Guest Post from Shriram Krishnamurthi: Growing respect for Research around Computational Learning and Thinking
Shriram and I had an email correspondence around the blog posts aboutrenaming the field and gaining respect for the study of how people learn and think about computation. He suggested a path forward that was about re-connecting to the fields that the CSEd community broke away from. I invited him to prepare a guest post that conveyed these ideas. Thanks to him for this!
Let me suggest you are probably trying to achieve two very different things here.
1. Create an actual community. There is real value to having all the interesting people from one area in one room. (This is why, despite the trouble it is to get there and back, I almost never say no to a Dagstuhl invitation.)
2. Have your students publish in venues such that, when they go out onto the job market, research universities such as yours (Georgia Tech) and mine (Brown) will notice and respect them, interview them, and make them tenure-track offers so they can have students of their own.
Unfortunately, I believe that right now these are fundamentally conflicting goals. SIGCSE, ITiCSE, and ICER address the former but not really the latter.
One, unrealistic, option is for 1 and 2 to merge. For this to happen, these venues need to become a whole lot better. I hear great things about the structure of ICER, but some of the papers are great while others are at best so-so. Changing the other two is harder than turning around an aircraft carrier. It may be possible to make ICER a stronger conference, but one small conference cannot really a whole area make. Plus, you still need to convince people to pay attention to it.
The only other option I see is to do both. Attend whatever conferences you need to to form a community. But get your students to publish at really good venues outside the area. That way, they can write a gung-ho application: “Look, I’m perfectly capable of holding my own in the open competition of conferences you respect”. People like Andy Ko—who published his work in conferences like ICSE—or my colleague Jeff Huang—an HCI person with strong publications in information retrieval—are examplars whose technical chops can’t be questioned.
In other words, this is a long response that could be abbreviated to “Yes, you should grow CSEd by sending it to more respected venues”, but I’m also showing you some of my thinking (because a good teacher grades the work, not just the answer!). A student who publishes a few papers in some conference already recognized as respectable technical CS is going to stand a far better chance. Once a dozen of those populate good departments and start producing students of their own, you’ve pretty much gotten over any prejudice and can then reset your standards. (Though I would still say it’s unhealthy to drop ties to these other areas and retreat into a CSEd shell.)
Which conferences, of course, depend on the student. For students doing HCI work, it might be SIGCHI; for those doing software engineering, it might be ICSE; for machine learning, ICML; for information retrieval, SIGIR; and so on. One good bit of advice to a young CSEd PhD student might be, “Find another area of CS in which you can demonstrate enough depth to publish papers in its good conferences and be able to hold your own in conversations with an expert in that field”.
Here are three other things to consider.
1. Being able to hold one’s own in another field creates natural allies in a department. A non-CSEd faculty member who realizes there will not be hires in their own area is likely to become an advocate for a CSEd candidate who has at least some presence in their area.
2. I feel the CSEd community has let itself be put into the “liberal arts ghetto” or, at the research university level, “instructor ghetto”. The leaders of “research” are tenured professors, but the leaders of “education” are Instructors, Professors of Practice, and so forth. This is a self-perpetuating cycle. For instance, who is the CSEd applicant going to get letters from? Getting a letter from an Instructor naturally makes the tenured faculty think, “Hmm, why should we take this person seriously?”
3. Finally, candidates need to be able to demonstrate a growth path. When I look at a candidate we’ve decided to interview, I’m only so interested in what they did before: their past achievements got them their interview, so now I’m interested in what lies ahead. I care to see what kind of agenda they have mapped out—is it interesting, is it hard, could someone else do it, etc.—and what skills they bring to the table (can they do it, and can they do it better than others).
I imagine this step is hard for some CSEd candidates. If you got a PhD studying some population, it may or may not be interesting to keep studying that population or to study the next such population or whatever. At the very least, then, if you intervened, showed an N% improvement, and have good plans to get to much more, and then show a path to bigger and more interesting problems, now you’ve got my interest. Put differently, think in terms of active interventions that demonstrate impact. Now you become comparable to students who are building or verifying software, deriving inferences from datasets, and so on. I don’t know whether CSEd students are getting advice in terms of presenting themselves this way.