Westlaw’s Moat

Westlaw has been more or less the same since the 1980s.

Lawyers often grow attached to the version of Westlaw that they grew up with; to take an extreme example, at least one judge on a federal court of appeals still uses command line Westlaw instead of the web version. Westlaw Next offers some handy annotating tools and a more modern interface, but also breaks key parts of Westlaw like database selection and a precise terms-and-connectors search where you feel you have control over what results are returned. I grew up with Westlaw and still prefer it to Next.

Lots of people have tried to compete with Westlaw on price: FindLaw; Lexis (formerly a major competitor, now not so much); Bloomberg; LoisLaw; Google Scholar. Others have attempted or are attempting to compete with Westlaw on search results: PreCyDent; Ravel Law; Judicata. PreCyDent used citations to weight search results, in an analogy to PageRank. Ravel Law uses an innovative (but not really useful in practice in its curent version) graphical interface to show how cases relate to (cite) each other. Judicata has a not-yet-unveiled technology for parsing cases and using the resulting structured data to improve search.

How does Westlaw compete with these challengers; how does it win despite being more expensive and providing worse search results? Content is Westlaw’s moat. The top of the market — great lawyers — that these companies are targeting cannot use a product that does not contain all the law that might be relevant: cases from all courts, going back to the dawn of time, and all statutes and rules in all their versions, going back to the dawn of time, at the very minimum. A search engine, no matter how good, that searches only 80% of the law (and most competitors have far less than that) is almost worthless — because a great lawyer can’t afford to have missed something that was in the other 20%.

(One way of putting this is as the difference between accuracy and recall. When you search Google, you want high accuracy, meaning that the results you get back and read, the first 10 or 20, are very much related to what you were searching for. You don’t care that there were many other also relevant things that you never see. When you search for primary law, you want high recall, meaning that if it exists, and its responsive, you will eventually get it, even if you have to spend a few hours reading through results to separate the wheat from the chaff.)

Isn’t this information, what lawyers call “primary law,” meaning the authoritative output of courts, legislatures, and agencies, freely available? Yes and no. It is relatively easy to get current opinions, statutes, and rules, although there is no standardized format and you would have to work separately with the many federal and state governmental bodies that promulgate primary law every day. But it is not as easy to get the historical stuff, opinions going back hundreds of years, earlier versions of statutes and rules, legislative deliberations, and the like.

None of this material is copyrighted, so a team with Google’s resources could simply buy and scan a complete set of books (an old school legal library), purge the scanned text of copyrighted material (annotations and the like) and thereby recreate West’s database of content. Or Google, which has perhaps done this for Google Scholar, could make its database public (mass content retrieval through Google’s APIs would violate the terms of service, and Google would cut you off long before you were done — a client of mine tried). Or the Government could undertake to create this resource, a free database of primary law, and release it to the public.

Until Westlaw’s competitors invest in creating or otherwise obtaining such a database, or until an enterprising technology company simply buys Westlaw for its content (and its customer relationships), I predict we will not see any serious movement of customers in the search space. Content is king.

No pre-work

Apple’s slogan a few years ago for OS X was that “It just works.” Apple didn’t provide instructions or manuals, and by and large users didn’t need them.  “It just works” is a great goal for a product, and we want users to think of Disco like Apple wanted people to think of OS X; “It just works.”

Other companies don’t seem to share that vision.  kCura, for example, has an entire website dedicated to Relativity documentation necessary to use the product.  No fewer than thirteen (13) reference guides can be found on how to use the product.  Even beyond those thirteen guides, users can find an explanation of how to use the “Starter Template.” The Starter Template, they say, isn’t required to use the software, but it does allow you to logically customize the software.  It seems to us that software should be logically set up in the first place so customization isn’t really necessary, and logical customization should never be necessary.

Before a user can use the Starter Template, though, that person has some “recommended pre-work” to do.  kCura recommends users first complete Relativity Admin training, Infrastructure setup, Data processing, and Client, Matter and Workspace creation.  Only then do they think users are ready to use the Starter Template.  That’s a lot of pre-work just to be able to customize software so it seems logical to the user!

We think no pre-work, no lengthy manuals, and no time consuming training should be necessary.  Software should work immediately and intuitively.  Software should also allow users to simply do their work without making users do pre-work.  Maybe we should adopt a slogan for Disco: No pre-work required, but a post-work beer recommended.

Counting Backward

Rule 4 of the Texas Rules of Civil Procedure governs “computation of time,” that is, if the rules require that something be done within a certain period of time, Rule 4 tells you how to calculate that period of time.

Rule 4. Computation of Time

In computing any period of time prescribed or allowed by these rules, by order of court, or by any applicable statute, the day of the act, event, or default after which the designated period of time begins to run is not to be included. The last day of the period so computed is to be included, unless it is a Saturday, Sunday or legal holiday, in which event the period runs until the end of the next day which is not a Saturday, Sunday or legal holiday. Saturdays, Sundays and legal holidays shall not be counted for any purpose in any time period of five days or less in these rules, except that Saturdays, Sundays and legal holidays shall be counted for purpose of the three-day periods in Rules 21 and 21a, extending other periods by three days when service is made by registered or certified mail or by telephonic document transfer, and for purposes of the five-day periods provided for under Rules 748, 749, 749a, 749b, and 749c.

The rule works like this: for periods over 5 days, you do not count the day of the event that triggers the period and you count every day after that. If the last day is a Saturday, Sunday, or legal holiday, then you get until the end of the next day that is not a Saturday, Sunday, or legal holiday. So if you are required to respond to interrogatories within 30 days of the date on which they are served, you do not count the day they are served, you count every day after that, and the period ends on the 30th day — unless that day is a Saturday, Sunday, or legal holiday, in which case you have until the end of the next day that is not. For periods of 5 days or less, however, you do not count intervening Saturdays, Sundays, or legal holidays, so that, for example, if you have five days from an event that happens on Thursday, you count Friday, you do not count Saturday and Sunday, and you count the next Monday, Tuesday, Wednesday, and Thursday so that your deadline is the Thursday after the event.

Complicated enough? Now consider what happens when you have to “count backward,” that is, when the deadline is something like “7 days before trial.” Do you literally count backward, so that you start, in the above example, on the date of trial, do not count that date, then count seven days back, and if the “last” day (which is earliest, because you are counting backward) falls on a Saturday, Sunday, or legal holiday, then you continue counting backward to the first day before that that is not a Saturday, Sunday, or legal holiday? Or does Rule 4 mean literally the “next” day that is not a Saturday, Sunday, or legal holiday, so that the provision for those kinds of days actually shortens the period in the case of counting backward when the last day is a Saturday, Sunday, or legal holiday (i.e., you count backward, hit one of those days, then count forward to the next day that is not one of those days)? Or does rule 4 simply not apply to counting backward?

Texas courts disagreed about this question; it was ultimately decided by the Supreme Court; and the answer was “none of the above.” In Sosa v. Central Power & Light, 909 S.W.2d 893, 895 (Tex. 1995), the Supreme Court held that Rule 4 means what it says and therefore applies to all time periods under the rules. But in a “counting backward” situation, that is, one in which you have to do something a certain number of days before something else, you apply Rule 4 by counting from the date you do the thing and then counting forward. If the count ends on or before the deadline event, then you are timely; otherwise not. So, for example, if you have to amend a pleading “7 days before trial,” and trial falls on a Wednesday, you start on the day you file the amendment, do not count that day under Rule 4, then count every day after that, and if your count hits 7 on or before Wednesday, you are timely. It doesn’t matter whether the first day (the day you filed the amendment) is a Saturday, Sunday, or legal holiday, because Rule 4 extends the deadline only if the last day, not the first day, is one of those days. See also Melendez v. Exxon Corp., 998 S.W.2d 266, 275-76 (Tex. App. — Houston [14th Dist.] 1999).

New rule 91a (a state motion to dismiss analogous to federal rule 12(b)(6)) provides that a respondent can nonsuit his claim, rather than face a ruling on a motion to dismiss and a mandatory award of attorneys’ fees, “at least 3 days” before the hearing. Under Sosa, Rule 4 applies; under Rule 4, because the period is less than 5 days, Saturdays, Sundays, and legal holidays are not counted; and under Sosa, counting starts on the day of the nonsuit and goes forward (not on the day of the hearing going backward). So, if the hearing is on Monday, the last day to file the nonsuit is Wednesday: you do not count the day of the filing, and you count each day after that, except that you do not count Saturdays, Sundays, and legal holidays because the period is less than 5 days. For a Wednesday filing, that would be Thursday, Friday, and Monday, which is the day of the hearing, so you have filed “at least 3 days” before the hearing. A nonsuit filed on Thursday would be late — and I have seen litigators with 10+ years’ experience make this mistake, exposing their clients to dismissal with prejudice and mandatory attorneys’ fees.

Never assume that something like “calculation of time” is clerical, or even that a careful reading of the rule alone will give the correct answer, or you might learn the hard way.