stronger statistical links for documents. West programmers are tweaking their algorithms all the time, and testing the system. Editors can correct the system if it seems that it has run off the rails in a particular instance.

This is not “googlization” of legal research. There is some similarity because they are looking at customer linking. But they are building an enormous “back end” to this research system that looks simple at the front. The system does in the background all the things law librarians have wanted good researchers to do: do background reading, get extra terms, carefully choose database or even combine them, So it is also not dumbing down legal research. It is, rather, doing it for you, automatically. However, it still gives you the choice, and the tools to do the Boolean search, to use the field searching, and to do all the powersearching things that long-time skilled searchers have learned to do.

So why do I feel sad? I am trying to get over this. I should feel happy. I know I should .