Joseph Keller used to jog
around the Stamford campus.
And he was struck by all the women
jogging there as well.
Why did their ponytails swing
from side to side like that?
Being a mathematician,
he set out to understand why.
(Laughter)
Professor Keller was curious
about many things.
Why teapots dribble,
or how earthworms wriggle.
Until a few months ago,
I hadn't heard of Joseph Keller.
I read about him in the New York Times
in the obituaries.
The Times had half a page
of editorial dedicated to him,
which you can imagine is premium space
for a newspaper of their stature.
I read the obituaries almost every day.
My wife understandably thinks
I'm rather morbid
to begin my day with scrambled eggs
and "Let's see who died today."
(Laughter)
But if you think about,
the front page of the newspaper
is usually bad news.
And [....] man's failures.
An instance where bad news
[...] accomplishment
is that the end of the paper,
in the obituaries.
In my day job,
I run a company that focuses
on future insights
that marketers can derive from past data.
A kind of rearview mirror analysis.
And we began to think,
what if we held a rearview mirror
to obituaries from The New York Times?
Were there lessons on how
you could get your obituary featured?
Even if you aren't around to enjoy it?
(Laughter)
What does go better with scrambled eggs?
(Laughter)
And so we looked at the data.
2,000 editorial nonpaid obituaries
over 20-month period
between 2015 and 2016.
What did these 2,000 deaths,
rather, lives,
teach us?
Well, first we looked at words.
This here is an obituary headline.
This one is the amazing Lee Kuan Yew.
If you remove the beginning and the end,
you're left with the beautifully
worded descriptor,
that tries to,
in just a few words,
capture an achievement of a lifetime.
Just looking at these is fascinating.
Here are a few famous ones --
people who died in the last two years.
Try and guess who they are.
[An Artist who Defied Genre]
That's Prince.
[Titan of Boxing and the 20th Century]
Oh, yes.
[Muhammed Ali]
[Groundbreaking Architect]
Zaha Haddid.
So we took these descriptors
and did what's called natural
language processing,
where you feed these into a program,
it throws out the superfluous words --
"the" and "and,"
the kind of words you can mime
easily in "Charades," --
and leaves you with the most
significant words.
And we did it not just for these four,
but for all 2,000 descriptors.
And this is what it looks like.
Film, theatre, music,
dance and of course art
are huge.
Over 40 percent.
You have to wonder why
in so many societies
we insist that art gets pr