Thursday, September 19, 2013

Don't show tables in your talk

Previously I have written about that tables are wonderful in papers.
This is because science is all about comparisons.

However, I think tables should not be shown [banned?] in seminars and conference talks.
Flashing a detailed table of data on the screen for a minute is useless. The audience does not have nearly enough time to absorb and process the table, even when you verbally explain the main points. This will increase the tendency of the audience to tune out.

I suspect computational chemists are particularly bad at this. They like to show all these quantities they have calculated with different levels of theory and different basis sets.

If you want to highlight a trend [or lack of one] you need to graphically represent the data. The audience can then quickly understand and assess the result.

I can think of a couple of exceptions from my own talks. Then I have merely a the table as an "existence proof".
For example,
"The parameters describing the spectral density for a wide range of chromophores in a wide range of solvents and proteins have been determined experimentally. Here is a table from my paper showing that. Nevertheless, some theorists write papers with unrealistic spectral densities."

1 comment:

  1. The couple of talks I've given on our paper PRB 87, 085411 (2013), I have shown Table I, which highlights that experimentalists have obtained many different, apparently arbitrary, values for a number which had previously been called a topological quantum number.

    However, on reflection, perhaps showing this variation in a graph is more talk-worthy. And in fact, there are nice examples of such plots (e.g., variation in measured gravitational constant from different groups, and variation in measured DC conductivity of graphene from different groups. The variation being much more than the error bars visually highlighting the lack of a good theory works much better in a plot than in a table).

    A (perhaps trivial) counter-example I can think of is the wonderful topological insulator table produced by Schnyder et al. In his and Ryu's talks, they tended to show the table (of course, the paper is about a classification scheme, so it's probably particularly appropriate). Something that works really well, from my perspective, is that they then put arrows onto the table that show what physical systems correspond to different positions on the table. It's quite visually striking because they're showing this unifying scheme that connects quantum Hall physics, polyacetylene, Helium 3, superconductors, Z2 top-insulators, etc. The table is also visually striking as it is, for instance if you increase dimension by one, the topological index moves down one spot on the table, so that there is this intimate connection between the symmetry groups and dimensionality.

    E.g., slide 13ff of this talk by Shinsei Ryu:

    So I think I'm against banning! But then perhaps I agree that finding a better way to present the point that the table is supposed to show, would be ideal, when possible. But also more work!