IBM’s General Counsel, Robert Weber, explains that while the technology behind Watson, the Jeopardy-playing supercomputer named for IBM’s first CEO, “could become a real boon to the legal profession” it “won’t ever be able to replace lawyers.”

Reference Tasks are teachable. In minutes, most high school students can be trained to retrieve a known Federal or state statute, judicial or administrative rule, form, or opinion. That’s what makes such tasks perfect for robots like Watson.

If you use GPS, you already know what it’s like to have a digital associate. Speak or enter your destination and follow the arrow. Even alternative routes are fully quantifiable.

In the practice of law, you know you’re performing a Reference Task when the objective is to retrieve a known item of information. As Weber explains, such tasks are ripe for robots.

A Research Task is the opposite of a Reference Task, because the needed item(s) of information are unknown to you.

You know the code for the murder statute and you pop it up. It’s as simple as retrieving something from the fridge.

But, you don’t know:

• how many judicial opinions in your state, if any, expressly cite your statute
• how many judicial opinions, if any, cite your statute for the individualized reason you care about
• which of these opinions, if any, are the most/least cited in relation to each other and why
• which of these opinions is the seminal case and which is the last word
• which of these opinions, if any, mirror or trace your fact pattern
• how the opinion(s) you are most interested in have been interpreted since the date of publication and why
• how the opinion(s) you are most interested in have been treated in non-binding jurisdictions and why

The retrieval of a known item of information is resolvable only to “yes,” whereas the search for unknown items of information are resolvable to yes/no/maybe.

For the 21st century legal researcher, this is not some fine point. Reuter’s recently released WestlawNext which relies, in part, on so-called “crowd sourcing.”

In the days of books, West Publishing built its brand on the back of employees who created various case finders that, on their best day functioned as a partial index of opinions.

It is only with the advent of compusearching that we now realize what we were missing. Don’t take my word for it. If you have access to a national plan, take your favorite oft-cited Federal statute, enter: [title number] /5 [section number]  - 18 /5 1346 for example – and make a note of how many hits you receive. Then pop up the annotations to the same statute and count them. Based on my experience, you will see a variance between the actual number of opinions that cite your statute and the number of referencing headnotes/annotations.

You should also find that many judicial opinions that give rise to the headnotes that link you to a particular key number or annotated statute are not included in the resulting index.

(Due to software licensing restrictions, you need to run these tests on your own.)

As a young litigator, if my search was driven by a statute, I grabbed one of those little brown books and turned to the page that listed the opinions citing my statute. It didn’t get any better than that. I knew that I’d found everything that was findable. I didn’t know what I was missing and neither did the judge or opposing counsel.

Today, I don’t have to wonder. I enter the code for the statute or I enter the citation for a case in quotation marks and miss nothing.

The reason crowd sourcing has no place in legal research is that it skews otherwise pure results. I have no interest in results that have been skewed by inferior researchers. I have no interest in search engines that distract me with advertising; whether it’s a banner ad or a series of monetized links to documents my infomediary wants to sell me. I am not alone in this concern:

“Have come across some very odd results with the West relevancy ranking. The most critical documents are often way down the list.  Associates should not rely on it blindly.” ~Summary of Law Librarian Opinions As To WestlawNext

In 2011, there is no need to rely on happenstance. Legal Research Tasks can be as process-driven and precise as Reference Tasks. You just need the right tool.

“I need all available Arizona cases where a school district has been held liable for the wrongful actions and assaults of a ‘bully student’ who injured another innocent student. In an important fact my case is that the bully’s parents work at the charter school and not surprisingly, the bully was never disciplined, suspended, etc.  Thus, school district should be held vicariously liable for bully’s actions and for bully never being disciplined.”

We know from the question that the binding jurisdiction is Arizona. It’s a state civil action. This search is driven by Concept, not Code. So, unfortunately, at the outset of the search we know we are limited to keywords.

I like the words: bully, vicarious, school. I can tell from the question where my subscriber has gone wrong. He searched on the phrase “bully student” found nothing and submitted his request for assistance.

For three reasons, I know before I even begin that I’m not going to find what he wants in Arizona: (1) Arizona state courts generate fewer than 2,000 citable opinions annually, (2) bullying is trending in litigation since Phoebe Prince and (3) this is the sort of thing that settles 99% of the time. As a general rule, you can’t have a citable opinion without an appellant.

My subscriber committed the sin of over-searching when he entered “bully student.” All I have to do to help the guy is activate the Arizona State Database only and search on bully. My results indicate that Arizona has less than five judicial opinions containing the term bully and they are all criminal cases.

In a click I Go National, searching on: bully and school and vicarious and receive 14 hits.

This is a start: fourteen citable opinions that contain at least on instance of each of our three terms somewhere in the opinion.

I email the query to my subscriber. I say: “If you agree with me that any opinion with the potential to resolve you search to ‘yes’ or ‘no’ must contain at least one reference to each of the three terms in our query, then we’ve covered the landscape for now.”

Resolving this question – for now – took less than thirty seconds. I can tell that my subscriber is going to be in the analogy business in this litigation. He’s going to have to research instances in which governmental instrumentalities are held vicariously liable for bodily injury inflicted upon an individual in their custody by another individual in their custody. I’m thinking prison cases – where a habitual bully inmate repeatedly attacks model prisoners and prison officials are guilty of willful blindness, reckless disregard – that sort of thing.

One of the most common mistakes people make when constructing a query is looking for too much too soon. The goal of legal research is not to find opinions that support your argument. The goal is to resolve your search to “yes” or “no.” The cases say what they say. The goal of every search is to avoid missing relevant law.

This search, for now, is resolved to “no” as to the binding jurisdiction and “yes” as to non-binding jurisdictions, including one opinion from a bordering state. California holds no more sway over an Arizona court than Maine. But, judges are human, California co-exists with Arizona under the same Ninth Circuit umbrella and the emotional connection to time and place can itself be persuasive.

In my home state of Vermont, which basically has far less citable law, practitioners are constantly looking next door to New York and Massachusetts for persuasive precedent.

The main reason I started TheLaw.net on January 1, 1999, was to ensure that every attorney had searchable access to all 315 Federal and state jurisdictions. Before TheLaw.net, my subscriber would have been limited to his Arizona Slice of West or Lexis. TheLaw.net changed that, redefining the entrepreneurial online law office. As a result, my subscriber has ready access to 14 opinions that may help or hurt, with no upsell, no distracting advertising and no monetized hyperlinks.

He may not cite them in his brief or memo. But, they may well influence the deposition of school officials and oral argument at a motion hearing or otherwise.

That’s up to him. For now, I’ve done my job and he’s on his way.

INTRODUCTION

Most of the knowledge contained in this post is portable for users of Westlaw, WestlawNext, Lexis and TheLaw.net. This post shows you how to quickly resolve any judicial opinion search to “yes” or “no” in just a click or two. This post is intended to put some meat on the bones of the foregoing decision tree graphic. I designed this decision tree in 1999 and have since used it on a daily basis.

I am the last stop on TheLaw.net’s Virtual Support Desk. I field questions from every practice area and jurisdiction. My decision tree makes it possible for me to know precisely the process I will deploy, even before I hear the question. Ninety-eight percent of support requests are resolved by others. About two percent are ratcheted up to my Inbox. If a subscriber can’t find something we promise to find it for them. If they’re struggling with a query, we build the query for them. Our goal is to keep our subscribers moving forward by quickly resolving their questions to “yes” or “no.” If a question makes its way to my Inbox, it’s a tough search. Sometimes it’s a search where they can’t find anything on point and they just need someone to confirm there’s nothing to find. Resolving a question to “no” is just as useful as resolving it to “yes.” The goal of every search is to ensure you never overlook relevant law.

BACKGROUND

Even though I oversee the marketing and operations of the most successful American owned legal research company, to this day I am a recovering book researcher. Book research was process driven. Everyone used the same two or three indexes of key numbers, statutes annotated and little brown Shepard’s books. Remember pocket parts? How would you like to have the pocket part concession today? Me neither!

The nice thing about book research is that you knew when you were done and you knew you didn’t miss a relevant case.

Three or four years ago, in the course of developing the latest version of TheLaw.net, I decided to compare Westlaw’s compusearch results with Westlaw’s handcrafted annotations and headnotes. The results – or I should say, the lack of results – knocked me out. Like you, I believed Westlaw when they told me their annotations were complete and comprehensive. For all I know, Westlaw believed its own claims, too.

If you’re a Westlaw user, I’d invite you to run some similar tests. I am most familiar with Federal criminal and civil rights law. The Federal statute I chose was the wire fraud statute. I found about 3,500 opinions nationally containing at least one express reference to the wire fraud code. But West’s Headnotes had less than 1,000 entries.

In the days of books this was no big deal. We used the same tools and missed the same cases. But the only way we could avoid reading all the opinions in the library was to use what we now know were inferior handcrafted tools like headnotes and key numbers and annotations. In my opinion, Reuter’s implicitly recognized this point with the release of WestlawNext. WestlawNext is similar to Westlaw in that the interface still looks like the inside of a cockpit. It still looks like the programmers are running the asylum. That said, WestlawNext at least feels like you’re searching on a computer, whereas Westlaw feels like book research on the computer.

That’s why I started TheLaw.net in 1999. Navigating Westlaw was like navigating a law library with the books all over the floor.

HOW LAWYERS SEARCH TODAY

Survey after survey makes plain that today the majority of legal researchers begin with Google because it is free and easy. The public web is a great reference tool. But, there is a difference between “reference” and “search.” We “reference” known items of information. We “search” for unknown items. Google is a great reference tool. It’s even a great research tool when you need a cite to fill in the blank on a routine 404(b) motion. I use Google, Bing, Yahoo and other advertising driven search engines all the time when supporting subscribers. I can’t get the perspective of a blogger in a commercial case law database. When it comes to unpublished opinions, sometimes the only place you find that item of information is on the freeweb, if at all. Commercially searchable collections of unpublished opinions vary. TheLaw.net has documents Westlaw and Lexis don’t have and vice versa. “Comprehensive” is an industry term of art that refers to published judicial opinions. As for Google, as long as it is a slave to advertising it will never, ever be a tool for serious legal research. Moreover, it is very difficult to impose your individual will on Google as they tweak their algorithm hundreds of times weekly. Free search engines are not maximized for case law research. They are maximized for advertising.

This superpost is for serious researchers focused on crafting the best possible work product.

MANMADE TAXONOMY v. ROBOT CREATED TAXONOMY

Humans create partial indexes of judicial opinions. Partial, by definition, means incomplete. Without the various casefinders created in the days of book research, researchers would have to read all the opinions in the library to resolve their question to “yes” or “no.”

The search engine robot, by contrast, indexes every word of every opinion. The ultimate question answered by this post is: “How do I find every opinion in the database that mentions my Code or Concept for the reason I am interested in?”

Before we deal with this question, let’s first understand our dataset. It doesn’t matter whether you use Westlaw, Lexis or TheLaw.net, your database of judicial opinions is tiny. It represents a small fraction of the law.

In the aggregate, Federal and state case law consists of less than six million published and unpublished entries. Ninety-eight percent of civil suits settle, resulting in no opinions. Ninety percent of criminal defendants plead guilty, resulting in no opinions related to pre-conviction proceedings.

When you search a database of case law, you have motivated losers to thank. If Gideon didn’t write his letter to the Supreme Court, there would be no Gideon v. Wainwright. Gideon, like all criminal and civil appellants, was a motivated loser.

The point is that the vast majority of law is unknown to us. My home state of Vermont generates less than 200 published opinions per year. The Vermont Statutes, by contrast, number in the tens of thousands.

The United States Supreme Court generates half the number of opinions annually as those generated by the supreme court in my tiny home state. The nearly 100 Federal districts, when aggregated, annually stamp fewer than 5,000 opinions for publication. Nearly 85% of all Federal opinions originate from the thirteen Federal circuits and the vast majority of them are unpublished.

The vast majority of Federal and state statutes and regulations have never been construed by a court. Another enormous subset has been mentioned only a handful of times. Another huge subset has been mentioned less than fifty times.

To the extent the practice of law is about drawing inferences and making analogies, it is because most of the law is unknown to us.

HOW TO RETRIEVE UNKNOWN OPINIONS

1. A Known Opinion is easy to find. Finding it is a reference task. Both TheLaw.net and WestlawNext have Smart Boxes. Westlaw, in contrast, gives you a lookup box, a validation box, a keyword box and a natural language box. Users of TheLaw.net and WestlawNext start every task from the same spot. (TheLaw.net, incidentally, had a Smart Box several years before WestlawNext. I’m not judging. I’m just saying!)

2. When we talk about legal research, we are talking about retrieving unknown documents from a database. For all of the CLE courses, for all of the law librarians, for all of the West and Lexis trainers trolling the campuses of all the law schools, no one has said: “Here’s the right way to search for Unknown Opinions.”

3. Judicial opinions consist of Codes and Concepts. Codes are: statutes, rules, regulations and citations.  Concepts include: facts, statutory and regulatory terms cribbed from the black letter law, terms cribbed from Federal and state constitutions and common/judge-made law.
(Personal Privilege: Lawyers are trained to think of Codes are code. They are not trained to think of Citations of code. Someday, the gatekeepers of legal research (I’m talking about you law librarians) will drop the phrase “CALR – Computer Assisted Legal Research” and simply call it “legal research.” Beginning in 1822, brief writers began abandoning the feather in favor of the dip pen. I have no doubt that the then fledgling law librarian community must have tried mightily to hasten the change by creating acronyms such as “DPAW” for “Dip Pen Assisted Writing.” If you tell me you’re doing legal research, I’ll just assume you’re using a computer. If not, just tell me you’re doing some “BAR” – “Book Assisted Research.” I feel better now.)

ANCHOR + FILTER = HOTLIST

When I field a subscriber’s legal research question – regardless of the practice area or jurisdiction – I ask myself: “Is this search driven by a statute, rule or regulation?” Seventy five percent of the time the answer is “yes.” Half of all searches are driven by statutes.

(If the answer is “no” then I know that I am temporarily dealing with a search that is driven by “concept.”)

Any search that is driven by a statute, rule or regulation is easy.

Just Anchor your query with the statute number and click SEARCH.

–Most states assign a unique number to each statute. This makes searching by statute number super-easy.

–Codes lead to desired results. Words lead to diffuse, ambiguous results.

–A few states and the Feds use a title/section format. If my search is driven by a Federal statute I enter: 18 and 1344 and click SEARCH.

—-Note: Modern case law search engines give primary weight to search term proximity. You don’t need to worry about so-called Within Connectors. Within Connectors are dead. As a general rule you should never use them. Similarly, you should never us /s or /p or /10. Search engines come preloaded with proximity logic and rank opinions accordingly. All such queries lead to arbitrary results. You don’t need to worry about whether the court wrote: 18 U.S.C. section 1344, or 18 U.S.C. sec. 1344 or 18 USC 1344 – you get the point. 18 and 1344 does the trick. If you must, go ahead and enter: 18 /5 1344 or 18 w/5 1344 which asks for the section number within five words of the title number. Just know that when you do this you are actually overriding the algorithm which is completely unnecessary. (The only time you should ever use a Within Proximity Connector is when you have actual knowledge of the range. search /2 seizure is a great example. Search engines see the word “and” only as a connector. search /2 seizure is a nice way to deal with this issue.)

Second, if I find too many cases, I filter my search by Concept.

Using the example above I would simply enter, for example: 1344 and “false loan application” and click SEARCH.

—-I can drop the title number from my query. The section number when combined with a term related to the facts or circumstances of my case is enough to generate my individualized Hotlist.
—-I search binding jurisdictions first. If I resolve my search to “yes” or “no” I’m done. If my binding jurisdiction search is inconclusive, I go national.
—-In Federal searches, once I have my Hotlist, if I want additional insight, using the same query, I expand my search to non-binding jurisdictions.
—-In state searches, once I have my Hotlist, if I feel like I need additional insight, I will replicate my search in selected states using analogous statutes.

Again, the purpose of asking “Code or Concept?” is that numbers lead to predictable, comprehensive results and words lead to diffuse, ambiguous, unmanageable results. Words combined with numbers lead to individualized results. 404(b) and limine is right. Limine is wrong.

WHAT TO DO WHEN YOU FIND A CASE YOU LIKE

Like WestlawNext, TheLaw.net automatically links you to every opinion in the database citing your opinion for any reason. (We actually did this three years before WestlawNext.)

If you’re using a database with a Smart Box that provides access to all Federal and state judicial opinions, and you have a case you like, Anchor your query with the citation and Filter by statute.

“100 F.2d 200″ and 1344

(Be sure to put your citation in quotes so the search engine sees it as a phrase)

The gatekeepers of legal research and apologists for the premium brands, sometimes refer to this as a a “Poor Man’s Shepard’s.” It is quite opposite. Citation indexes were essential to research in the last century. Today, they are a crutch; a one-size-fits-all, mass-customized holdover for the attorneys who openly brag that they still don’t know how to turn on a computer.

An effective legal research class would teach every student how to say: “Yo Robot – Give me every case that cites my case for my point of law!”

You can’t do that with KeyCite. KeyCite is preoccupied with things it wants to sell you.

Everybody knows that appellate opinions typically resolve three to five questions. Who doesn’t want a Hotlist of every opinion, from your binding jurisdiction(s), citing your case for your reason? I mean, it doesn’t get any better than that. That’s why we call it a “Hotlist.” You want to check every opinion on that list.

Citators like KeyCite, do not allow you to individualize the validation process. The Smart Boxes provided by WestlawNext and TheLaw.net allow you to do just that. (This assumes your WestlawNext subscription includes all Federal and state case law.)

If you want to know whether your case is still “Good Law” KeyCite spits out a list of every opinion in every jurisdiction that has ever cited your case for any reason, together with a list of secondary resources you don’t care about and a list of legal memoranda West mined for free from Pacer that it now wants to sell you and it makes you browse that list. Remember, you just want to know if your opinion is Good Law.

Validation is about three things: jurisdiction, citation and point of law.

Using TheLaw.net or WestlawNext, activate your binding jurisdiction databases, enter your citation and statute number (or concept) and click SEARCH. (You may be able to do this on Lexis. I haven’t checked. The reason I haven’t checked is that whenever I use Lexis I have this sudden urge to put a pistol in my mouth.)

The upshot? Instead of browsing an endless list for Red Flags that have been planted by who knows who, for who knows what point of law, you’re looking a discrete list of binding opinions that cite your case for your individualized reason. Along the way, you pick up all the good news about your case that outmoded Red Flags invite you to gloss over.

This is the difference between a generic solution that plays into the hands of vendors who want to lead you into purchasing items of information that are outside your contract and a powerful, individualized solution that ends information overload by producing a discrete Hotlist of opinions that you wouldn’t want to miss!

ANALYTICS

By default, Westlaw results are ranked Newest to Oldest. A lot of commentators are flipping out over the superior results they are obtaining on WestlawNext. If you are a Westlaw user, you can replicate these results by drilling down in your Settings and changing the default ranking to Relevance. You will thank me.

For a lot of years, TheLaw.net poked fun at Westlaw for ranking results Newest to Oldest. TheLaw.net Equalizer has always provided default results by Relevance. We even assign percentages so you can assess the relevance of opinions in relation to each other. In determining relevance rankings, Citetrak, our best-of-breed, first-mover algorithm, considers search term proximity, density, diversity and numerosity.

TheLaw.net’s Results Report is dynamic. In a click you can alphabetize A-Z. In another click Z-A. Another click sorts Newest to Oldest. Another Oldest to Newest. Another click sorts by Citation Frequency Nationally. Another limits the sort to Citation Frequency within your results only. Another subsets results by Jurisdiction. Thereafter, you can sift and sort as above.

The upshot is that at-a-glance, you know that the best opinions are the ones that share high relevance and high citation frequency. And – ah – those are the ones you read.

As President of TheLaw.net Corporation I have access to vast information resources that are similar to what you might expect to find in a large law firm or government office. That said, one of the best tools for tracking the activity surrounding an opinion is Google Alerts.

I enter the citation and Alert’s robot emails me updates for free.

The judicial opinions you find in commercial databases do not come from Westlaw, as some industry gatekeepers might have you believe. They come from courts. They appear first on the websites of courts before they appear in commercial databases. Google indexes the websites, sometimes even before opinions appear in Westlaw, Lexis and TheLaw.net. In addition to black letter judicial opinions, Alerts emails me links to emerging Pepper related news and comment.

Alerts provides a way to robotically force your will on Google. Google is getting harder and harder to control. Programmers tweak the algorithm hundreds of times a week. The search engine has a mind of its own and that mind is focused on generating revenue for advertisers. If you search Google by citation you will receive diffuse links to the freeweb where it still looks like a law library with the books all over the floor.

An Alert is not a substitute for Validation or for comprehensively searching West, Lexis or TheLaw.net by citation and point of law. It does, however, keep me in the real-time reference loop of what’s happening with opinions I’m interested in tracking and delivers the perspective of commentators who are interested in the same thing I’m interested in.

Searching West and Lexis by citation only works if you have access to all 315 Federal and state jurisdictions. You can use KeyCite or Shepards, but forcing your individual will on KeyCite is impossible. Forcing it on Shepards is cumbersome.

TheLaw.net provides all subscribers with all 315 Federal and state jurisdictions for less than $50 a month. Subscribers easily search by citation and point of law to build their HOTLIST of relevant opinions in seconds.

ANCHOR + FILTER = HOTLIST

ANCHOR your query with the citation and FILTER by point of law: “130 S.Ct. 3499″ and resentencing and rehabilitation gives me opinions citing my case for the reason(s) I care about from the court(s) I care about.

Try using this process as an end-run around West’s Keycite and Lexis’ Shepards. You’ll be thrilled with the results. You won’t overlook opinions that have been red-flagged for a point of law that is unrelated to your search and you’ll learn the good news about cases that are buried in your citator results.

In other words, you’ll be the most informed lawyer in the room.

Here’s a portable search tip created in the support office of TheLaw.net Corporation. When your search is driven by statute, before you begin, print the statute and place the hard copy next to your keyboard.

I am continually stunned by lawyers and paralegals who commit the sin of word-guessing.

The web is atwitter with the news that WestlawNext now assists enables word-guessing. Enter “sexual battery” and it may ask: “Did you mean “sexual assault?” Perhaps one day it will ask: “Did you mean Mike Tyson?” Who knows? And, that’s the point.

Book research yielded pure results. People aren’t so sure about compusearching.

Humans word-guess. Robots crowdsource. And, that’s fine if you’re searching for Snowball The Dancing Cockatoo. It’s not fine when you’re client is facing several years in prison.

If your client is accused of aggravated assault and you guess “sexual battery” when the statute says “sexual assault”, you either didn’t read the statute or you’re working from a faulty memory. You need a simple, reliable process.

This is an example of when old technology – simply printing the statute and having it at your fingertips for reference - is the only solution that guarantees you will not miss relevant law. It guarantees your query is rooted in the plain language of the law, ensuring you retrieve all relevant law.

At the outset you need to know the size of the dataset you’re working with. How many opinions mention your statute? Usually there aren’t that many. Many statutes have never been cited. Even fewer have been cited by the court you care about for the reason you care about.

When searching by statute, first search using only the statute number. If your client is accused of committing a murder in Florida he/she is charged under 782.04. That’s your Anchor. Enter that first. You don’t search on murder because that leads to diffuse results, returning, for starters, both criminal and civil cases.

If you find too many opinions, filter by concept:

782.04 and carjacking

782.04 and “aggravated child abuse”

782.04 and “home-invasion robbery”

I didn’t invent the phrase home-invasion robbery, I cribbed it directly from the statute. I consistently query in the precise language of the controlling law to ensure I never miss relevant law. I want to see every opinion that expressly mentions my statute number, together with the general relevant circumstances. That’s my Hotlist.

Why rely on woefully incomplete annotations that were compiled when all we had were books or the happenstance of crowdsourcing?

When your search is driven by a statute or regulation, all you need is this simple process to find all relevant law, every time, guaranteed.