
in 2005 the Year Hurricane Katrina
struck New Orleans the average 48 hour
track error for hurricane predictions
was a hundred and ten nautical miles
this means two days before a hurricane
made landfall we could expect the
prediction for where it would strike to
be wrong by an average 125 statute miles
125 miles is about the difference
between evacuating New Orleans Louisiana
and Pensacola Florida in 1990 that error
was 200 miles today it’s about 65 200 to
125 265 in 27 years what’s changed
meteorology has gotten better satellites
have gotten more sophisticated yes
there’s one particular tool in
scientists and forecasters bag of tricks
it gets exponentially better every year
and perhaps more than anything else has
been instrumental to their growing
capability to make predictions and to
save lives and it has to do with
make-believe all over the world today
there are big powerful supercomputers
whose job it is to pretend to be storms
we can see how they’ll move how they’ll
strengthen and weaken and every year
these machines get bigger and faster and
therefore better able to pretend to be
those storms technically we call it
simulation and modeling but it’s really
about making the computers pretend and I
know that sounds like a play school kind
of thing to say but why use a $5 word
when a 50 cent one will work we make the
computer pretend there’s all sorts of
really interesting things going on in
this world that are either too big or
too small or too fast or too slow or too
expensive or too dangerous for us to
touch with our hands or
watch with our eyes but we can make the
computer pretend now hurricanes are big
and fast and dangerous
how about something small our bodies are
made up in part of proteins which
perform many important tasks in order to
function
these proteins first have to assemble
themselves or fold into particular
shapes now there’s a lot about the
process of protein folding inside the
body that we don’t yet understand but we
do know that when this process goes
wrong we can wind up with conditions
called
protea path ease these are things like
Huntington’s Parkinson’s Alzheimer’s we
would like to have the ability to watch
these proteins miss fold inside the body
but that’s not practical now we do
understand most of the physics and
chemistry involved so we can make the
computer pretend watch the proteins fold
and miss fold inside the computer
simulation watch for the critical moment
where things go wrong and use that
knowledge to suggest drugs that could
help now the rules of this game and the
outcome have to be managed by experts in
this field biochemists perhaps but it
gives them an incredible and potentially
life-saving capability that they
otherwise would not have let’s talk
about something expensive
in 2003 Goodyear the largest American
tire maker was in trouble they were
falling behind their overseas
competitors they had a product cycle
that took three years to go from
inception to market and it was based in
large part on educated guesswork they
would build tires test how they
performed and importantly they would see
how they would wear out over time and
that year they made a key decision stop
doing start pretending they used super
computers to pretend to be tires watch
how they performed on pavement on gravel
on water on sand but most importantly
they used those simulations to predict
how the tires would age over time and
the result was that three year cycle
became a 1 year cycle and building and
testing tires went from 40 percent of
their R&D budget to 15 three years later
they posted record profits how about
something slow computers can pretend to
be the world too but not yet all of it
climate change may be the key issue of
our time and computing can let us see
how our decisions today impact the
future we can model the atmosphere and
make the computer pretend to be the
decades and centuries that pass as we
adjust carbon or methane or ozone or ash
but there’s not yet enough computing
power in the world for us to turn as
many of those knobs as quickly and in as
many combinations as we would like but
we’re on our way and the day will come
that the countries of the world will be
able to come together negotiate climate
treaties armed with computers powerful
enough to pretend to be the planet and
know quickly and confidently what the
decisions they make that day will mean
for the future of the earth a lot of
human effort is spent trying to figure
out how the world works trying to
understand the unknown and faced with
the unknown we have some choices doing
is very important but we can also
pretend and this is why computing
matters at virtually every scale and in
every field in fact where I work at the
tandy supercomputing Center we play
everywhere along this spectrum from
helping frontline university researchers
do basic science to helping transfer
technologies out of the university
research lab into the marketplace to
helping small and mid-sized companies do
today what Goodyear did 15 years ago but
it’s not just about big huge million
dollar machines if you recall that
protein folding problem
I talked about a little bit ago when you
go home today you can actually join your
PC or even your phone to a worldwide
network of computers pretending together
to fold proteins
it’s called folding at home and there’s
a lot of other really worthy projects
out there too that I hope you’ll check
out because when the experts set the
rules of the game and let computers play
pretend the results can change the world
thank you
[Applause] [Music]