The other day I wrote a post imploring academic computer scientists to at least occasionally break out of their research bubbles and engage with real engineering problems where their research matters. That post was, admittedly, a bit facile about how one might make this engagement happen. This piece suggests some ways. I don’t claim any particular authority or expertise, though I have — except where noted — tried out all of the suggestions below. Perhaps others will chime in with what has worked for them.
Talk to a lot of people and create a network. Networking is a really important skill. Back in the day Phil Agre’s Networking on the Network was what students would read. It predates modern social networks but contains a lot of timeless advice and in any case I don’t know of a better choice.
Attend conferences outside the academic mainstream. In some cases, such as SIGGRAPH and Supercomputing, there’s plenty of non-academic content at the conference you’re attending anyhow, but most of us aren’t so lucky. There’s a really wide range of industrial conferences; some of them definitely won’t be very interesting for academics so you need to do some homework ahead of time to find the one where the right people will be. For my current work, the LLVM Developers Meeting is the main event, almost all of the community’s heavy hitters are there and the technical sessions are amazing. The security community has plenty of great non-academic events. I’ve heard good things about !!Con and Strange Loop. In any case, the point of attending these conferences (besides the technical content) is to meet people who aren’t professors or students.
Spend a day visiting a company or a government agency. You need an invitation, but you can invite yourself if you can make a contact for example via your advisor, a mutual friend, or by meeting people at conferences. Talk to people there, get a sense of what they’re doing with their days, what they’re worried about, what they’re frustrated with. Give a talk about your work if possible, but this isn’t necessary. It often makes sense to do these visits when you’re already traveling.
Spend longer visiting a company or government agency. Depending on your career stage this could be an internship, a postdoc, a sabbatical, a summer, or a leave of absence. This is a chance to work closely with people for an extended period of time. A lot of people do this at some point in their careers and I think it’s a really good idea.
Engage on twitter. It’s a weird, crowded, noisy place and it can take a while to find the right people to follow and longer to get them to follow you. The advantage of Twitter is that there’s a huge number of smart people are out there and communicating with them is almost frictionless.
Blog. People are far more likely to read a blog entry than a paper, in my experience. Also, the readership is different, because non-academics are even less likely to read a paper than academics are. Realistically, starting a blog only makes sense if you have a fairly consistent stream of things to say that don’t fit into tweets and don’t really belong in academic papers. Building an audience takes time and requires a certain amount of regularity in writing; these don’t necessarily fit in very well with the academic binge model of working that many of us subscribe to. Another issue is that blogging doesn’t pay the bills, academically speaking — you should only do it because you want to, not because you expect any direct benefit to your career. I waited until getting tenure to start a blog for this reason, and also to make sure that I had at least a few years’ worth of ideas lined up.
Find people who are great at external engagement. Emulate them, collaborate with them, or join them. The Racket folks are amazing at this.
Release software. Put your stuff on Github, polish it up, and then tell people about it. Get users, accept pull requests, respond to feedback, fix bugs, add features, cut releases, and repeat. Either your code will provide people with a good value proposition or it won’t — either way you learn something. The caveats are that building a user base takes time, creating realistically usable software is like 25 times as much work as creating research-grade crapware, and only a small subset of computer science professors will value your contributions in this area. But it is enormously fun and anyway you don’t want to make the mistake of caring too much what professors think.
Engage with existing open source software. For many of us, there’s an obvious open source project that we could be contributing to or otherwise helping out. Find that project and read their mailing lists, look into the issue tracker, build and use the code, read the code, and maybe submit a patch or two. Beyond these things, consider attending their meetings or BoF sessions, if these exist. A reasonable long-term goal would be to use your work to make a significant improvement to the open source program.
Start a company. This one I haven’t done, though I know many people who have. It is a somewhat extreme option, as much a lifestyle choice as research engagement strategy. Details are out of scope of this post and anyway I don’t know anything about doing this.
Ok, with all that said, I’ll make a prediction or two about what will happen if you follow these suggestions. First, you’ll often find it frustrating and some of the time you invest will be wasted. I’ve burned months on things that never bore the tiniest fruit; if I knew how to tell you to avoid this, I certainly would. Second, you’ll discover that the problems that people are having out there aren’t the ones that you would have predicted, nor are they the ones that your CS friends and colleagues predicted. You need to learn to listen to people, but often even the people having problems aren’t actually having the problems that they think they’re having (anyone who has worked tech support will tell you this is the case more often than not). You need to learn to observe carefully and read between the lines to figure out what’s really going on. Third, at some point you will run into the distinction between problem-driven research and solution-driven research. The former is like trying to cure cancer or put a person on Mars: the problem is everything and you’ll consider any solution that might work. The latter is where your research hammer is a damn good one and you’re never going to put it down: if it can’t solve someone’s problem, you’ll move on and find a different problem. Obviously there’s a spectrum — but you’ll need to decide where on it you sit.