Thread by @JakeVigdor

A few (more) thoughts on the evolution of scholarly publishing, abetted by the metrics computed by the @eigenfactor project.

Authors supply content, journals demand it. Journals publish when quality exceeds a certain threshold.

What happens when supply increases?

Journals can respond by holding the threshold constant and publishing more, or raising the bar for publication. Or somewhere on the continuum between the two.

In economics over the past half-century, the dominant choice has been to raise the bar. Why?

Two words: impact factor.

Impact factor is a measure of average article quality. In a world where there is competition among journals, and the "winner" is the one with highest impact factor, expanding capacity when supply increases is a losing strategy.

This is where the @eigenfactor measures come in.

One can rank a journal by citations per article, or by total citations. @eigenfactor does both.

This image shows trends at the Journal of Political Economy: brown line is based on cites/article, blue is total cites. JPE has boosted the former at the expense of the latter.

Let's try that image again.

Here's the AER for comparison. The AER made a conscious effort to expand available space, and we see that the total impact measure more closely tracks the per-article impact measure.

Ranked by impact per article, JPE>AER. But by total impact the rankings reverse.

The top five journals, between them, publish just over 400 articles per year. So every year there's a competition by authors to see if their article ranks among the top 400. It often requires submission to 3-5 journals in sequence, and typically takes years.

Meanwhile, if you're a chemist, there's a competition to see if your article ranks among the top 3,000 or so. You just need to submit to one journal (JACS), and typically the time to acceptance is under two months.

The economics system is incredibly inefficient.