What About Eli?

One of the most important projects of my professional life has been to create, along with Jeff Grabill and a fantastic group of our colleagues, a software service called Eli Review. It started as a project in the Writing in Digital Environments (WIDE) research center at Michigan State University and, in 2012, it became a service available to others.

Eli Workflow: Write, Review, Revise…and Reflect!

The Business Side

Working with Michigan State University’s technology transfer office, we created a business — Drawbridge Inc. — that offers Eli to customers around the world. We had support from the MSU Foundation during our launch, including leadership assistance from a former Board member, Greg Bradbury, who served for many years as our CEO before his passing last year.

Jeff and I also created very strict and clear Conflict of Interest agreements with MSU. Mine is overseen and reviewed each year by the Dean(s) and Provosts’ office. Over time as my role at the university changed, the agreements have been updated to ensure that my (or Jeff’s) positions did not introduce new conflicts. Among the terms we agreed to is that we take no salary from Eli. We have an ownership stake but no direct compensation as employees. We also have never and will never employ MSU students to work in the company or on Eli projects.

The benefits to me for working on Eli are largely the same as those for any major research or publishing project. Though it wasn’t always the case, my home department’s bylaws today count software that has an educational or research purpose for tenure and promotion. And I have of course spoken and published a great deal about Eli and about research on learning and feedback, which also helps in my career advancement and merit reviews.

Apart from that, MSU is also an owner of Eli and the MSU Foundation has a seat on the Board alongside Jeff and Me. In fact, the company pays MSU every year a licensing fee to use the technology that Jeff and I created. Over the years, we have paid our employer hundreds of thousands of dollars to use the “invention” that we, ourselves, made. So we pay them. Not the other way around.

All of this might sound strange. And it was certainly new to us when we started, but much of it is standard in Tech Transfer at universities. It is perhaps rare in the Humanities but much more common in Engineering or Bio Tech or Agriculture. For my part, I have worked hard to create each piece of my COI agreement to ensure that on any given day my employer, my colleagues, and my students can trust that I am making Eli as an extension of my research and teaching work.

Because we have folks counting on us to make their classrooms a success, our business goals are to be sustainable and reliable as a service to the tens of thousands of students and teachers on the service every day during the academic year.

As a business there are a few things I’m proud of that have nonetheless been big challenges. First is that we are still going after ten years, many changes in technology and our team. Second is that we have no debt, and we have never taken infusions from venture capital. We run the business on our revenue stream. Our price to students is $20 for six months or a semester. A little less (but not half) for a quarter and a little more (but not double) for a year. At that rate, we have a very slim margin. And we’ve kept the price the same, except for one adjustment to make it lower, for ten years.

All of these things, combined with limits on what we will and will not do technologically have meant that we are an unattractive target for acquisition by players looking for a “hockey stick” growth curve. Not our thing. But for the same reasons, we are still here, still operating in the black with nobody but us making the important decisions about how our service is designed, engineered, and distributed.

I am proud to say that our vision for the product as a formative learning space for students and teachers remains our sole focus today.

The Service

Eli is a space for students to practice giving and using feedback with the guidance of teachers. The research that inspired us to create Eli — conducted by others and supported over the years by our own original work — consistently shows that when students practice giving feedback and practice using feedback to revise, they become much more confident and proficient writers. In 2008 when we looked around at our options, we did not have a good one for getting students to practice review and revision. So we made Eli to see if it would help. And it did.

Over the years, the original design choices we made for Eli have remained:

1. It facilitates formative feedback only. No summative feedback at all. No grading. No gradebook. No means to export data from Eli to the LMS gradebook. Just feedback for writers from writers.
2. No peer grading. Some call this “calibrated peer review.” You might remember it from school as passing papers to the left and having another student mark them as the teacher reads the answers? This outsources grading to students. WE DON’T DO THAT. There are many products that do. You can look them up. But just to reassure you if you’ve never seen Eli work: our review model does not allow for one-to-many reviews (all students calibrating on a sample text). Nor does it impose a standard rubric (teachers make their own). The reasons we don’t do those things are simple: the research shows they are not drivers of student learning. Peer grading is not peer learning. It may drive down the cost of instruction (for big sections), but its benefit for learners is questionable at best. So we are uninterested in it.
3. Students own the writing they do in Eli and teachers own theirs. We do not claim copyright or ownership of students’ work. We ask their permission to show it to other students in the class(es) they are in, and that’s it. Teachers own their assignments. We ask permission to show them to their students and that’s it. Apart from that, if we want to do a study or have a workshop where we use samples from folks, we ask permission. Just like good colleagues should.
4. We take great care, too, with students’ data. We collect some, all for formative feedback purposes, and we show it to students and teachers. We make all of the data we can see about an instructors classroom available to the teacher in a variety of ways, including on screen in tabular form and in visualizations as well as in spreadsheets for download. There is no trove of data we have that our colleagues and partners do not have. Students can see their results too, of course, and download their materials at any time for their own portfolio.
5. Over the years, our core service has changed but not dramatically and in some cases not as much as we would like. We still don’t have some useful features like notifications, for instance. This is because most of our limited engineering and development resources have gone into making sure our service is reliable, accessible, secure, and safe. We have users all over the globe now and a 24/7/365 customer base. On peak days we’ll see 20,000 or more people using Eli at the same time. We integrate with all the major Learning Management Systems on campuses that have institutional agreements to offer Eli and this means we have passed hundreds of rigorous security reviews with IT departments, and we meet high standards for accessibility and data privacy for states like California and also the EU. We also have automatic failover and disaster recovery technology, instant backup with redundant hosting onshore in the United States (on both coasts), and a staffed customer support help line that ensures folks get prompt responses if they have a problem.

The Goal

When I think about what we have done with Eli, I am proud of what we have done. But I am just as proud of what we have NOT done. We have not caved to the many people who advised us to monetize student data or to add plagiarism detection or to build in calibrated peer grading. Other people have made money doing those things.

We also have not pursued features that threaten the work or expertise of teachers or that make students’ data a product. This is not because our research and tech chops do not include machine learning or AI. I study that stuff too. But we have taken great care to engineer Eli as a learning studio, not as a robot learning factory.

And the reasons why are not due to an inflated sense of self-righteousness. This is not, as some have claimed, a “purity test.” But it is a character test. If we say we are making technology that helps students learn, we want to do that. Helping robots learn is another matter.

All the things we haven’t done over the years come down to that one idea: we don’t think that any of those things help students learn. That is our only goal. Being self-funded and independent as a business has allowed us to stay with the goal.

Well before I ever imagined starting a business (it was NOT on my to do list) something like creating Eli has been a personal goal for my whole career. I have made the case for doing this — embedding our disciplinary expertise and knowledge about rhetoric and writing in software and systems — since my undergraduate days and it is a defining theme in my career.

For having this goal and being unashamed to relentlessly pursue it, I have gotten sneers from both academics and business people. “It’s like selling broccoli when you could sell pizza” one business bro said to me. I mean I like broccoli?

Academics are more complicated. Assuming they will consider a piece of software from an academic worth talking about at all, they go through a series of reactions. Let me walk you through how it goes

• “you’re an academic not a maker…” — we make
• “ok, but it can’t work the way you say…” — it works
• “ok but it won’t last” — it lasts
• “well you are doing X (something we aren’t doing)” — we aren’t, we don’t

It doesn’t bother me too much now, because this, too, is feedback. And in the Ed Tech landscape, plenty of skepticism is warranted! Bring your critical eye. But also look and read carefully. You might see something unusual, something that surprises you. A research project, a bit of theory and pedagogy made into software, that — in this procedural format — can help people learn. That’s what we’re shooting for.

Enough people have seen and used Eli in the last decade or more, and enough colleagues have studied and reviewed and published about it, that nobody has to take my/our word for any of it anymore. So you can google and ask them too.

Find a teacher or student who has used it and ask: “so… what about Eli?”