That’s No Plagiarism Checker

I finally went on spring break as in “I actually got away from work and stuff”.  We took our first cruise.  For a social and institutional economist with a critical bent that just loves to observe people and capitalism in the wild, let’s just say that a cruise offers a target rich environment.  I’ll have more on that in some other post yet to be written.  But what I did want to comment on was the news last week in the edtech sector.

Last week Advance  acquired Turnitin, the notorious stealer of student intellectual property doing business under the guise of offering “plagiarism checking” services. Turnitin also has some other related businesses such an auto-grading service, etc.  The price was apparently $1.75 billion dollars.  That’s billion with a b.

I’m not here to talk about how awful the pedagogy of mistrusting all students is or how it’s immoral and unethical to steal/coerce student’s copyrights away from them.  These are all horrid aspects of Turnitin and among the reasons why I’ve always, for over a decade, opposed and fought against use of it at my school.  I’m not here to talk about those aspects because people much more knowledgeable than I have been saying that a lot – people like Jesse Stommel and Sean Michael Morris.  And the sale announcement last week has brought a lot of faculty and teacher anger out in public about the sale.

What I want to talk about is the numbers on this deal (I am an economist, after all).  $1.75 billion for this company. That’s a pretty hefty valuation. Especially for a company that really isn’t that big and hasn’t been a huge growth tear.  Yes, it’s been growing and the core product/service probably has a lot of room to expand internationally. But from what I can tell on the Interwebs, Turnitin probably has annual revenue in the $127.7 million range. That’s million with a m.  1.75 B to buy an annual 127.7 M. That means Advance is paying approximately 14 times annual revenue to buy it.  Valuation of a company as a multiple of the annual revenue is common way in finance of comparing whether a deal is highly valued or cheaply valued,  especially for technology companies and startups.  Turnitin isn’t exactly a startup – it was founded circa 1998, but it’s still a “tech” company.  For a tech company 14 x revenue isn’t out of range, but it’s not cheap either. Tech companies, especially ones that are expected to grow fast don’t usually have strong current earnings (profits), so revenues times a multiplier is used to estimate value.  So what this valuation tells us is that Advance expects Turnitin to produce some very significant growth – probably much faster than Turnitin has achieved so far in it’s 20 year life.

And that’s what scares me. Advance isn’t really just a “family-owned company” as some reports have it.  Yes, the ownership of Advance is private and dominated by descendants of Newhouse family (think newspaper publishers). But “family-owned” sounds warm and fuzzy like the diner down the street where you get breakfast.  Advance is a serious technology, publishing, and communications conglomerate. And they’ve got ambitions. And they’re serious with their money.  They think like venture investors. If they invest $1.75 billion, they’ll expect to turn it into $100 billion or more. That’s the game.

Even if they had modest ambitions and only wanted to turn Turnitin into (read those last three words again just for fun)  a modest 10 or 20 billion dollar company, they have to do something big and different. Turnitin isn’t getting to that range on it’s own by doing what they currently do: call students cheats and check for plagiarism.

So how does Advance expect to get it’s money back multiple times?  I don’t know. They didn’t share their thoughts with me.  But they did share them The Chronicle:

Chris Caren, chief executive of Turnitin, said the company’s next step is to become a platform for colleges and high schools to submit all types of student assignments, digital or on paper. It would then use AI to help instructors review that work to, among other things, spot at-risk students and devise remediation plans. The company is also developing Turnitin’s software to branch out into the STEM fields and detect plagiarism in coding, for example. In other words, it hopes to become a one-stop shop for all sorts of tech-driven teaching services.

Advance, which owns companies like Condé Nast, has recently begun investing in data and analytics companies, said Janine Shelffo, Advance’s chief strategy and development officer. Turnitin’s strong market presence and its advanced technologies, said Shelffo, make Turnitin a valuable investment. “There’s a whole road map where we can see where tech innovation will increasingly power personalized learning and enhance outcomes for all students.”

It’s time we connect the dots, folks.  Advance isn’t just a newspaper publisher. They’re  adept at cookies, tracking of readers on the web, and data collection just as Facebook and Google are.  They publish online magazines (Conde Nast, hello?). They publish Reddit and Arstechnica and other sites. They also have very capable big data analysis capabilities (1010data) that “transforms Big Data into smart insights to create the High-Definition Enterprise that can anticipate and respond to change” for 850 large companies.

In the Chronicle article, Sean Michael Morris rightly observes that Advance/Turnitin could develop profiles of students using their data and monetize that via marketing and advertising.  That’s true but I think we’re missing the mark. We’re being distracted by the fact the big two of early surveillance capitalism, Facebook and Google, have monetized their vast troves of surveillance data by using it to sell advertising. Rule of thumb in business strategy:  there’s only room for 2-3 big monsters in any particular big industry.  I don’t think Advance/Turnitin will go that way.

I think it will be worse. They’re already pushing automated-grading systems and student “feedback” systems.  There’s no technological difference between a system that checks a student’s written posting or submission for plagiarism against a database of collected writings and a system that checks those same posts/submissions against a database of “approved” thoughts and phrases. In a way, isn’t that what grading is anyway? Advance/Turnitin can easily morph into the thought police.

But can thought policing be monetized?  You bet. First, any nation that thinks a social credit system for controlling ordinary behaviors like jay-walking is a good idea, will love the idea of policing thoughts and utterances. That will be worth a pretty penny. Let’s suppose that social credit would never take off in the US or Western Europe (an assumption I am loath to make).  Many, many institutions of higher education will jump at such a system.  It’s just an extension of grading – until the database of vetoed utterances, ideas, or word strings is expanded to include controversial ideas.  Are you going to tell me there aren’t some religious based schools that would buy a system that automatically rejects papers or forum postings that suggest abortion is acceptable? Of course, the system won’t just “reject”, it will provide “feedback”.  Those controversial ideas can be handled automatically and merged with the “inclusive” repository of acceptable learning materials (free!) which is also, ironically, the exclusive source of learning materials.

picture of Star Wars deathstar with caption "that's no plagiarism checker, that's a teaching death star"For profit colleges will love the systems since they’ ll allow further cutbacks on faculty. Replacing labor with capital investment is one the oldest tricks in the capitalist toolchest. And, those colleges will get automated stats and data “proving” their students learned!  Not only were the students’ papers “correct” but everything they’ve said on Reddit and other social forums since has conformed to the acceptable.

Forget social credit. We’ve got the possible (probable?) platform for thought control scoring.

By all means let’s complain, scream, and object to the abuses of the Turnitin plagiarism model. But let’s keep our eyes open for the next big data thing.  That’s no plagiarism checker.  That’s the teaching death star.