Anatomy of a botched USNWR law ranking leak

For the past few years, USNWR has emailed all law schools deans an embargoed PDF listing the tentative law school rankings about a week before their formal release. And for the past few years, within minutes (and in disregard of that embargo), that email is leaked to a private consulting company, which then posts the rankings on its corporate blog, where the rankings then spread via social media and gossip sites.

This year, USWNR did something different. It released most of its graduate school rankings in an Excel spreadsheet on a password-protected site around 8 am ET on Tuesday, March 5. But it did not release the law school full time rankings, nor the business school rankings. (I guess we know which schools are beholden to these rankings and where USWNR sees its value!) 

Instead, shortly after, individuals at schools received their own school's ranking, and nothing more. This makes leaking much more challenging. If you leak your own school's ranking, it's obvious you leaked it, and USNWR may punish you by not giving you access to that embargoed data early next year. 

But around 5 pm ET on Tuesday, March 5, USNWR sent out a new update. Its Academic Insights database would now have the 2020 rankings data (that is, the rankings data to be released March 12, 2019). 

Academic Insights is a USNWR platform that law schools purchase a license to access and use. It has rankings data stretching back to the beginning. It offers multiple ways to view the data inside AI, or to pull the rankings data out of the database. 

It's user friendly, but it isn't always the easiest to operate, and like many web databases it can suffer from some wonky behavior. It makes leaking a trickier proposition.

Around 7 pm ET March 5, the private consulting company posted the rankings. But the rankings made it very obvious that there were errors, and it also provided clues about how those errors came about.

To leak this information to someone, some law school administrator made a data request from the database and exported the rankings information to a CSV file. The “Leaderboard” AI database is a swift way to see the ranking of law schools compared to one another across categories. (Recall that the database stretches back through the history of USNWR, so it includes all schools that were ever ranked over the last 30 years, whether or not they’re ranked, or even exist, this year.)

The list then included as “N/A” (i.e., “unranked” this year) schools like Arizona Summit and the University of Puerto Rico. This is unsurprising because USNWR doesn’t rank (1) provisionally-accredited schools, (2) schools under probation, and (3) the schools in Puerto Rico.

But the leaked ranking included other bizarre “unranked” choices: Hamline University; Pennsylvania State University (Dickinson) pre-2017; Rutgers, The State University of New Jersey--Camden; Rutgers, The State University of New Jersey--Newark; Widener University; and William Mitchell College of Law (among others). These schools all no longer exist (along with a couple of others that have announced closures). Why list them as “unranked”?

Separately, the leaked rankings omitted information for Penn State - University Park, Penn State - Dickinson, Rutgers, Widener University (Commonwealth), Widener University Delaware, and Mitchell|Hamline. Why aren’t these schools in the database?

These are obviously not random database omissions. They're omissions of schools that split or merged. Their old schools are in the database. But the Leaderboard database pull request omitted those schools. (Why, I don't know.)

But there are ways of requesting school-specific data. You could request the specific institutional data in the AI database for, say, Penn State - University Park or Rutgers, and the data is now available for your review—including those institutions’ ranks. Of course, a few schools might ultimately be "rank not published," or "Tier 2" schools in the rankings. But they're not "unranked."

(Incidentally, from the revealed metadata, we know a lot of information about which person at which law school leaked the rankings, but that’s not what this blog post is about.)

The real botching came when the leaked ranking included these strange inclusions and omissions (with some noticeable gaps—think two schools listed at 64, followed by a school ranked at 67, which means there’s an omission in the 64 ranking) was posted and began to spread. Panicked students and prospective students at places like Penn State and Rutgers asked what happened. The private consulting company replied that it “appeared” the schools were “unranked.” That spawned a great deal of speculation and worry on behalf of these students.

Of course, that wasn’t the case. The statements speculating that these schools appeared to be “unranked” were reckless—that is, they were based without an understanding of how the database operates and based instead on speculation—and false—because, as I noted, each of these omitted schools had a ranking in the database, simply not in the CSV leaked to this private consulting company. (Later statements began to concede that these schools would probably be ranked, but those statements came only after worry and misinformation had spread.)

I pushed back against this false news last week in a couple of social media outlets, because it does no good to perpetuate false rumors about these law schools. These law schools, I insisted, would probably be ranked. They were ranked at the very moment in the AI database; and, barring a change, they’ll be ranked when the rankings were released (i.e., now). (Of course, some schools, like those under probation or those in Puerto Rico, were never going to be ranked.) 

The backlash I received on social media was impressive. I confess, I'm not sure why so many prospective law students felt threatened by my insistence that someone had disclosed bad information about schools like Penn State and Rutgers to them! (Happily, such comments roll off easily.) After that, apparently, USNWR asked for those rankings to be taken down, and they were. (Of course, they still floated around social media and gossip sites.)

But we know that leaking USNWR information from the AI database presents complications for future leaks. Failure to understand how to operate the database may leave an incomplete and inaccurate picture, as occurred this year with the botched leak. We’ll see what USNWR does for the 2021 rankings—are total but accurate leaks better, or incomplete but inaccurate leaks better? We shall see.

And for those relying on leaks in the future? Read skeptically. The leak included material errors this year, and I wouldn't be surprised to see material errors in future leaks.