A Walk Down Memory Lane
A million years ago when I went to school, capstone projects didn’t exist. Students looking for real-world experience had to volunteer, find internships, and/or obtain part-time work on their own.
I should know. I remember the not-so-good old days.
During my 1995/6 winter break at Cornell, I drove one hour at 7 a.m. in 15-degree weather every morning for free just so I could put a my three-week NYSEG internship on my resume. (I didn’t even receive gas money for my efforts.)
Times have certainly changed. For a while now, capstone projects have allowed students to gain valuable real-world experience over the course of a semester. Typically the students work for credit, not money.
Two years into my professor career, I have learned a few things about teaching these courses. Beyond my own experience, I’ve spoken with plenty of my colleagues at other institutions to effectively compare notes. It turns out that there’s more than one way to skin a cat.
Let Students Select Their Own Capstone Projects
Quite a few companies hold online contests to solve data- and tech-oriented problems. Two of my favorites are TopCoder and Kaggle (now part of Google).
Students benefit from being able to select projects that more closely align with their skills and long-term interests. Beyond this benefit, this method is easiest for the professor. In this scenario, the professor may be the de facto sponsor or client. For the most part, no one needs to sign a non-disclosure agreement (NDA).
The deadlines or a project probably don’t coincide with the professor’s and school’s deadlines. What’s more, the potential for cheating is significant. If 100 groups are working on the same Kaggle contest, it doesn’t take a rocket surgeon to figure out how to game the system.
Next, to the extent that these projects are often remote, in-person contact between sponsor and students is limited if not impossible. Finally, the students might not know what they can handle over the course of the semester. They may over- or under-shoot.
Assign Students Simulated Capstone Projects
This is related to the prior method, with one key difference: the professor finds or creates the assignments and then assigns them.
The professor can limit the projects to those that s/he consider meaningful. NDAs cease to be an issue here because these projects are, for want of a better word, hypothetical. Hypothetical or not, the fact that no one else is working on the project minimizes the opportunity for students to cheat. Finally, students can easily ask the sponsor questions because the professor is the sponsor.
The deadline issue mentioned above still exists. Students won’t receive job offers or recommendations from third-party sponsors.
Find Real-World Capstone Projects
This is the approach that I have chosen to take. As you’ll see below, there are both high costs and high benefits.
This is by far the most valuable approach for students for a number of reasons. First, they get to help an actual organization solving an actual problem. Second, they increase the chances of receiving employment references as well as job offers. (No, neither is guaranteed, but it’s a nice carrot from the student’s perspective.) Next, the potential for cheating is minimal. It’s not as if 20 other teams are trying to complete the same project.
Students can ask the sponsors questions, present iterations of their projects, and seek clarity.
Here’s another massive benefit: interactivity. Students can ask the sponsors questions, present iterations of their projects, and seek clarity.
I almost always speak with sponsors before they even fill out the form. Even a 15-minute conversation can allow sponsors to get to know me and vice versa. I can answer their questions—an important aspect since most have never been through this process before.
Finally and perhaps most important, the students learn about prosaic but formidable obstacles that they’ll face upon graduation. Here I’m talking about client availability, unclear business requirements, legacy technology, and impure data.
There tends to be a positive relationship between costs and benefits—and the same is true here. For starters, this approach means that professors need to constantly network and find projects for their students. To this end, I send out a link to this post several times each semester.
Dealing with attorneys and NDAs represents a considerable source of friction for professors and students alike. I understand the need for relative secrecy on certain projects, but injecting one month of back-and-forth never made anyone’s life easier.
Of course, professors can always employ hybrid approaches. That is, they can find real-world projects but let students discover and work on their own if desired. For instance, if they really want to work with a particular technology or programming language, they can pitch their own ideas.
Simon Says: The juice is worth the squeeze.
All things being equal, I have found that students tend to benefit more from real-world projects than simulated ones. It’s more work on my end, but my students ultimately get more out of the experience.
What say you?