[Applause]
technology has created grave threats to
life on our planet and now we need
technology to help us address those
climate change is one example where
fossil fuel energy largely created the
problem and now we need clean energy
technologies to help us solve it but
we’re not actually that good at
technological innovation to solve these
big problems we’re a whole lot better at
creating new gadgets with new
functionality that we didn’t even know
we needed for example I recently heard
about something called the umbrella
drone which is a drone that’s
specifically designed to perfectly
balance an umbrella but why aren’t we
better at solving technological problems
that really matter well as I found over
the years I think the answer is that
we’re missing key knowledge about why
technology improves and how to measure
it
and so this makes it really difficult to
be deliberate about solving those big
societal problems so let me give you an
example of this and what an example of
how I spoke to a room of energy experts
last week and and and why that
discussion made me think about this
problem in this way namely that we’re
not as good as we might be in explaining
where technological progress comes from
so the question that I asked this room
of energy experts was why do you think
solar energy costs have fallen and
specifically I asked how much of the
cost decline do you think came from
government funding for research and how
much for how much came from government
incentives to grow markets for this
technology and this is a technology
solar energy that’s often held up as a
success story of clean energy innovation
and this room of experts focused on
renewable energy but there was no
agreement on the answer and the answers
were all over the place and here’s an
example of some of the challenges that
we see in measuring technological
progress let’s take the example of the
electric car everybody knows that the
electric car has evolved over time but
how close is it to being an option that
can see widespread adoption and
specifically can we answer the question
how many cars on the road could be
replaced by a low-cost electric vehicle
today and still meet our daily energy
driving needs and what I found is that
people aren’t able to answer this
question this very basic question but if
we don’t know how to explain techno
technological progress as we saw from
the example of solar energy and we’re
not really able to measure it as we saw
in the example of electric cars then how
are we going to know how to invest our
time and money best in solving these
problems well so this is really what
motivated my research and what I can say
now is that I think we can fill these
gaps and our knowledge about what drives
technological progress and how to
measure it and as a result we’re going
to be able to make better decisions
about how to solve these problems like
climate change and maybe other big
problems in the future so let me go
through how we can actually answer these
kinds of questions
let’s take solar energy so solar module
costs have fallen by two orders of
magnitude over the last 40 years and 90%
in the last decade why did this happen
this is a technology that’s seen faster
and more sustained cost improvement than
other energy technologies where did this
come from to answer this question what
we did in my research
group was to break the technology up
into all of its cost components there
are a few different categories of cost
components materials labor equipment and
then we want to ask what changed about
this device and what changed about the
manufacturing process so there are a
number of different changes but they can
be grouped into these three categories
changes to the device physics changes to
the manufacturing process and changes to
the manufacturing scale now these what
we’re calling low-level mechanisms these
changes act on different components of
the device and some may act on one
component and some may act on many more
and the challenge the real
methodological challenge here was how to
model this mathematically to capture the
effect of each mechanism on the total
cost of solar energy but we did this and
after doing this we can say how much
these low-level mechanisms mattered so
which one was most important what we see
is that the improvement in conversion
efficiency that’s the efficiency with
which you’re converting the energy and
sunlight to electricity is the single
most important mechanism but also we see
that the reasons the fact there were
many different reasons that the cost
came down there are many different
mechanisms that were at play and I think
that may be why this technology has seen
more sustained improvement over time now
that we know what happened to the device
and the manufacturing process we can say
something about what the human efforts
were in the company strategies that led
to this really impressive cost decline
an unprecedented cost decline among
energy technologies and these are what
I’m calling the high level mechanisms
research and development that’s research
by companies as well as
government-funded research
learning-by-doing which is the process
of getting better at doing something the
more you do it and faster at it and also
economies of scale now we can answer the
original question I asked the room of
energy experts which is how much of the
cost decline in solar modules came from
government invest
research and how much came from
government incentives to grow markets
that led to lots of efforts by many
different private companies around the
world roughly 70% of the cost decline
came from market expansion policies and
the rest from government-funded R&D so
now we can answer this question how much
of the solar costs decline came from
government funding how much from market
expansion policies and this is the right
answer now let’s take the example of
electric cars and ask ok how these
technologies are improving but how can
we measure progress toward where we want
to go toward our end goal one of the big
barriers to our end goal to the adoption
of electric vehicles is to meet the
range requirements of drivers right so a
lot of people are deterred from buying
an electric vehicle because they have
what’s called range anxiety which is the
fear that their car will run out of
juice without the possibility to
recharge so his range anxiety justified
well it turns out that on the vast
majority of days it’s not and I’m going
to show you why and how we can go about
answering this question so I’m going to
show you the answer for this one car the
Nissan Leaf but you can check out other
cars on this app that my group released
last year but we’ve actually just
updated today to add more models and as
you can see the Nissan Leaf it’s it’s
labeled number one there and this is a
cost this is a car that costs below that
that’s a low cost car its costs are well
below the average car the average new
car sold in the US today and also even
with today’s electricity supply mix this
is a car that can save substantial
emissions 30% or more relative to the
internal combustion engine a similar
internal combustion engine car so
already today if you buy this car you
can reduce your greenhouse gas emissions
but can you buy this car will it meet
your range requirements and will it meet
the requirements of the entire
population in the US so that’s the
question we want to answer for this
Nissan Leaf and I’ll go through how we
can do this well to do this we actually
have to study how
people are driving on a second-by-second
basis across the country
that’s because depending on how you’re
driving on a second-by-second basis your
energy consumption can be very different
the range of your car is not a single
number depending on how you’re driving
it can fluctuate substantially as I’m
showing here where I’m showing different
what are called drive cycles for the
same car so how far your car can go
depends on how you’re driving as well as
a number of other factors so how do we
study an entire US population based on
how they’re driving on a
second-by-second basis the data didn’t
exist to do this when we started out
doing the research and so we developed a
model to probabilistically match
information and different kinds of data
sets to build a data set on the
second-by-second driving behavior of
millions of cars and people across the
US across all of these no road networks
and now we can answer the question how
many cars could be replaced by a Nissan
Leaf and still meet that daily energy
needs of those drivers without having to
recharge and that number is 87% for that
particular car the number is actually
remarkably similar across more sprawling
and more dense cities so even though the
likelihood that you own a car and you
drive a cars very could be very
different in Houston and New York
there’s a certain similarity in how
people drive that we discovered through
this research which means that this
adoption potential is really quite
similar across these cities so now we
can answer the question what percentage
of cars on the road could be replaced by
the Nissan Leaf or another similar kind
of low-cost electric vehicle like the
Ford Focus and there are other options
and still meet daily energy needs and
that number as I showed you before is
about 90% we can do this for thousands
of other car models looking forward as
batteries improve you can increase that
adoption potential but what we also see
in this research is that there’s a small
number of high-energy days which are
going to be difficult to satisfy it’s
going to be difficult to satisfy those
needs with an electric vehicle
so one of the things that we’re doing is
to use the predictive power in our model
to develop an app that will allow people
to find the the vehicle that meets their
needs – so to match cars to people’s
needs on a particular day as well as
when they’re considering what what car
they may want to buy so this may help
people become more comfortable those
that want to buy an electric vehicle
give them the information empower them
with the information to perhaps get over
their range anxiety and you know there
are other lessons that come from this
research as well so it in the solar
energy example with this model of why
technology’s particular technology
improves and where that progress comes
from we’re also able to provide advice
on to engineers as well as companies and
government policymakers as to where what
strategies might be most effective to
drive the cost down further and as I
mentioned in the example of the electric
vehicle now that we can measure progress
against people’s needs we can also
provide some insight on how to develop
this technology further both of these
are technologies that could reduce
greenhouse gas emissions and there are
many other technologies that that could
could help with that as well where we
can apply this approach so what does all
of this mean well as I as I’ve shown
through these two examples we actually
can measure progress toward
technological progress toward these
broad societal goals and we can explain
where it comes from this will help us
make better decisions about how to
invest our time and money in developing
technologies that will help us solve
problems like climate change and maybe
other problems as well in the future so
this doesn’t have to be guesswork as an
engineer as a company as a government
policy maker as a private individual
this doesn’t have to be guesswork we can
we can actually say something about how
to choose technologies and develop
technologies to get where we
want to go we’re really good at
developing technologies that we don’t
need now the thought I’d like to leave
you with is that now let’s use
science-based data-driven approach to
develop technologies to make
technologies to adopt technologies to
solve the big problems that matter most
thank you
[Applause]
technology has created grave threats to
life on our planet and now we need
technology to help us address those
climate change is one example where
fossil fuel energy largely created the
problem and now we need clean energy
technologies to help us solve it but
we’re not actually that good at
technological innovation to solve these
big problems we’re a whole lot better at
creating new gadgets with new
functionality that we didn’t even know
we needed for example I recently heard
about something called the umbrella
drone which is a drone that’s
specifically designed to perfectly
balance an umbrella but why aren’t we
better at solving technological problems
that really matter well as I found over
the years I think the answer is that
we’re missing key knowledge about why
technology improves and how to measure
it
and so this makes it really difficult to
be deliberate about solving those big
societal problems so let me give you an
example of this and what an example of
how I spoke to a room of energy experts
last week and and and why that
discussion made me think about this
problem in this way namely that we’re
not as good as we might be in explaining
where technological progress comes from
so the question that I asked this room
of energy experts was why do you think
solar energy costs have fallen and
specifically I asked how much of the
cost decline do you think came from
government funding for research and how
much for how much came from government
incentives to grow markets for this
technology and this is a technology
solar energy that’s often held up as a
success story of clean energy innovation
and this room of experts focused on
renewable energy but there was no
agreement on the answer and the answers
were all over the place and here’s an
example of some of the challenges that
we see in measuring technological
progress let’s take the example of the
electric car everybody knows that the
electric car has evolved over time but
how close is it to being an option that
can see widespread adoption and
specifically can we answer the question
how many cars on the road could be
replaced by a low-cost electric vehicle
today and still meet our daily energy
driving needs and what I found is that
people aren’t able to answer this
question this very basic question but if
we don’t know how to explain techno
technological progress as we saw from
the example of solar energy and we’re
not really able to measure it as we saw
in the example of electric cars then how
are we going to know how to invest our
time and money best in solving these
problems well so this is really what
motivated my research and what I can say
now is that I think we can fill these
gaps and our knowledge about what drives
technological progress and how to
measure it and as a result we’re going
to be able to make better decisions
about how to solve these problems like
climate change and maybe other big
problems in the future so let me go
through how we can actually answer these
kinds of questions
let’s take solar energy so solar module
costs have fallen by two orders of
magnitude over the last 40 years and 90%
in the last decade why did this happen
this is a technology that’s seen faster
and more sustained cost improvement than
other energy technologies where did this
come from to answer this question what
we did in my research
group was to break the technology up
into all of its cost components there
are a few different categories of cost
components materials labor equipment and
then we want to ask what changed about
this device and what changed about the
manufacturing process so there are a
number of different changes but they can
be grouped into these three categories
changes to the device physics changes to
the manufacturing process and changes to
the manufacturing scale now these what
we’re calling low-level mechanisms these
changes act on different components of
the device and some may act on one
component and some may act on many more
and the challenge the real
methodological challenge here was how to
model this mathematically to capture the
effect of each mechanism on the total
cost of solar energy but we did this and
after doing this we can say how much
these low-level mechanisms mattered so
which one was most important what we see
is that the improvement in conversion
efficiency that’s the efficiency with
which you’re converting the energy and
sunlight to electricity is the single
most important mechanism but also we see
that the reasons the fact there were
many different reasons that the cost
came down there are many different
mechanisms that were at play and I think
that may be why this technology has seen
more sustained improvement over time now
that we know what happened to the device
and the manufacturing process we can say
something about what the human efforts
were in the company strategies that led
to this really impressive cost decline
an unprecedented cost decline among
energy technologies and these are what
I’m calling the high level mechanisms
research and development that’s research
by companies as well as
government-funded research
learning-by-doing which is the process
of getting better at doing something the
more you do it and faster at it and also
economies of scale now we can answer the
original question I asked the room of
energy experts which is how much of the
cost decline in solar modules came from
government invest
research and how much came from
government incentives to grow markets
that led to lots of efforts by many
different private companies around the
world roughly 70% of the cost decline
came from market expansion policies and
the rest from government-funded R&D so
now we can answer this question how much
of the solar costs decline came from
government funding how much from market
expansion policies and this is the right
answer now let’s take the example of
electric cars and ask ok how these
technologies are improving but how can
we measure progress toward where we want
to go toward our end goal one of the big
barriers to our end goal to the adoption
of electric vehicles is to meet the
range requirements of drivers right so a
lot of people are deterred from buying
an electric vehicle because they have
what’s called range anxiety which is the
fear that their car will run out of
juice without the possibility to
recharge so his range anxiety justified
well it turns out that on the vast
majority of days it’s not and I’m going
to show you why and how we can go about
answering this question so I’m going to
show you the answer for this one car the
Nissan Leaf but you can check out other
cars on this app that my group released
last year but we’ve actually just
updated today to add more models and as
you can see the Nissan Leaf it’s it’s
labeled number one there and this is a
cost this is a car that costs below that
that’s a low cost car its costs are well
below the average car the average new
car sold in the US today and also even
with today’s electricity supply mix this
is a car that can save substantial
emissions 30% or more relative to the
internal combustion engine a similar
internal combustion engine car so
already today if you buy this car you
can reduce your greenhouse gas emissions
but can you buy this car will it meet
your range requirements and will it meet
the requirements of the entire
population in the US so that’s the
question we want to answer for this
Nissan Leaf and I’ll go through how we
can do this well to do this we actually
have to study how
people are driving on a second-by-second
basis across the country
that’s because depending on how you’re
driving on a second-by-second basis your
energy consumption can be very different
the range of your car is not a single
number depending on how you’re driving
it can fluctuate substantially as I’m
showing here where I’m showing different
what are called drive cycles for the
same car so how far your car can go
depends on how you’re driving as well as
a number of other factors so how do we
study an entire US population based on
how they’re driving on a
second-by-second basis the data didn’t
exist to do this when we started out
doing the research and so we developed a
model to probabilistically match
information and different kinds of data
sets to build a data set on the
second-by-second driving behavior of
millions of cars and people across the
US across all of these no road networks
and now we can answer the question how
many cars could be replaced by a Nissan
Leaf and still meet that daily energy
needs of those drivers without having to
recharge and that number is 87% for that
particular car the number is actually
remarkably similar across more sprawling
and more dense cities so even though the
likelihood that you own a car and you
drive a cars very could be very
different in Houston and New York
there’s a certain similarity in how
people drive that we discovered through
this research which means that this
adoption potential is really quite
similar across these cities so now we
can answer the question what percentage
of cars on the road could be replaced by
the Nissan Leaf or another similar kind
of low-cost electric vehicle like the
Ford Focus and there are other options
and still meet daily energy needs and
that number as I showed you before is
about 90% we can do this for thousands
of other car models looking forward as
batteries improve you can increase that
adoption potential but what we also see
in this research is that there’s a small
number of high-energy days which are
going to be difficult to satisfy it’s
going to be difficult to satisfy those
needs with an electric vehicle
so one of the things that we’re doing is
to use the predictive power in our model
to develop an app that will allow people
to find the the vehicle that meets their
needs – so to match cars to people’s
needs on a particular day as well as
when they’re considering what what car
they may want to buy so this may help
people become more comfortable those
that want to buy an electric vehicle
give them the information empower them
with the information to perhaps get over
their range anxiety and you know there
are other lessons that come from this
research as well so it in the solar
energy example with this model of why
technology’s particular technology
improves and where that progress comes
from we’re also able to provide advice
on to engineers as well as companies and
government policymakers as to where what
strategies might be most effective to
drive the cost down further and as I
mentioned in the example of the electric
vehicle now that we can measure progress
against people’s needs we can also
provide some insight on how to develop
this technology further both of these
are technologies that could reduce
greenhouse gas emissions and there are
many other technologies that that could
could help with that as well where we
can apply this approach so what does all
of this mean well as I as I’ve shown
through these two examples we actually
can measure progress toward
technological progress toward these
broad societal goals and we can explain
where it comes from this will help us
make better decisions about how to
invest our time and money in developing
technologies that will help us solve
problems like climate change and maybe
other problems as well in the future so
this doesn’t have to be guesswork as an
engineer as a company as a government
policy maker as a private individual
this doesn’t have to be guesswork we can
we can actually say something about how
to choose technologies and develop
technologies to get where we
want to go we’re really good at
developing technologies that we don’t
need now the thought I’d like to leave
you with is that now let’s use
science-based data-driven approach to
develop technologies to make
technologies to adopt technologies to
solve the big problems that matter most
thank you
[Applause]
Please follow and like us: