Friday, July 04, 2008

That Conference in Changsha, China (Part I)


Two weeks ago, Tobi and I were in Changsha, China attending a conference at the Frontiers of Algorithmics Workshop (FAW) 2008. While I thought we delivered a rather exquisite 20-minute presentation of our work on "Versioning Tree Structures by Path-Merging", we both found that there were several, ah, "speakers" among the conference delegates, who, in our opinion, strived very hard to give seriously bad and ineffective talks! Although, strangely, I must confess, that they did write excellent and well-organised articles in their contributions to the conference proceedings -- such constrasting malediction!

And so, as would anyone who are forced to listen to a series of uncomprehensive repertoires of monotonic regurgitations of Lemmas and Theorems in between bland syntaxes and numerical tables, Tobi and I made a list of what we thought were going on in the minds of these highly esteemed people whose only mission seem to drive everyone sitting in the audience to lose interest / sleep / be dilatory / stare blankly at walls / unnecessarily draining litres of bottled water (to keep awake) / (and thus) making excuses to visit the toilets / etc... etc...

The cogent title we gave, in our quest to be superb raconteurs of accounting the details of the three-day conference proceedings, was:

"How To Give a BAD Talk"

Read on, and be amused...


    [Day 1: 19 June 2008]
  • Present 6 huge and extremely complicated Lemmas followed by a Theorem explaining the universe in the span of 20 minutes.

  • Show 100 lines of source codes in a separate editor with font size 10.

  • Prepare enough slides to give a 4-hour long talk and present them all in 20 minutes.

  • Don't bother to explain any of the abbreviations / acronyms / unit measurements / etc. that are on the slides -- but keep mentioning them in the talk all the time since everyone else are assumed to have already know what they mean.

  • Present, on each slide, at least 6 very complicated mathematical formulae, using more than 2 super-/sub-scripted indices without explaining the significance of the variables.


  • [Day 2: 20 June 2008]
  • Place 5 Definitions and 3 Lemmas on 2 slides, about 25 lines each copied from LaTeX.

  • Show 20 lines of fully commented pseudo-"pseudo-codes" on 1 slide.

  • Present experimental results by showing endless columns of numbers without properly explaining their relevance / consequence / importance / purpose.

  • Don't ever use graphs to represent these endless columns of numbers!

  • Don't ever use figures to illustrate graph-theoretical principles. Instead, draw them impressively in the air with your fingers, some distance above the podium, while talking.

  • Periodically ask for the audience's approval, after the delivery of every thought / idea, by saying, "OK?".

  • If your time is over, defy the session chair's suggestion to end the talk and continue to keep reminding the audience that there is still a lot more contents to come.

  • Create slides that dont't match the screen dimensions; for example, use A4-portrait.

  • Include example figures in the slides. Then ensure that the audience sees them -- or rather, make sure they only manage a quick glance at them; then quickly move on to other another slide. The figures only serve to distract.

  • Skip the motivation for solving the problem, simply assume that the audience already knows.


  • [Day 3: 21 June 2008]
  • Don't bother to do a final review of the contents in the slides to spot for typos or something similar. The slides are always already perfect the first time round.

  • Copy everything from LaTeX to PowerPoint, and don't follow up to ensure the correctness of contents. For example, in LaTeX: "$c n\log n$" -- simply copy and remove the "$" after placing in PowerPoint.

  • Add references such as [1],[2],... without ever letting the audience know the concrete sources. Until, of course, we reach the end of the talk. These references are only meant for our own conveniences, and no one else's.

  • Use tiny font sizes in all the example figures in the slides. This prevents the audience from spotting your mistakes, which, of course, are non-existent.

  • Bring the audience's attention to your figures on the slides only by their label names, and don't bother to explain the significance of those figures. For example, say something like, "Here are our results. Look at Figures (a) and (b)." Then pause for 2 seconds, and quickly turn to the next slide and conclude the talk. Then watch the stunned reactions of the audience.