140 likes | 247 Views
Publication Etc. Disclaimer. This is a complex and emotional topic. There are many facets of the problem. Any “solution” will have good and bad points and be hard to implement. We will not design a solution today. Plan for talk. Some thoughts on the problem.
E N D
Disclaimer • This is a complex and emotional topic. • There are many facets of the problem. • Any “solution” will have good and bad points and be hard to implement. • We will not design a solution today.
Plan for talk • Some thoughts on the problem. • A straw-man “solution,” just to: • Be concrete. • Provide a target to attack for discussion. • We should focus much more on the problems that exist rather than this straw-man – even I won’t advocate implementing it without a lot of refinement. • Most of the session will be discussion (or lunch will be early).
What is the problem? • Too many people striving to write too many papers. • Among the results: • Conferences are either too large and diverse or ridiculously selective. • PCs are enormous, fickle, and have a hard time finding good programs. • Researchers face a depressing task of generating all these papers and getting them past program committees.
The coming deluge… • Scary scenario: • As more of the world develops research universities… • As more existing universities “fill up” with researchers… the problem will get (much) worse. • We are living in the “good old days” of 10 years hence.
What the problem is not • Young researchers don’t have the high morals and research taste of the old guard. • Young researchers have a genetic deficiency that leads them to prefer LPUs over high impact fundamental work.
Some observations on papers • Once-in-a-decade papers tend to appear only about once every 10 years. • Paradigm-change causing papers tend to appear only about as frequently as paradigm changes occur. • Asking everyone to only write once-a-decade, paradigm-changing papers is stupid.
So why do we have so many papers? • Two factors: • Many more researchers. • Many more papers per researcher. • These researchers are smart: • Publishing a lot of papers in good places guarantees [a moderate level of] success. • Going for lots of singles has a higher expected payoff than always swinging for the fences. So they write a lot of papers.
How can we address the problem? • Supply – reducing the supply of papers. • Demand – reducing the demand for these papers. • Capacity – creating scalable publishable systems that can handle the deluge.
Reducing supply • Make our field really unattractive so no one enters it? • Ask everyone to please write fewer papers? • Shame people by calling their work LPUs? • Establish and enforce quota systems? All are doomed; like fighting drugs by asking suppliers to stop.
Reducing demand • Thought experiment: if writing 5X papers gave no benefit over writing X, the problem would be lessened (perhaps by a factor of 5). • So how can we do this? • Tell everyone to stop counting papers in evaluations? • Many of us already do not count papers in evaluations, but unlikely to succeed worldwide anytime soon. • Guess: solution must be linked to capacity.
Straw-man idea • All papers go to one clearing house. Every paper is “accepted” there. • Conference PCs comb this clearing house for papers. • Each PC member picks a few papers to bring to the PC meeting (small, and face-to-face). • Each PC member is an advocate for his/her selections. • No obligation to review any paper in the clearing house. • Conferences compete for the best papers. • It will be brutal but life is hard.
Creates a two-tier system • Clearing house publishes everything. • Everyone gets the experience of performing research and writing it up. • The ecosystem continues to be fed with interested people dedicated to interesting projects. • The lucky few will get recognized with conference paper invitations. • Want to be a researcher? • Figure out how to get invitations. • Fail to get invitations? • Have great career in industry.
Discussion? • Your comment/idea goes here…