Thursday, July 6, 2017

Are theoretical physics and chemistry amenable to online collaboration?

Last week at UQ we had a very nice mathematics colloquium, "Crowdsourcing mathematics problems" by Timothy Gowers.
He first talked about the Polymath project, including its successes and marginal contributions.
He then talked about a specific example of a project currently underway on his own blog, concerning transitive dice. This was pretty accessible to the general audience.

This is where a well defined important problem is defined on a blog and then anyone is free to contribute to the discussion and solution. A strength of this approach is that it makes use of the complementary strengths, experience, and expertise of the participants. Specifically, solving problems includes:
  • selecting a problem that is important, interesting, and potentially ripe for solution
  • defining the problem clearly
  • breaking the problem down into smaller parts including conjectures
  • sketching a possible heuristic argument for the truth of the conjecture
  • giving a rigorous proof of the conjecture
  • finding possible counter-examples to the conjecture
  • connecting the problem to other areas of mathematics
This can be efficient because dead ends and mistakes are found quicker than someone working in isolation. 
People are more motivated and engaged because they are excited to be working on something bigger than themselves and what they might normally tackle. And they enjoy the community.
What about assigning credit in such group work? There is a clear public record of who has contributed what. Obviously, this does not work for bean counters looking at traditional metrics.
This approach mostly attracts senior people who are secure in themselves and their career stage and more interested in solving problems than getting individual credit.

The cultural differences of pure mathematics and physics was striking. The talk was given on whiteboards and blackboards without notes. No powerpoint! The choice of research problems was purely based on curiousity, not any potential practical value or the latest fashion. It is fascinating and encouraging that the pure mathematics community is still "old school" with the focus on quality not quantity.

Aside: Gowers is also well known for initiating a boycott of Elsevier journals.

Now, my question. 
What is stopping theoretical chemistry and physics from such a "crowd sourcing" approach? 
Is it that the problems are not amenable? 
Or is it largely that we are too driven by a system that is fixated on individual credit?


  1. I would fully endorse it for theoretical physics if you were to start such a program --- it needs someone established/prominent and with this mind-set to kick things off.

  2. The problem for that idea in theoretical chemistry is that the "big" problems are seen as being in physics (left to physicists) or biology, where big ideas or big computers are seen as what is needed, not big teams.

  3. There is a nice example in physics for determinant quantum Monte Carlo (DQMC) of open science working really well. Terrance Tao is even involved. I couldn't do a better summary than the original source I read it on, but the entire process was done on mathoverflow, and there is a resulting paper (in PRL). The paper is interesting to read because it tries to give credit to everyone that contributed, but many were anonymous.

    Here is the original blog post I read about it:

  4. I think the problem is that even in the most theoretical parts of chemistry and mathematics there are very few formalised conjectures. Henry's example is not really an instance of open science in physics, as the open part was purely mathematical: proving the positivity of a particular determinant.

    It seems that the problems that work best involve both a conjecture and a strategy, perhaps one that is "elementary" (in the mathematical sense of not involving preexisting technical machinery) so that one can most benefit from the wisdom of the crowd.

    One CAN think of sharp questions in theoretical physics. An example: Is there long range order in the ground state of the 2D spin-1/2 Heisenberg antiferromagnet? Although almost everyone believes this on the basis of approximate treatments and numerics, very few physicists have the inclination or ability to address the proof. I would probably go as far as to say that once a question has been posed sufficiently precisely to be understood unambiguously, it has effectively become mathematics. That's why you can put up a cash prize for a solution of a specific math problem.

    Would love to hear of counterexamples, of course!

  5. Sorry to nitpick, but "Is theoretical physics and chemistry amenable to online collaboration?" -> "Are theoretical physics and chemistry amenable to online collaboration?" .. ?

  6. Fundamental or theoretical physics one can infuse into theoretical chemistry online. The other way around one doubts it. The fact that more physicists have won Nobel prizes in chemistry is a testimony to this. Even in engineering many aspects is physics on a large scale. Mass transfer with chemical reaction considered as the intellectual triumph of chemical engineering can be described as the physics of chemical kinetics. (Ed Cussler, Chem Tech article).

  7. "The talk was given on whiteboards and blackboards without notes. No powerpoint! ....

    The Cognitive Style of PowerPoint Prof Edward Tufte Yale Univ His criticism
    Criticism of PowerPoint[edit]
    Tufte has criticized the way Microsoft PowerPoint is typically used. In his essay "The Cognitive Style of PowerPoint", Tufte criticizes many aspects of the software:

    1. Its use to guide and reassure a presenter, rather than to enlighten the audience;
    2.Its unhelpfully simplistic tables and charts, resulting from the low resolution of early computer displays;
    3. The outliner's causing ideas to be arranged in an unnecessarily deep hierarchy, itself subverted by the need to restate the hierarchy on each slide;
    4.Enforcement of the audience's lockstep linear progression through that hierarchy (whereas with handouts, readers could browse and relate items at their leisure);
    5. Poor typography and chart layout, from presenters who are poor designers or who use poorly designed templates and default settings (in particular, difficulty in using scientific notation);
    6. Simplistic thinking—from ideas being squashed into bulleted lists; and stories with beginning, middle, and end being turned into a collection of disparate, loosely disguised points—presenting a misleading facade of the objectivity and neutrality that people associate with science, technology, and "bullet points".
    7. Tufte cites the way PowerPoint was used by NASA engineers in the events leading to the Space Shuttle Columbia disaster as an example of the many problems. The software style is designed to persuade rather than to inform people of technical details. Tufte's analysis of a NASA PowerPoint slide is included in the Columbia Accident Investigation Board’s report—including an engineering detail buried in small type on a crowded slide with six bullet points, that if presented in a regular engineering white paper, might have been noticed and the disaster prevented.[14][15]

    Instead, Tufte argues that the most effective way of presenting information in a technical setting, such as an academic seminar or a meeting of industry experts, is by distributing a brief written report that can be read by all participants in the first 5 to 10 minutes of the meeting. Tufte believes that this is the most efficient method of transferring knowledge from the presenter to the audience and then the rest of the meeting is devoted to discussion and debate.