BLOG@CACM

Enjoying Reading Research

By Greg Linden
May 8, 2009
Comments

Research papers take a long time to read.  They are dense, narrowly focused, often seem abstract and detached from practical issues, and occasionally require much knowledge of prior work to grasp.

Given all that work, why bother?  After all, as many of my colleagues in industry say, due to the many assumptions about the quality of the data, needs of the users, performance of the algorithms, or size of the data, academic research often is unusable to them as is.

What I find most valuable about research work is that someone smart has spent a long time thinking about a particular problem.  Someone has spent much effort describing a problem, why it is important, what has been tried in the past, and what should be tried. 

The authors are working to advance the state of the art, but the solution often is less valuable than the journey.  For those that are trying to solve similar problems, it is the discussion of the paths taken and not taken that illuminates the road.

If you also believe this, then the way you read papers might change.  Years ago, I used to turn first to the implementation and experimental results, then push the paper away if I found the evaluation lacking. 

Nowadays, I turn first to the introduction, related work, conclusion, and future work.  I seek to understand the problem, why it is important, what has been tried, and what still needs to be tried.  I try to see why the authors chose to spend part of their lives pursuing solutions to this task and what insights they gained.  I think about how I would solve the problem myself.  And only then do I turn to their solution.

Read this way, it is much easier to bask yourself in the flow of academic publications, letting the thoughts and insights wash over you.  The papers become an easy joy to read, like having a conversation over coffee with the authors.  It becomes what research should be, the sharing of ideas.

---

No entries found