Does Anyone in IT Read Academic Literature?

digital book worm asks: "I'm a soon-to-be fresh-out with a Masters degree in Computer Science. For the last several years I've had paper after paper shoved in my face, and following some of the major publications (such as Communications of the ACM, IEEE Software, and some of the IEEE Transactions series) and conference proceedings has become second nature. I can't help but wonder though: For those of you outside of academia, do you find that there is any benefit to following the latest research on a regular basis? Clearly there are times when it is appropriate to go back and look for papers that solve a particular problem you're working on, but I'm finding it a little difficult to believe my professors' hype that the latest and greatest is the stuff that goes on in these papers, and that I should make it a life-long goal to keep current on many of the publications listed above."

Comments

Digests

[Pentagram]

There is a great deal of dross that is published in even the better scientific journals and it is quite a task keeping up with even a small field. I can't imagine many people who are not directly involved in research would be able to find the time to keep abreast.

I'd like to see better summaries of research published; something available in between reading all the abstracts and interesting papers in the top journals of the field and just reading the occasional flasgship paper in /Science/ or /Nature/.

Re:Digests

[AtrN]

I find the ACM's Computing Reviews to very useful.

Yes.

[Roadkills-R-Us]

When I was in software design (with some hardware work), I followed several academic journals and a plethora of trade papers to keep up.

Now, after a career veer into network and systems administration, I haven't run across much worth reading on a regular basis. It may be there, but I haven't found it.

Even with the "dross", and the stuff I didn't care about, it was worthwhile when I was in software and hardware.

They should!

[ponos]

There's a saying:

A few months in the laboratory (or coding, in your case) can often save you a few hours in in the library.

The hard part is actually being able to keep up (i.e. knowing the language, which journals to trust, what's going on etc). Since you are following the literature closely you should not abandon it. Those that don't follow it will definitely have a hard time starting, though.

P.

go with the flow?

[macshit]

I certainly don't religiously keep up with things, but I love periodically going to the library and checking out the latest journals etc. because there's always something cool to read.

I do seem to find the most interesting stuff through references from other sources though, e.g., someone will mention a paper on a mailing list I read, or I'll do a google search on some term I see in patch which will turn up relevant publications.

So my impression is that you can keep abreast of things pretty well simply by "remaining engaged" in whatever activity/community interests you, and that the current interesting ideas and research will inevitably pop up if you do. Reading journals and conference proceedings (or at least browsing through the contents to find the cool stuff) can be part of that, but it hardly seems necessary to worry about it too much.

There's some merit to either viewpoint.

[mellon]

On the one hand, if you stay on top of research, even though most of what you read is useless, you'll be in a good position to integrate what you read and it will probably kick off thinking processes that will be of value. Having a good overview of what's going on helps you to see connections between work that the authors may not even realize is related.

On the other hand, there's a tremendous amount of stuff out there. I don't mean crap - some of it's quite good. But staying on top of it could easily be a full-time job, and then some.

So I think your professor is right, in a sense, but what he's saying may not be practical in the real world, even if it makes sense in academia.

I would suggest that you decide how much time you're willing to spend reading papers, and make a habit of doing it, and when you run out of time, stop. If you have nothing that's obviously interesting in your stack, skim what's on top. If there's something interesting, skim that, and if you still think it's interesting, read it more carefully (if you need to - often skimming is all you need). But always cut yourself off when you get to the end of the allocated time - don't bog down reading papers all day.

Heck, this is advice I should probably take. Thanks for asking! ;')

Re:No

[ivan256]

I have exactly the opposite perspective. Every successful, novel product I've worked on has had foundations in recent academic papers. Successful design level engineers read the lastet stuff, and put the ideas they get from the papers they read into practice.

Anybody that tells you they can launch a tech product that is both revolutionary and successful without learning from the recent innovation of others is lying, and the people in academia are the ones that share their recent innovations.

No

[Ratbert42]

Part of the reason I bailed on my PhD was that one of my profs was an editor for one of the big journals. I probably reviewed a couple dozen papers a year and saw how tit-for-tat things could be. An unknown with a good paper could still get published but a incrementally improvement by one of the big research schools was always rated higher than it should have been. The papers aren't written for most of the audience anyways; they're written to impress reviewers. So that's one reason I don't read them.

The other is that they just aren't that relevant to my career. I once heard cutting-edge software engineering literature called "intellectual porn". Just like porn, all the little imperfections are smoothed over and if you read enough of it you believe that every company (or girl) is perfect and yours is the only one that's this messed up. Talk to guys that have been around (jobs or girls) and they'll tell you that they're all messed up. The ones that look perfect are the worst of all.

Personally I try to keep up with trends that have caught on. I didn't jump on Agile or Extreme Programming even though parts of it sounded good. Over the last several years I've learned Java, Photoshop, ASP, Perl, JSP, Flash, UML, Design Patterns, J2EE, and now DotNet. Not once did I pick up a journal article for those, but just looking at job postings, resumes, and the shelves at mega bookstores will show you what's hot. Sure you'll be a little behind the curve. If you want to stay in front of it, get to DotNet or Java user groups and watch the respective web sites to keep up with the up-and-coming features.

Current stuff? How about CLASSIC stuff?

[xtal]

You're kidding, right?

I've been able to do very well for myself by being able to apply classic engineering and IT things - you know, searches that aren't linear brute force, some of the more useful design patterns, knowing what a code profiler is. How many programmers out there have even tried to read the Art of Computer Programming, by the man your God, Knuth?

How many people who call themselves C and C++ programmers do you know, who haven't read the C Programming Language, or the C++ Programming Language? ..for that matter, how many have studied the documentation that came with their compiler? understand the operation at a general level of the OS they work with? have a knowledge of security problems? care about any of that? ..becoming more cynical, know what the stack and heap are?

My experience is that a large segment of the IT world is ignorant or dismissive of the classical stuff, so to hope for them to be aware of current developments that may or may not be useful to anything being done now is probably a far reach at best.