[mxschumacher]

I wonder how the "Publish the Same Result Several Times" recommendation ties into the "publish or perish"-mantra. I suspect that 1997 was a different time. Many academics have a super-human publication cadence, not so much because of all the new knowledge they've extracted from the universe but because they aggressively re-publish their existing results. Very low read- and citation counts are common implications.

From the outside, it seems hard to sift through all the redundancy. At the edge of knowledge it gets harder to verify what is genuinely new and what is a rehash.

It feels like the infrastructure and format around scientific publishing could benefit from compression, consistency and version control.

[generationP]

> "Publish the Same Result Several Times"

Do not take this too literally. Usually it won't work or will get you branded as a self-plagiarist (not exactly a deadly sin, but not a good reputation either; people will discount your work for this). In Rota's example of Riesz, the first publication was "in some obscure Hungarian journal", the second one was in Comptes Rendus, and the third was in a "real" journal. The choice of venues is no accident; you couldn't do it the other way round. Comptes is specifically for short (ca. 4 page long) communications that mostly serve to announce results and ideas; it's not unlike the "extended abstracts" you submit to conferences. And there are no obscure Hungarian journals any more in the age of the Internet. You won't get the same result published in 3 "normal" journals unless at least two referees are asleep at the wheel.

The modern way to get high publication count "for free" is doing a lot of incremental work (as opposed to trying to write up something definitive and general) and splitting your papers up into "least publishable units". These aren't bad things to do, though: Often you have to do incremental work as a warm-up before you have a chance to see the full picture (unless you are Grothendieck, presumably), and the thinner your papers are, the easier (and faster) they will get through peer review. It's a matter of knowing where to stop.

[MaxBarraclough]

There's a Wikipedia article on this: https://en.wikipedia.org/wiki/Least_publishable_unit

[Forge36]

I think this was the version control when paper was big :)

Over of the other points on the article was "lengthen your summary" which was to address that very low read count!

I wonder if it's possible to make a "github of research" where articles are more publicly revised only showing the most recent (celebrating Theory 2.0! And 2.0.1 -fixed typos)

I took the publish process to be: iterate on your thesis.

I saw the pattern as: Ask for feedback from a small group. Improve ask a larger, more reputable group. Repeat.

I'm not current on what constitutes a prestigious Journal today (is Nature still a top Journal?).

[ctchocula]

arxiv.org lets you submit revisions to a paper you have previously submitted and putd the label v2, v3, etc. A lot of preprints will go up for physics and computer science papers, while medrxiv.org fills the same role for health sciences.

[anjc]

> they aggressively re-publish their existing results. Very low read- and citation counts are common implications.

I think it's generally the opposite to a great degree. Some people spend their PhD and postdoc doing one large piece of work, overly concerned about its rigour/correctness/validity, eventually publishing just one paper that they're happy with (perhaps a summary of their thesis). Others publish smaller pieces of work frequently, at conferences, workshops etc. In this time the latter will have built up a public academic profile, while the former may be essentially anonymous to the field. This will be reflected in greatly more citations and reads for the latter.

You could argue that there's value to the former approach, but if you can't disseminate your work effectively - i.e. nobody knows that you or your work exist - then it's useless (imo).

[projektfu]

The way it’s done now is to publish tiny incremental results. For example, you can get funding to study a class of a dozen drugs in a dozen exotic species. The results can be written in any number of papers from 1 to 144. Add two different test methods, and you can multiply as you see fit. If the journals reject a paper for being too small, just lump a few together and resubmit.

[FabHK]

One professor of mine talked about the LPU, the least publishable unit that is sufficient for a paper.

Even has a wikipedia page, which also mentions "salami publishing"...:

https://en.wikipedia.org/wiki/Least_publishable_unit

[ketzu]

Personally I interpreted the remark differently. But it also may depend on the field you are in.

First, publish the idea, the rough concept. You can get valueable feedback from this.

Second, publish an implementation of the idea and the evaluation thereof. This is the actual result.

Depending on limitations it can be worthwhile to publish an extended and revised version of the second one. You can go in excruciating detail on this and include unremarkable side results.

Depending on your field you might be limited to 8 pages for your second publication, which is often very limiting, or your third publication might be your thesis linking several results together.

Version control would be nice though. Especially when you notice errors yourself you would like to fix.

[czep]

> what is genuinely new and what is a rehash.

I struggled with this in grad school, as it often felt I was rereading the same material in multiple journals, and the examples of my successful peers appeared to me to be the same paper with 8 different introductions!

But the example Rota gives of Riesz is illuminating: by refining an idea over time, not simply rehashing, the product improves and eventually becomes "perfect". I think this is the original and benign intent of what has often devolved into the appearance of "publish or perish".