Monthly Archives: December 2013


Sebastian Wernicke: Lies, damned lies and statistics (about TEDTalks)

This is one of those Ted talks that I’ve considered writing about back when I had first started.  (I even had to double check my history to make certain I didn’t write about it already.)

It is a very entertaining piece about the meta data of Ted Talks and how at the end of it all, everything can be summarized ‘poorly’ using data science.

Views to date: 1,355,648

Now I do enjoy a good statistic, its a topic I don’t talk frequently enough about.

Odds are, statistics aren’t pointed out.  (give or take a 5 percent variance)


Ah forget the failed attempts of humor, here’s some quick factoids that this TED Talk brought to mind.

The term, “Lies, Damned Lies and Statistics” is a commonly used saying made popular by Mark Twain in 1906 when he published “Chapters from My Autobiography”.  The statement itself was a variation on another phrase used much earlier in 1885 (just not as popular) in the description of witnesses: “Liars, damned liars, and experts.”

Statistics, represents that expert which is also a jerk.  As it has also become incredibly common to use numbers in everyday use from commercials (“3 out of 4 dentists agree”), symbolic translations (“your voice is currently at a 7, I need you to be at a 3”) and of course, the modern day meme (“its over 9,000!”)

Interestingly enough, the very first time I heard the phrase, it was uttered by the villain known as ‘The Penguin’ in the issue where Batman fights The Actuary!  (Detective Comics #683)

Little more on the history of the phrase here:,_damned_lies,_and_statistics

And a link for more Batman:

(Why does it mostly just show the cover?  Well as people we relate to pictures, etc etc)

Of the three topics he did statistical analysis for: topic, delivery and visuals

If we wanted to find where the statistics ‘lie’, it would be in the correlation between these three (plus an unmentioned fourth).  In other words, chances are for a very technical Ted talk (topic) will be using technical jargon (in its delivery) and images used are more for proof of concept rather than pure entertainment (aka having visuals that make you wish you were blind). 

In other words, if the topic is about apples, it will also include the words like ‘delicious’ and involve colors like red and green.  If a Ted Topic is happens to be popular, the accompanying words and images will be popular.

And that is where the fourth element kicks in, the would be the audience member.  Going on a ‘statistical guesstimate’, I’d stake that audience members who believe the ‘average joe’ would enjoy the Ted Talk would be shared more often (and those that require an engineering degree to appreciate, might only be shared among fellow engineers).

The one thing that Ted Talk statistics can’t account for, is the changing platform of people’s tastes.

Statistics, from the internet I stab at thee!

Lies, Damned Lies and Share,


Leave a comment

Posted by on December 13, 2013 in Uncategorized