Worst talk ever

Mea Culpa

This post wouldn't be here if the presenter of said talk felt like he had really worked on his talk, because, this would be impolite pointless overkill.  As it is, the presenter was quite clear that he had prepared nothing, nor was he in the right venue for the talk.  So if anybody is to be embarrassed about the situation, it is the organizers and the audience, i.e. yours truly.

The venue

As some of you know, Woodcoin was developed during my stay in Oxford, England.  While there I regularly attended lectures in the Mathematics department.  The venerable Invariant Society of Oxford University meets there and is an interesting venue for Mathematics talks.  One of the things I like about this informal venue is that the presenter is not expecting anyone in the audience to later be reviewers for his or her grant proposals, nor are the audience members on the speaker's PhD committee.  Neither is the presenter earning a salary for a course directly tied to the talk.  This clears the air a bit of various political or other intrigues that might not be entirely encouraging to the growth of intellectual curiosity.  The few talks that I did see with this group were of quite high quality.  Professors, emeritus and otherwise, chose it as a venue to describe in detail some set of problems they had considered over the years, complete with history, equations, props, anecdotes, and a candor that one might struggle to find elsewhere.

The talk

So!  Boy was I in for a surprise when on May 13, Alex Davies began his talk: "Machine Learning and Sentiment Analysis, or How to find Happiness".  The previous talk I had attended was called "The 27 Lines and Other Stories" presented by Oxford Professor Nigel Hitchin, which had been a surprising romp through relatively complex geometric spaces, complete with odd physical props and personal stories relevant to the authors publications.  So, what would be in store for us today?

To be fair, Mr. Davies didn't really intend to give a good talk.  He began by saying "well sorry, I've prepared nothing, but I gave this talk to some blokes in London the other day, maybe you'll be interested".  Jeans and a tee shirt in Oxford are not as commonplace as other venues.  OK, not a great start, but hey ..  it gets worse.

Lets cut right to the chase.  The man's talk was about Twitter.  About how useful the analysis of tweets can be to judge the happiness of a population.  I kid you not.  At one point he asked us "what is the happiest nation?"  Of course, the crowd was dumbfounded already at this point and nothing would get a sound out of us.  "It is Germany" he said.  "They have the highest fraction of smiley faces in their tweets".   Wow, so happy!

Let me go out on a limb here with a perhaps radically more accurate judge of happiness related to twitter: when you are feeling good, you aren't tweeting.  Right?  Doncha think?  Like imagine those moments of satori, being either one with the universe, connecting to people, doing something well that you like to do.. who knows, du vin, du poesie, ou du virtue..  Were you logging into freaking twitter at that moment?  Uh, no.  Sorry Alex.

So get this.  In the midst of this hopelessly incorrect thesis being presented, with no interesting work to back it up, in an established mathematics department of a reputable institution, he then: apologizes for showing us an equation.

This is always a mistake in any venue.  What you are actually apologizing for is showing us something you don't understand, or showing us something that you won't take the time to explain.  Sure, that might happen.  So say it!  Don't apologize for two things being equal to each other.  Especially not to an audience which is absolutely desperate for any kind of factual statement whatsoever to be presented.  In this case, it was a definition of Bayesian statistics, harvested from wikipedia I assume.  Not any clear relevance to the presentation but hey - statistics right?  I'll wait while you go clean up the vomit.

Feeling better yet?  Don't worry, the worst is over.  A few plots showing numbers of smiley faces vs. sad faces in tweets, and of course displaying some lengthy URLs where one could obtain this oh-so-useful data of people's tweets, and then we were free.  For me, it was the last Invariant Society talk I was able to see.  I'm sure they have recovered, they seemed to be a strong lot.