Sunday, February 16, 2020

Pre-(Publish and Perish)

Guest post by Evangelos Georgiadis

Quite a few posts have recently focused on papers,publications and venues; "optimal" venues for papers under different objective functions,e.g. minimizing carbon footprint while maximizing community building, networking as well as information sharing, see Moshe Vardi.

Here we would like to take a closer look at one of the key assumptions -- the paper. In order to generate a paper, one needs to come up with a result, something novel, fresh or interesting to say. The question that has baffled this author is what represents a conducive or perhaps even optimal setting for generating papers. Since papers come in different flavors ranging from "solid technical papers to risky innovative ones" the settings may vary; but ultimately, what would be interesting to investigate (or for that matter crowdsource) is whether there is a common denominator in terms of setting or environment, a necessary but not sufficient condition (so to speak).

Here are some accounts of others which may be helpful as reference points.

Knuth's papers entitled "Semantics of context free grammar" along with "The analysis of algorithms" represent two instances that suggest research institutes might not provide an optimal environment for idea generation.

As Knuth points out in "Selected Papers on Computer Languages" (Chapter 18, p. 431):
Perhaps new ideas emerge most often from hectic, disorganized activity, when a great many sources of stimulation are present at once -- when numerous deadlines need to be met, and when other miscellaneous activities like child-rearing are also mixed into the agenda.
Knuth goes on to say, that it was challenging to do creative work in office and that finding a few hideaways provided some form of solution -- aka sitting under 'that' oak tree near Lake Lagunita. That said, the inspirational setting for getting into the zone for the aforementioned two papers were provided by (Californian) beaches. Hold that observation. Is this not something we have come across somewhere else ? Fields medalist Stephen Smale in "Chaos: Finding a Horseshoe on the Beaches of Rio" suggests that some of his best work happened at his "beach office". Whether beaches do provide for a good setting remains to be shown; perhaps for very innovative ideas, oceanic freedom is necessary. That said, the author recalls (hopefully accurately enough) an account by the young James H Simons, who attended a conference in Japan in the early days. Instead of choosing a spacious accommodation (which he was able to afford), he restricted himself to the typically confined room type -- not only confined by space, but also pressured by time, young Simons was able to generate an interesting result for that conference. (This probably demonstrates that technical results don't necessarily require 'oceanic freedom'.)

Some meaningful probabilistic advice comes from the fat-tails department, in "The Black Swan" by Nassim Taleb (on page 209) : "Go to parties! If you're a scientist, you will chance upon a remark that might spark a new research. "

Murray Gell-Mann provides an interesting collective account in his Google Tech Talk entitled "On Getting Creative Ideas." He recollects a workshop he attended in 1969 in Aspen that focused on the experience of getting creative ideas, not just among mathematicians and theoretical physicists but also poets and artists. This account seems to neglect the actual setting that might nurture creative thought process, but provides interesting references to people such as Hermann von Helmholtz, who happened to have thought about this topic and partitioned the process in terms of "saturation, incubation and illumination".

For those interested in an account that focuses on the Eureka moments of exclusively mathematicians/theoretical physicists see Jacques Hadamard's book "The Mathematician's Mind". Hadamard iterated on Helmholtz's 3 stage process and it's worth taking a look at what he came up.

At last, what are good venues or workshops for generating papers ? Or let's rephrase that a bit, what type of atmosphere at venues fosters creativity -- what food for thought to provide participants and how to distribute that food for thought over a given day ? Ryan R Williams proposed (as practiced by 34th Bellairs Winter Workshop on Computational Geometry) "... easy problems, informal atmosphere focusing exclusively on thinking about problems in a cycle of down-time where one meets in two intense sessions and have free time otherwise." (This type of setting seems to resonate with the 3 stages of "saturation, incubation and illumination".)

That said, most workshops including the Simons workshops don't seem to follow such a recipe. They are more geared towards the follow-up step, namely, communicating what people have found, rather than collaborating with them to tackle open problems. Perhaps some re-evaluation might be required in how workshops are run.