22 October 2012

High information vs. low information science

“Cluttering up the literature.”

This is the sort of complaint that I see from time to time. And when I saw it, I never got why people got upset about it. There was something about it that always seemed snooty to me.

It occurred to me that I never quite got the ire because in my scientific research, I am operating in a low information environment. I’m confident that I have surveyed almost all the relevant literature for some of the species I work with. There’s not a huge number of people in the field. Every new paper gives something new to work with. And the rate of directly relevant new papers is measured in a few per year.

A lot of people, however, are working in high information environments. Forget about knowing all there is about a single species; so much is known that people have problems knowing about one small aspect of one species. Relevant papers probably come out weekly.

Intellectually, I knew are more active than others. But I don’t think I appreciated how much that affects how people view the “problems” of the scientific literature. Scientist in high information environments desperately want filters. They want glamour mags to tell them what’s important. Scientists in low information environments want more. They want to know why is nobody researching what to them are completely obvious questions.

Related posts

I’m a pebble in the avalanche

1 comment:

Bjoern Brembs said...

'high information flow' really depends on how far you cast the net of your literature coverage. What is still relevant for you and what isn't? This is not only confined by your field but also by your own interests as to where you get interesting ideas from.
My field is so small, I probably have one or two relevant papers a year. However, there are so many closely related fields in many models systems, humans and robotics, that I basically have to follow the entire literature somehow.
And since there is no empirical evidence that journal rank does anything, journal rank is probably the most useless filter one can think of. You might as well have a script pick out random papers from the entire literature:
Thus, what we need is a smart way of tracking the relevant literature. Some attempts to provide such a service here and there (and I'm involved in beta-testing some of them), but at the moment, this simply does not exist.
Consequently, I spend about 6-10 hours every week searching the literature. Ironically, this is so much time, that there is usually hardly any time left to actually read what I found.