February 29, 2024 • 38 mins
Why are they using more electric cars in Norway than in the US? What are the underlying mechanisms that drive technological change, and how can we influence them? What is ‘soft technology’ and what role can it play in lowering carbon emissions? In the season 2 finale of Data Nation, IDSS professor Jessika Trancik explains how her lab attempts to measure and model a technology’s real or potential impact, beneficial or harmful, on people, the environment, society — and on critical outcomes like mitigating climate change.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:09):
The newest term in AI that we're hearing over and over again is soft technology,
so let's find out what this really means from the leading expert Jessica Trancik,
on Data Nation from MIT's Institute of Data, Systems and Society.
I'm Liberty Vittert and today my co-host Munther Dahleh, the founding director of
(00:32):
MIT’s Institute for Data, Systems, and Society and I are speaking with Jessica Trancik. As an
MIT professor at the Institute for Data, Systems, and Society, Jessica has seen
first-hand how AI tools can impact research business and careers. Her research examines
the dynamic cost performance and environmental impacts of energy systems to inform climate
(00:58):
policy and accelerate beneficial and equitable technology innovation. So let's get started.
Munther (01:05):
Well thank you first, Jessica for taking the time to do this and I'm going to
just jump into you know some of the core area of research that you've been doing,
and in one area is to try to understand the important factors that impact the technology,
the reduction of cost and the adoption factors and so forth. One area you've
(01:26):
been talking about recently is soft technologies and so maybe you can tell
us a little bit more about what you mean by soft technologies and how those were identified to be
critical factors in large infrastructure, large renewable type energy and so forth.
Jessika (01:43):
Yeah, sure. Soft technology is something that I define as codified knowledge that doesn't
take on a physical form. So the definition of technology that I apply in my research is any
kind of transformation of raw materials that provides a useful service. So those
(02:06):
raw materials can be physical, they can be mined materials, for example, but they can also be the
information processing power of the human brain, they can be biological in nature. So this is the
broad definition of technology that I apply in my research and some of that technology takes on a
physical form and other technology doesn't, it's basically information-based. So software is an
(02:31):
example of soft technology but we have other forms of codified knowledge that fall into that category
of soft technology. So these can be checklists, these can be processes that are codified in some
way through best practices, for example. These can be institutional designs that are codified,
(02:52):
can be repeated and essentially reproduced. So that's how we define soft technology,
and the reason I became very interested in this with respect to climate change and the energy
transition is that what we see is that the costs associated with soft technology are becoming a
larger fraction of the overall cost of some really critical clean energy technologies. We're not
(03:17):
seeing as fast improvement in soft technology as we are in hardware. Of course, software
is one type of soft technology that improves rapidly, but there are many others that matter
for the clean energy transition. So these can be, for example, processes of installing different
technologies like building a nuclear fission reactor or installing a solar energy plant.
Liberty (03:42):
This might sound like a really dumb question. It's pretty easy for me to
imagine what technology is - what is innovation that's not technology? What would be an example?
Jessika (03:52):
So if we define innovation as some improvement in technology, so some
improvement in that codified knowledge that is then implemented right? You also have invention,
so you can invent something, some new some new way of doing something and you can have technologies
that are invented that can apply to technology, but then those technologies aren't always adopted
(04:17):
they don't always make it outside of the lab. So those would be inventions, and then once they are
adopted they become innovations. We could think of examples of innovation, I suppose, that are
not focused on technology. So you would maybe have like a one-off improvement that's adopted
maybe just from one period of time to the next, so maybe, I don't know, we're organizing a one-day
(04:43):
conference and you know the morning is organized in some way and then through some process of
innovation the afternoon is organized in some other way, but it's not codified, then it wouldn't
fit under that definition of technology. Now, I'm not saying my definition of technology is the one
that you want to use in all cases, but because I'm interested in many societal impacts of technology
(05:07):
including effects on the environment, humans and sustainability-related concerns, then it's very
useful to use this broad definition so that we can study all of these different transformations
of raw material inputs. So on this planet, we have access to biological, geological materials, people
(05:29):
are on trying to access resources from other planets as well, and then the question is what
are we doing with those and those transformations that then become codified and are repeatable fall
under that definition of technology. So, the theories and the models and the different insights
that we work on developing in my research group apply often generally to many forms of technology,
(05:57):
and then we look at specific cases and focus a lot on climate change mitigation.
Munther (06:02):
So maybe I'll just ask one question for the technical
part of our audience here which is - trying to understand what impacts the outcome in climate
change or energy changes and so forth, there's so many different parameters,
so many different factors at play in and somehow you can isolate, maybe,
(06:24):
the impact of soft technology or soft skills or so forth. So the question is - what techniques
and methodologies do you use to actually extract this impact and causal effect of one particular
parameter or change on the outcome of a problem that is not very easy to experiment with?
Jessika (06:43):
Exactly. There's kind of two classes of models that we develop in my group, and one
would be data-driven models. So there we're not trying to get at causality, those would be used
for things like making forecasts. You know, you always want to, of course, consider uncertainty
in those forecasts but then we also work on mechanistic models. So those mechanistic models
(07:07):
are really the ones that we've developed to get at some of these questions about what is driving
technological change, what are the underlying mechanisms and one of the approaches that we've
developed is to look at the underlying mechanisms of technology improvement. So a key step there
is to start with things that you can measure. So whatever type of performance you're interested in,
(07:32):
often in the case of climate change mitigation it's cost per unit service, you look at all of the
different inputs to arriving at that cost per unit service. So you typically have price variables,
and then you have what are called usage ratios. So the amount of whatever input it is that you
(07:54):
need to provide a certain service, and then the price variables would be the cost that you pay
the price that you pay as a consume, the consumer here would be, for example, a manufacturer for
that input material. And then what you do is you can write out an equation that relates those input
variables to the cost-per-unit service or other performance realized at a given point in time,
(08:21):
you can look at then how those variables change over time and from there you can estimate the
effects of different mechanisms. So this is what we use to get at this question of soft technology.
The initial observation was that in the case of many clean energy technologies it looks like soft
(08:42):
technology is the part that isn't improving as much and it's coming to represent a larger and
larger fraction of the cost over time, so then the question is why is that? And it turns out to be
an interesting relationship between hardware variables and the soft technology variables
and so we had to really get into those details to understand what was going on.
Liberty (09:04):
To get really into one of the big issues that I know you've been looking at
which is climate change, how is it that we really evaluate the role of innovation and
soft technology in particular in mitigating climate change? Or also, how do we evaluate
the role of what it's doing furthering climate change? How do we evaluate it on both sides of
(09:27):
what it's doing to help mitigate and what it's also doing to further climate change in bad ways?
Jessika (09:33):
Yeah, so usually we kind of go back to the basics and we know that technology has
supported a growth and an improvement in living standards in many places, and if you look around
the world you know there are different wealth levels, there's different access to technology,
but overall since the onset of industrialization globally we've adopted more and more technology to
(09:58):
do various things, to do physical work, to augment our own human power that we have in our own
physical bodies, but also to process information. So a lot of economic activity today is supported
by technology. Now, that technology requires some energy resource to drive it, right, so you always
(10:22):
require some energy to process information, to store information and certainly to do physical
work. Around the late 1800s, we started to use fossil fuels for this and then that's
leading to this buildup of greenhouse gases in the atmosphere which is what is driving human-caused
(10:43):
climate change, so that's the concern. Then the question is what do you do about it? Do you stop
using technology? Well, many people today would say ‘well, no I'm not willing to give up all of
the things that I do.’ In the US, for example, an individual uses about 100 times the power of their
own physical body through all of their activities. So that's taking into account economy-wide
(11:09):
activities and so forth. They're using a lot of additional power outside of what they can generate
that - if you want to keep doing those things, that has to come from somewhere. There's a lot
of room to improve in terms of efficiency and probably do just as much with less energy, but
you can't get all the way back to just subsistence living and basically using our own physical power
(11:32):
and still support the standard of living that many people enjoy today and many people aspire to enjoy
in the future. It’s important to note that in many countries in the world they're not using nearly
as much energy and they're not consuming nearly as much fossil fuels per capita. So there's a need to
(11:53):
rely on other energy resources to address climate change and bring those greenhouse gas emissions
down while still supporting a high standard of living and supporting human well-being.
So that's really the challenge, and that's how technology factors in. We can't address climate
change simply by our own individual choices and our own behaviors. Certainly we can choose - let
(12:16):
me revise that a bit, we can choose different technologies, but technology really has to be a
key part of the solution. Now it's important to note that technology is not perfect,
right? Any technology is going to have some negative environmental and probably societal
consequences. So environmental consequences because you're accessing these raw materials,
(12:39):
you're consuming them, you're transforming them. That's going to do something that's probably going
to bring about some harm to the environment and then societal consequences often come in the form
of inequitable outcomes. So some people benefit from this technology, others actually lose,
right? So there's forms of pollution that disproportionately affect less wealthy individuals
(13:04):
and they're not the ones that are consuming as much of that technology-based service.
Munther (13:10):
Just to segue from that a little bit, Jessica, in an area that you've done extensive
work is electric vehicles, and I was just reading an article in The Economist just a
few weeks ago that the United States is lagging behind in the adoption of electric vehicles from
the expectations where we thought the United States would be at. What are your thoughts
(13:33):
about the adoption aspect of electric vehicles, at least in the United States, but maybe worldwide?
Jessika (13:40):
Actually, the adoption of electric vehicles is growing rapidly around the world
and different countries are further along those growth curves, but generally we can approximate
those growth curves by exponential growth curves. Now, that doesn't continue forever, at some point
it levels off when you reach market saturation. The US is earlier on than some other countries.
(14:05):
If we look at Norway, most new vehicles sold are electric - fully electric - vehicles in Norway
today. The US isn't there yet, so in 2022, about 8 percent of vehicles purchased in the US were
electric, but that still represents a significant increase relative to 2021, and then in the last
(14:29):
quarter, if we include hybrids in there in the last quarter of this year, we've seen increases
in the adoption of electric vehicles reaching up to close to 20 percent, that's including hybrids
as well. So you have about half of that from fully electric vehicles and then about very roughly half
from hybrids, and that's just one-quarter of data, so you really have to look at the
(14:52):
longer term trends. We'll see what 2023 comes out at in terms of fully electric vehicles.
Adoption is growing, but it's still a small fraction, and I think in terms of what needs
to be done to support the adoption of electric vehicles which, by the way,
is one way to reduce greenhouse gas emissions from transportation,
(15:15):
even today with the electricity supply that we have in this country and in most places,
you get very significant reductions in greenhouse gas emissions. Upwards of 30 percent, in many
places 40 percent, even with the electricity supply mix that we have today, and that takes
into account all of the emissions incurred in producing the cars and so forth. But, I mean,
(15:37):
a couple things have to happen. Right now we're seeing mostly wealthier individuals are able to
access these electric vehicles, it's mostly people that are using them as second cars. Expanding
charging infrastructure is really important and in different locations. So residential
charging infrastructure for people that don't have off-street parking in urban areas, for example,
(15:59):
is really important. Fast charging along highways, and really being strategic about where to place
chargers is key, but then also offering more vehicles that are lower in price. So less focus
solely on the luxury cars, and we are seeing this happen, but it hasn't been fully realized
(16:21):
yet. Access to financing is also important, because although you can save over in terms
of the total cost of ownership of these cars, you often have to pay more up-front for these cars.
Liberty (16:32):
You know usually we listen to Munther’s rants on this podcast but now you're going to get
one of mine. So I have a Tesla and I have to get rid of it. I love it, it's wonderful, it's a great
car, but I'm now in West Virginia and to go from West Virginia to DC in a regular car is 5 hours,
to go in my Tesla it’s six-and-a-half because of the charging. It's just not sustainable,
(16:56):
it's just a nightmare with the charging, I love it otherwise. So I'm getting rid
of it. So I'm now dropping out of that eight percent that are rising in terms
of the electric vehicles. So is there - and I'd love to have one - but is there a more
practical solution involved with owning an EV and a conventional car to address daily use issues,
(17:17):
as well as my long-distance driving issues? Or if we only have one car, what do we do? What are what
are sort of the practical examples that people can at least do now until we do have fast charging?
Jessika (17:30):
Yeah, so great question, and I think a lot of people have similar questions, you’re
not the only one. And just one quick check on the numbers, that eight percent, by the way, is new
cars being sold. So if we look at the percentage of cars on the road, it's a much lower number. But
(17:51):
yeah, so we've done a lot of work on this and without getting into too many of the details,
thinking about how to offer convenient charging is really important, I think, and it's something
that requires considering people's behaviors in cars and what you want to do is to be able to
develop predictive models so that you know when people tend to stop where and for how long, and
(18:16):
then you expand charging infrastructure based on that knowledge. And one of the key locations for
expanding charging infrastructure and specifically fast chargers would be along highways. So
the example that you presented would be one of your like ‘higher energy’ days is what we call
them. When we look at the overall population, around 95 percent of days for most people
(18:43):
given how they're driving and given very accurate estimates of the energy consumption in their cars,
so that's something we've worked on a lot, about 95 percent of days for the population,
and for most people will be covered on a single charge. So an overnight charge, when we look at
(19:03):
the lower-cost electric vehicles. So the cheaper vehicles, you just have to charge them overnight,
if they're ready to go when you access them in the morning on 95 - and for many people it's
higher than that - percent of days, you don't have to think about recharging. So, you don't have to
stop at gas stations, you're not losing time in that way. That's one nice thing about electric
(19:26):
vehicle charging is that it can fit into people's activities if you're really strategic about where
to place those chargers. Now, on those high energy days, on that five percent of days, maybe you have
to stop for an additional 15 minutes, 20 minutes for most. I think the delay you were describing
(19:48):
sounds long, so I'm not sure what the power of those chargers were that you were accessing,
maybe they were the lower power chargers, but so - it's a different experience obviously, those few
long trips you take per year, are you willing to delay your trip a little bit in order to then save
(20:11):
time the rest of the year? That's one question, and is it just about getting used to it? I mean,
I think it's not just about that because it's also about expanding that charging infrastructure, and
often those fast chargers are not available. So, in your case it could be that along that route,
those fast chargers weren’t available and then for some people that are driving a lot,
they have more high energy days, those would be outliers in the population, but we have to
(20:37):
think about solutions in those cases as well. So I think it's a great question, and really people
are going to adopt the cars that work for them. So it's about providing information, but it's
about providing access to really convenient charging infrastructure. It's a process,
it's not going to happen overnight. We do see in countries that are further along those exponential
(20:59):
growth curves that the markets gain momentum. People start to factor in also the resale value
of their cars. If the transition is happening, are people going to want that fully electric car
some years down the road - five years, six, seven, eight, ten years, whenever I want to sell my car,
and then that's factoring in, I think to people's decisions as well. But for personal cars certainly
(21:23):
in this country, it's not a one-size-fits-all solution. So, we really need to consider
lots of different ways in which people use their cars and preferences and all of that as well.
Liberty (21:34):
Just to follow up on that really quickly, is there data on why the adoption has been so much
greater in Europe and other places than the US? Is it the infrastructure that's done it,
or is there some other factor involved besides just infrastructure that's created that adoption?
Jessika (21:49):
It's a combination of income levels, incentives from the government. So in Norway,
that example that I mentioned a minute ago, there were very significant incentives for Norwegians to
purchase electric cars, and then what you see with charging infrastructure is that that's expanding,
and it's usually just a little bit ahead of the adoption levels, as we could expect, right? So
(22:13):
what people need, for the kinds of people that are adopting cars early on, it does tend to be
wealthier individuals. Which, by the way, is a huge problem if you're thinking about where are
government incentives going, where are dollars going. I mean, that's something that is very
much the focus of policy efforts in many countries including in the US is to think about how to make
(22:35):
sure those dollars are not disproportionately helping certain individuals, and sort of
making those benefits more evenly and more equitably distributed, that's really important.
Munther (22:47):
I want to go back to something you said Jessica, and I don't want to overlook it,
and that is the fact that electric cars are still contributing to mitigation of pollution,
even if electricity is generated by fossil fuels. Can you explain that because it
does sound a little counterintuitive, that you're creating more electricity
(23:11):
through kind of burning coal or gas or something and so you're - it seems
like a zero sum game at best. So how come that we're actually benefiting?
Jessika (23:22):
If we look at the different fossil fuels, they have very different what are called carbon
intensities. So they emit more or less carbon per unit energy. They all have significant
emissions associated with them, but natural gas has half of the carbon intensity of coal,
for example. Now that reduction of half isn't enough to get us to net zero, which is where you
(23:46):
need to go in order to stay within global climate targets, but they're not all the same, and so when
you look at a functioning energy system like in the US the carbon intensity of electricity
is around that of natural gas. So you have some coal, you have some natural gas, you have hydro,
you have renewables, you have nuclear fission, and so all of that gives us a certain carbon
(24:11):
intensity. So if we look at the US as a whole, those numbers that I mentioned a minute ago,
those 30 and 40, even 50 percent reductions, it depends on the vehicle model that you get
from switching to a battery electric vehicle are based on those carbon intensities of electricity
versus the carbon intensity of oil, and then all of the losses as you go from this primary
(24:36):
fuel inputs to electricity, the masses of coal, the amount of natural gas, etc. The inputs to
providing that tractive energy that moves the car forward in an internal combustion engine if you're
using gasoline or diesel. So you have different losses along the way and that is what gives us
(24:58):
that answer. So I think sometimes people think about fossil fuels and then they
think about the highest intensity fuel which is coal, but only in a few cases,
like if we look at the US there's some regions of the US where you use a lot of coal,
but even in those cases like in certain parts of the Midwest, you still have emission savings in
(25:25):
switching to a hybrid or a fully battery electric vehicle. The savings of going to all electricity
in those few cases which are outlier cases are not as significant. So you get that significant
savings in going to a hybrid or a battery electric vehicle, but then using only electricity you don't
(25:46):
get as much savings. So that's why you know it's important to look at individual locations,
the energy mix and in my group we've published a website called carboncounter.com
which is informational, people can go there, look up their region. It's focused on the US,
we also have one for Europe, and they can go there, look up their regions, see what the
(26:09):
emission savings would be for different vehicle models that they might be interested in. So that
takes into account the full life cycle emissions, and one thing people often wonder about is the
impact of the batteries. One issue that has been very much discussed is this question of how good
(26:30):
or bad are batteries when you consider the fact that switching to an electric vehicle means that
you need a larger battery. What does that do in terms of contributing to those overall emissions?
And the numbers that I mentioned before take into account the emissions of producing the battery.
So you still get those significant savings. So although you have to mine materials to build that
(26:55):
bigger battery, you still overall get savings. The other thing that I'll mention is that the input
to building that bigger battery, those inputs mean that we have to mine a bit more material,
but that additional material is still going to be a small percentage of the overall material
(27:19):
that we're mining economy-wide. We only have to kind of look around the rooms that we're sitting
in and kind of look at all of the different materials, you might have plastics in your room,
those are often petroleum-based and so they're coming from extracted crude oil. You probably
have metals in your room, so those are all mined materials. So if you do a check and say ‘okay,
(27:44):
look at all these extracted materials that I have around’ and then look outside at the cars and how
big is the car, how big is the battery? You can see that the battery and the mined materials that
go into the battery in terms of all of the impacts - and it's not just greenhouse gas emissions there
are many different societal environmental impacts from extractive industries - but those batteries
(28:06):
are just a small part of the economy-wide impact. So I think, on the one hand, there's been a lot of
focus on mining for batteries and my hope would be that that would put more focus on the overall, the
bigger problem which is mining overall. Batteries are just a very small part of that and if you look
(28:26):
at the full set of impacts, compare an internal combustion engine vehicle and a battery electric
vehicle, you see that there are still significant benefits from the battery electric vehicle. But,
of course no technology is perfect and there are always going to be some negative impacts.
Liberty (28:42):
You've talked about, sort of, the virtuous cycle of innovation and adoption
leading to cost reduction. Do we have precedents that we can highlight to sort of raise awareness
and drive more adoption of alternative sources of energy? Are there use cases
where we can be like – look, this works!
Jessika (29:00):
Yeah, this is something that I think is really interesting, and it's something that
actually goes back to one point that Munther and I were talking about a minute ago,
which is that we can actually look back in time and we can say ‘okay, why did solar energy costs
fall by more than 99 percent? Where did that come from?’ If you start with something measurable, you
(29:26):
start with the inputs to that device, you consider the physics, you bring all those variables I was
talking about into the equation and you can actually see if you predict the answer. If
your equation is representing all of the inputs, because you see, do you actually predict at some
point in the past what the cost was, for example. So, you can start with that and then see why did
(29:49):
this technology improve? And what we learn when we do that is that it was a combination of government
policies that funded, did research and development in government funded labs, but also government
policies that stimulated market growth, that kicked off a lot of private sector innovation.
We can actually estimate the effects of these different kinds of policies, so we can actually
(30:14):
then study these effects quantitatively which is really interesting. But I think what's important
is that you can start to get at causality, and you can see what's driving what, and it's very
clear because we've started with the device and the physics and what's actually changing,
you can go all the way up to, then, what companies did and then the policies that help drive that.
(30:37):
So this virtuous cycle is not just some nice kind of fluffy term, we actually see in the
data that it's very clear that government policy kickstarted a lot of innovation, and then markets
began to grow and now there's a lot of momentum in these markets, and so this is the virtuous
(30:58):
cycle. Can this apply to other technologies? Well, we're seeing this to some extent for lithium-ion
batteries, we've looked at that as well. Although in that case it's not all policy driven,
because you had a lot of development that came through incentives to support the adoption of
better personal devices, digital devices there was incentive already without policy to improve
(31:23):
lithium-ion batteries for that, but there's also a policy component there. And you can
really look at any technology in this way, either retrospectively or prospectively,
and obviously what we care about is what could happen in the future. And for both soft technology
and for hardware, you see that there is a lot of potential to be strategic about what mechanisms
(31:46):
you're investing in, what strategies companies pursue, what engineering approaches people take,
for example, in the field and on construction sites, and what policies are supported. So we
talked before about soft technology, I think there is room to be more deliberate about improving soft
technology by really understanding what driving good performance or not-so-good performance.
Munther (32:13):
So this is really taking me to, sort of, the whole question of understanding the
system-level problem with climate change and then the impact of technology and so forth,
and I think part of the confusion that you get in being an observer and not being a researcher
in that area is that the different articles you read give you a different sort of perspective. You
(32:38):
read about electrification for transportation and then you read something about the fact that ‘well,
it's going to be very difficult for putting that into ships and air and so forth as opposed to
land transportation’, and then you sort of walk away with the confusion that - to what
extent one technology is able to help a particular sector, what's your perspective on this? I mean,
(33:00):
obviously we can be hitting in all different technologies and we can be hitting for hydrogen
and looking for different modes of renewable, but from an economics perspective, right, there's an
investment that has to go somewhere, and over time - infinite time, maybe - we'll figure it all out,
but tomorrow where should we be putting our money and what are the factors? How do we
(33:21):
think about the system-level questions? It's just - it's daunting a little bit.
Jessika (33:25):
Yeah, you know we've all heard people say ‘well let's invest in all of the above,
just invest in everything’ but if you're faced with a decision, you have to make choices. We
always have limited time, we have limited money, we have to decide how much to allocate across
different approaches, different technologies we might be pursuing. So, you know, one very useful
(33:46):
way of thinking about this is to apply portfolio theory. We can draw some concepts from finance,
there are some differences when we look at technology portfolios, but what I always say
is that although forecasting is uncomfortable, we never know exactly what's going to happen in the
future, if we take information from data-driven forecasts and also understanding of the underlying
(34:12):
mechanisms of what can improve technology more rapidly, more slowly, which technologies improve
more slowly, more quickly, then we can do better than random guesses and making forecasts. So the
way to solve this technology portfolio problem that you're talking about out is to forecast,
although it's uncomfortable. You always want to understand and ideally forecast uncertainty
(34:37):
associated with your prediction, but you also want to understand if youcan't quantify that
uncertainty. That's going to push you to more diversification, that's going to push you closer
to that ‘all of the above’ solution, but you're never going to be right there at the ‘all of the
above’ solution and economy-wide, we now know some things are working, some things are not as
(34:59):
far along in terms of certain technologies. Like if we take shipping fuels for example, there's a
lot of uncertainty about will it be ammonia? Will it be methanol with carbon capture and storage?
What portion of ships can we electrify, etc. That's an area where you want to diversify more,
but this portfolio theory, and thinking about the problem as a technology portfolio problem I think
(35:22):
is very useful for informing our decisions. We shouldn't be aiming for perfection, but
we can do better than randomly guessing, and the benefits of that can be huge, because if you're
thinking about technologies changing by orders of magnitude sometimes over a couple decades,
and you forecast that they're not going to change at all or you didn't take into account past
(35:44):
information and informing your decisions, you will be very far off from an optimal investment. Now,
you might not have been able to perfectly forecast what happened but you could probably do better,
and maybe you’re one order of magnitude closer, the returns to that are huge and the returns
here are not just financial returns, they're of course returns in terms of societal well-being and
(36:07):
our ability to actually address climate change.Munther: But has this been done? These forecasting
and the portfolio optimization problem?
Yeah, it's happening more and more. So what I would say is,
if we look back at the forecast by major institutions and different groups that that are
(36:28):
in the business of forecasting and data analysis and also groups that rely on expert elicitation,
we've seen that many forecasts of technological change in the past have been very far off. They've
essentially not been very accurate at all, they've been wrong. But what we are starting to see now
(36:49):
is a movement toward more data-driven approaches, more systematic approaches and not purely relying
on subjective inputs from experts. You can also bring those expert inputs into this whole process,
and so we actually are seeing - Munther, it's interesting - we are seeing a movement toward
(37:10):
using these methods, these more formal methods and making forecasts and people do talk about
applying technology portfolio theory, I would say that there's still a ways to go there. I think
as humans we're often very driven to follow our intuition and even if we do the analysis,
(37:33):
we see some number it's like ‘oh no, this can't possibly…’ for whatever reason, it's
forecasting too fast technological improvement. We know there's what's called a pessimism bias
in some cases for humans in terms of technology improvement. So there can be some discomfort,
there's still a push to follow intuition. I'm not saying throw that out, but we are seeing
(37:57):
these methods, these more formal methods gain more attraction.
Liberty (38:02):
This has been fabulous, thank you so much.
Jessika (38:05):
Thank you so much.
Liberty (38:08):
Thank you for listening to this month's episode of Data Nation. You can get
more information and listen to previous episodes at our website, idss.mit.edu,
or follow us on Twitter and Instagram at MIT IDSS. If you liked this podcast,
please don't forget to leave us a review on Spotify, Apple or wherever you get your podcasts.
(38:29):
Thank you for listening to Data Nation, from the MIT Institute for Data, Systems, and Society.
The newest term in AI that we're hearing over and over again is soft technology,
so let's find out what this really means from the leading expert Jessica Trancik,
on Data Nation from MIT's Institute of Data, Systems and Society.
I'm Liberty Vittert and today my co-host Munther Dahleh, the founding director of
(00:32):
MIT’s Institute for Data, Systems, and Society and I are speaking with Jessica Trancik. As an
MIT professor at the Institute for Data, Systems, and Society, Jessica has seen
first-hand how AI tools can impact research business and careers. Her research examines
the dynamic cost performance and environmental impacts of energy systems to inform climate
(00:58):
policy and accelerate beneficial and equitable technology innovation. So let's get started.
Munther (01:05):
Well thank you first, Jessica for taking the time to do this and I'm going to
just jump into you know some of the core area of research that you've been doing,
and in one area is to try to understand the important factors that impact the technology,
the reduction of cost and the adoption factors and so forth. One area you've
(01:26):
been talking about recently is soft technologies and so maybe you can tell
us a little bit more about what you mean by soft technologies and how those were identified to be
critical factors in large infrastructure, large renewable type energy and so forth.
Jessika (01:43):
Yeah, sure. Soft technology is something that I define as codified knowledge that doesn't
take on a physical form. So the definition of technology that I apply in my research is any
kind of transformation of raw materials that provides a useful service. So those
(02:06):
raw materials can be physical, they can be mined materials, for example, but they can also be the
information processing power of the human brain, they can be biological in nature. So this is the
broad definition of technology that I apply in my research and some of that technology takes on a
physical form and other technology doesn't, it's basically information-based. So software is an
(02:31):
example of soft technology but we have other forms of codified knowledge that fall into that category
of soft technology. So these can be checklists, these can be processes that are codified in some
way through best practices, for example. These can be institutional designs that are codified,
(02:52):
can be repeated and essentially reproduced. So that's how we define soft technology,
and the reason I became very interested in this with respect to climate change and the energy
transition is that what we see is that the costs associated with soft technology are becoming a
larger fraction of the overall cost of some really critical clean energy technologies. We're not
(03:17):
seeing as fast improvement in soft technology as we are in hardware. Of course, software
is one type of soft technology that improves rapidly, but there are many others that matter
for the clean energy transition. So these can be, for example, processes of installing different
technologies like building a nuclear fission reactor or installing a solar energy plant.
Liberty (03:42):
This might sound like a really dumb question. It's pretty easy for me to
imagine what technology is - what is innovation that's not technology? What would be an example?
Jessika (03:52):
So if we define innovation as some improvement in technology, so some
improvement in that codified knowledge that is then implemented right? You also have invention,
so you can invent something, some new some new way of doing something and you can have technologies
that are invented that can apply to technology, but then those technologies aren't always adopted
(04:17):
they don't always make it outside of the lab. So those would be inventions, and then once they are
adopted they become innovations. We could think of examples of innovation, I suppose, that are
not focused on technology. So you would maybe have like a one-off improvement that's adopted
maybe just from one period of time to the next, so maybe, I don't know, we're organizing a one-day
(04:43):
conference and you know the morning is organized in some way and then through some process of
innovation the afternoon is organized in some other way, but it's not codified, then it wouldn't
fit under that definition of technology. Now, I'm not saying my definition of technology is the one
that you want to use in all cases, but because I'm interested in many societal impacts of technology
(05:07):
including effects on the environment, humans and sustainability-related concerns, then it's very
useful to use this broad definition so that we can study all of these different transformations
of raw material inputs. So on this planet, we have access to biological, geological materials, people
(05:29):
are on trying to access resources from other planets as well, and then the question is what
are we doing with those and those transformations that then become codified and are repeatable fall
under that definition of technology. So, the theories and the models and the different insights
that we work on developing in my research group apply often generally to many forms of technology,
(05:57):
and then we look at specific cases and focus a lot on climate change mitigation.
Munther (06:02):
So maybe I'll just ask one question for the technical
part of our audience here which is - trying to understand what impacts the outcome in climate
change or energy changes and so forth, there's so many different parameters,
so many different factors at play in and somehow you can isolate, maybe,
(06:24):
the impact of soft technology or soft skills or so forth. So the question is - what techniques
and methodologies do you use to actually extract this impact and causal effect of one particular
parameter or change on the outcome of a problem that is not very easy to experiment with?
Jessika (06:43):
Exactly. There's kind of two classes of models that we develop in my group, and one
would be data-driven models. So there we're not trying to get at causality, those would be used
for things like making forecasts. You know, you always want to, of course, consider uncertainty
in those forecasts but then we also work on mechanistic models. So those mechanistic models
(07:07):
are really the ones that we've developed to get at some of these questions about what is driving
technological change, what are the underlying mechanisms and one of the approaches that we've
developed is to look at the underlying mechanisms of technology improvement. So a key step there
is to start with things that you can measure. So whatever type of performance you're interested in,
(07:32):
often in the case of climate change mitigation it's cost per unit service, you look at all of the
different inputs to arriving at that cost per unit service. So you typically have price variables,
and then you have what are called usage ratios. So the amount of whatever input it is that you
(07:54):
need to provide a certain service, and then the price variables would be the cost that you pay
the price that you pay as a consume, the consumer here would be, for example, a manufacturer for
that input material. And then what you do is you can write out an equation that relates those input
variables to the cost-per-unit service or other performance realized at a given point in time,
(08:21):
you can look at then how those variables change over time and from there you can estimate the
effects of different mechanisms. So this is what we use to get at this question of soft technology.
The initial observation was that in the case of many clean energy technologies it looks like soft
(08:42):
technology is the part that isn't improving as much and it's coming to represent a larger and
larger fraction of the cost over time, so then the question is why is that? And it turns out to be
an interesting relationship between hardware variables and the soft technology variables
and so we had to really get into those details to understand what was going on.
Liberty (09:04):
To get really into one of the big issues that I know you've been looking at
which is climate change, how is it that we really evaluate the role of innovation and
soft technology in particular in mitigating climate change? Or also, how do we evaluate
the role of what it's doing furthering climate change? How do we evaluate it on both sides of
(09:27):
what it's doing to help mitigate and what it's also doing to further climate change in bad ways?
Jessika (09:33):
Yeah, so usually we kind of go back to the basics and we know that technology has
supported a growth and an improvement in living standards in many places, and if you look around
the world you know there are different wealth levels, there's different access to technology,
but overall since the onset of industrialization globally we've adopted more and more technology to
(09:58):
do various things, to do physical work, to augment our own human power that we have in our own
physical bodies, but also to process information. So a lot of economic activity today is supported
by technology. Now, that technology requires some energy resource to drive it, right, so you always
(10:22):
require some energy to process information, to store information and certainly to do physical
work. Around the late 1800s, we started to use fossil fuels for this and then that's
leading to this buildup of greenhouse gases in the atmosphere which is what is driving human-caused
(10:43):
climate change, so that's the concern. Then the question is what do you do about it? Do you stop
using technology? Well, many people today would say ‘well, no I'm not willing to give up all of
the things that I do.’ In the US, for example, an individual uses about 100 times the power of their
own physical body through all of their activities. So that's taking into account economy-wide
(11:09):
activities and so forth. They're using a lot of additional power outside of what they can generate
that - if you want to keep doing those things, that has to come from somewhere. There's a lot
of room to improve in terms of efficiency and probably do just as much with less energy, but
you can't get all the way back to just subsistence living and basically using our own physical power
(11:32):
and still support the standard of living that many people enjoy today and many people aspire to enjoy
in the future. It’s important to note that in many countries in the world they're not using nearly
as much energy and they're not consuming nearly as much fossil fuels per capita. So there's a need to
(11:53):
rely on other energy resources to address climate change and bring those greenhouse gas emissions
down while still supporting a high standard of living and supporting human well-being.
So that's really the challenge, and that's how technology factors in. We can't address climate
change simply by our own individual choices and our own behaviors. Certainly we can choose - let
(12:16):
me revise that a bit, we can choose different technologies, but technology really has to be a
key part of the solution. Now it's important to note that technology is not perfect,
right? Any technology is going to have some negative environmental and probably societal
consequences. So environmental consequences because you're accessing these raw materials,
(12:39):
you're consuming them, you're transforming them. That's going to do something that's probably going
to bring about some harm to the environment and then societal consequences often come in the form
of inequitable outcomes. So some people benefit from this technology, others actually lose,
right? So there's forms of pollution that disproportionately affect less wealthy individuals
(13:04):
and they're not the ones that are consuming as much of that technology-based service.
Munther (13:10):
Just to segue from that a little bit, Jessica, in an area that you've done extensive
work is electric vehicles, and I was just reading an article in The Economist just a
few weeks ago that the United States is lagging behind in the adoption of electric vehicles from
the expectations where we thought the United States would be at. What are your thoughts
(13:33):
about the adoption aspect of electric vehicles, at least in the United States, but maybe worldwide?
Jessika (13:40):
Actually, the adoption of electric vehicles is growing rapidly around the world
and different countries are further along those growth curves, but generally we can approximate
those growth curves by exponential growth curves. Now, that doesn't continue forever, at some point
it levels off when you reach market saturation. The US is earlier on than some other countries.
(14:05):
If we look at Norway, most new vehicles sold are electric - fully electric - vehicles in Norway
today. The US isn't there yet, so in 2022, about 8 percent of vehicles purchased in the US were
electric, but that still represents a significant increase relative to 2021, and then in the last
(14:29):
quarter, if we include hybrids in there in the last quarter of this year, we've seen increases
in the adoption of electric vehicles reaching up to close to 20 percent, that's including hybrids
as well. So you have about half of that from fully electric vehicles and then about very roughly half
from hybrids, and that's just one-quarter of data, so you really have to look at the
(14:52):
longer term trends. We'll see what 2023 comes out at in terms of fully electric vehicles.
Adoption is growing, but it's still a small fraction, and I think in terms of what needs
to be done to support the adoption of electric vehicles which, by the way,
is one way to reduce greenhouse gas emissions from transportation,
(15:15):
even today with the electricity supply that we have in this country and in most places,
you get very significant reductions in greenhouse gas emissions. Upwards of 30 percent, in many
places 40 percent, even with the electricity supply mix that we have today, and that takes
into account all of the emissions incurred in producing the cars and so forth. But, I mean,
(15:37):
a couple things have to happen. Right now we're seeing mostly wealthier individuals are able to
access these electric vehicles, it's mostly people that are using them as second cars. Expanding
charging infrastructure is really important and in different locations. So residential
charging infrastructure for people that don't have off-street parking in urban areas, for example,
(15:59):
is really important. Fast charging along highways, and really being strategic about where to place
chargers is key, but then also offering more vehicles that are lower in price. So less focus
solely on the luxury cars, and we are seeing this happen, but it hasn't been fully realized
(16:21):
yet. Access to financing is also important, because although you can save over in terms
of the total cost of ownership of these cars, you often have to pay more up-front for these cars.
Liberty (16:32):
You know usually we listen to Munther’s rants on this podcast but now you're going to get
one of mine. So I have a Tesla and I have to get rid of it. I love it, it's wonderful, it's a great
car, but I'm now in West Virginia and to go from West Virginia to DC in a regular car is 5 hours,
to go in my Tesla it’s six-and-a-half because of the charging. It's just not sustainable,
(16:56):
it's just a nightmare with the charging, I love it otherwise. So I'm getting rid
of it. So I'm now dropping out of that eight percent that are rising in terms
of the electric vehicles. So is there - and I'd love to have one - but is there a more
practical solution involved with owning an EV and a conventional car to address daily use issues,
(17:17):
as well as my long-distance driving issues? Or if we only have one car, what do we do? What are what
are sort of the practical examples that people can at least do now until we do have fast charging?
Jessika (17:30):
Yeah, so great question, and I think a lot of people have similar questions, you’re
not the only one. And just one quick check on the numbers, that eight percent, by the way, is new
cars being sold. So if we look at the percentage of cars on the road, it's a much lower number. But
(17:51):
yeah, so we've done a lot of work on this and without getting into too many of the details,
thinking about how to offer convenient charging is really important, I think, and it's something
that requires considering people's behaviors in cars and what you want to do is to be able to
develop predictive models so that you know when people tend to stop where and for how long, and
(18:16):
then you expand charging infrastructure based on that knowledge. And one of the key locations for
expanding charging infrastructure and specifically fast chargers would be along highways. So
the example that you presented would be one of your like ‘higher energy’ days is what we call
them. When we look at the overall population, around 95 percent of days for most people
(18:43):
given how they're driving and given very accurate estimates of the energy consumption in their cars,
so that's something we've worked on a lot, about 95 percent of days for the population,
and for most people will be covered on a single charge. So an overnight charge, when we look at
(19:03):
the lower-cost electric vehicles. So the cheaper vehicles, you just have to charge them overnight,
if they're ready to go when you access them in the morning on 95 - and for many people it's
higher than that - percent of days, you don't have to think about recharging. So, you don't have to
stop at gas stations, you're not losing time in that way. That's one nice thing about electric
(19:26):
vehicle charging is that it can fit into people's activities if you're really strategic about where
to place those chargers. Now, on those high energy days, on that five percent of days, maybe you have
to stop for an additional 15 minutes, 20 minutes for most. I think the delay you were describing
(19:48):
sounds long, so I'm not sure what the power of those chargers were that you were accessing,
maybe they were the lower power chargers, but so - it's a different experience obviously, those few
long trips you take per year, are you willing to delay your trip a little bit in order to then save
(20:11):
time the rest of the year? That's one question, and is it just about getting used to it? I mean,
I think it's not just about that because it's also about expanding that charging infrastructure, and
often those fast chargers are not available. So, in your case it could be that along that route,
those fast chargers weren’t available and then for some people that are driving a lot,
they have more high energy days, those would be outliers in the population, but we have to
(20:37):
think about solutions in those cases as well. So I think it's a great question, and really people
are going to adopt the cars that work for them. So it's about providing information, but it's
about providing access to really convenient charging infrastructure. It's a process,
it's not going to happen overnight. We do see in countries that are further along those exponential
(20:59):
growth curves that the markets gain momentum. People start to factor in also the resale value
of their cars. If the transition is happening, are people going to want that fully electric car
some years down the road - five years, six, seven, eight, ten years, whenever I want to sell my car,
and then that's factoring in, I think to people's decisions as well. But for personal cars certainly
(21:23):
in this country, it's not a one-size-fits-all solution. So, we really need to consider
lots of different ways in which people use their cars and preferences and all of that as well.
Liberty (21:34):
Just to follow up on that really quickly, is there data on why the adoption has been so much
greater in Europe and other places than the US? Is it the infrastructure that's done it,
or is there some other factor involved besides just infrastructure that's created that adoption?
Jessika (21:49):
It's a combination of income levels, incentives from the government. So in Norway,
that example that I mentioned a minute ago, there were very significant incentives for Norwegians to
purchase electric cars, and then what you see with charging infrastructure is that that's expanding,
and it's usually just a little bit ahead of the adoption levels, as we could expect, right? So
(22:13):
what people need, for the kinds of people that are adopting cars early on, it does tend to be
wealthier individuals. Which, by the way, is a huge problem if you're thinking about where are
government incentives going, where are dollars going. I mean, that's something that is very
much the focus of policy efforts in many countries including in the US is to think about how to make
(22:35):
sure those dollars are not disproportionately helping certain individuals, and sort of
making those benefits more evenly and more equitably distributed, that's really important.
Munther (22:47):
I want to go back to something you said Jessica, and I don't want to overlook it,
and that is the fact that electric cars are still contributing to mitigation of pollution,
even if electricity is generated by fossil fuels. Can you explain that because it
does sound a little counterintuitive, that you're creating more electricity
(23:11):
through kind of burning coal or gas or something and so you're - it seems
like a zero sum game at best. So how come that we're actually benefiting?
Jessika (23:22):
If we look at the different fossil fuels, they have very different what are called carbon
intensities. So they emit more or less carbon per unit energy. They all have significant
emissions associated with them, but natural gas has half of the carbon intensity of coal,
for example. Now that reduction of half isn't enough to get us to net zero, which is where you
(23:46):
need to go in order to stay within global climate targets, but they're not all the same, and so when
you look at a functioning energy system like in the US the carbon intensity of electricity
is around that of natural gas. So you have some coal, you have some natural gas, you have hydro,
you have renewables, you have nuclear fission, and so all of that gives us a certain carbon
(24:11):
intensity. So if we look at the US as a whole, those numbers that I mentioned a minute ago,
those 30 and 40, even 50 percent reductions, it depends on the vehicle model that you get
from switching to a battery electric vehicle are based on those carbon intensities of electricity
versus the carbon intensity of oil, and then all of the losses as you go from this primary
(24:36):
fuel inputs to electricity, the masses of coal, the amount of natural gas, etc. The inputs to
providing that tractive energy that moves the car forward in an internal combustion engine if you're
using gasoline or diesel. So you have different losses along the way and that is what gives us
(24:58):
that answer. So I think sometimes people think about fossil fuels and then they
think about the highest intensity fuel which is coal, but only in a few cases,
like if we look at the US there's some regions of the US where you use a lot of coal,
but even in those cases like in certain parts of the Midwest, you still have emission savings in
(25:25):
switching to a hybrid or a fully battery electric vehicle. The savings of going to all electricity
in those few cases which are outlier cases are not as significant. So you get that significant
savings in going to a hybrid or a battery electric vehicle, but then using only electricity you don't
(25:46):
get as much savings. So that's why you know it's important to look at individual locations,
the energy mix and in my group we've published a website called carboncounter.com
which is informational, people can go there, look up their region. It's focused on the US,
we also have one for Europe, and they can go there, look up their regions, see what the
(26:09):
emission savings would be for different vehicle models that they might be interested in. So that
takes into account the full life cycle emissions, and one thing people often wonder about is the
impact of the batteries. One issue that has been very much discussed is this question of how good
(26:30):
or bad are batteries when you consider the fact that switching to an electric vehicle means that
you need a larger battery. What does that do in terms of contributing to those overall emissions?
And the numbers that I mentioned before take into account the emissions of producing the battery.
So you still get those significant savings. So although you have to mine materials to build that
(26:55):
bigger battery, you still overall get savings. The other thing that I'll mention is that the input
to building that bigger battery, those inputs mean that we have to mine a bit more material,
but that additional material is still going to be a small percentage of the overall material
(27:19):
that we're mining economy-wide. We only have to kind of look around the rooms that we're sitting
in and kind of look at all of the different materials, you might have plastics in your room,
those are often petroleum-based and so they're coming from extracted crude oil. You probably
have metals in your room, so those are all mined materials. So if you do a check and say ‘okay,
(27:44):
look at all these extracted materials that I have around’ and then look outside at the cars and how
big is the car, how big is the battery? You can see that the battery and the mined materials that
go into the battery in terms of all of the impacts - and it's not just greenhouse gas emissions there
are many different societal environmental impacts from extractive industries - but those batteries
(28:06):
are just a small part of the economy-wide impact. So I think, on the one hand, there's been a lot of
focus on mining for batteries and my hope would be that that would put more focus on the overall, the
bigger problem which is mining overall. Batteries are just a very small part of that and if you look
(28:26):
at the full set of impacts, compare an internal combustion engine vehicle and a battery electric
vehicle, you see that there are still significant benefits from the battery electric vehicle. But,
of course no technology is perfect and there are always going to be some negative impacts.
Liberty (28:42):
You've talked about, sort of, the virtuous cycle of innovation and adoption
leading to cost reduction. Do we have precedents that we can highlight to sort of raise awareness
and drive more adoption of alternative sources of energy? Are there use cases
where we can be like – look, this works!
Jessika (29:00):
Yeah, this is something that I think is really interesting, and it's something that
actually goes back to one point that Munther and I were talking about a minute ago,
which is that we can actually look back in time and we can say ‘okay, why did solar energy costs
fall by more than 99 percent? Where did that come from?’ If you start with something measurable, you
(29:26):
start with the inputs to that device, you consider the physics, you bring all those variables I was
talking about into the equation and you can actually see if you predict the answer. If
your equation is representing all of the inputs, because you see, do you actually predict at some
point in the past what the cost was, for example. So, you can start with that and then see why did
(29:49):
this technology improve? And what we learn when we do that is that it was a combination of government
policies that funded, did research and development in government funded labs, but also government
policies that stimulated market growth, that kicked off a lot of private sector innovation.
We can actually estimate the effects of these different kinds of policies, so we can actually
(30:14):
then study these effects quantitatively which is really interesting. But I think what's important
is that you can start to get at causality, and you can see what's driving what, and it's very
clear because we've started with the device and the physics and what's actually changing,
you can go all the way up to, then, what companies did and then the policies that help drive that.
(30:37):
So this virtuous cycle is not just some nice kind of fluffy term, we actually see in the
data that it's very clear that government policy kickstarted a lot of innovation, and then markets
began to grow and now there's a lot of momentum in these markets, and so this is the virtuous
(30:58):
cycle. Can this apply to other technologies? Well, we're seeing this to some extent for lithium-ion
batteries, we've looked at that as well. Although in that case it's not all policy driven,
because you had a lot of development that came through incentives to support the adoption of
better personal devices, digital devices there was incentive already without policy to improve
(31:23):
lithium-ion batteries for that, but there's also a policy component there. And you can
really look at any technology in this way, either retrospectively or prospectively,
and obviously what we care about is what could happen in the future. And for both soft technology
and for hardware, you see that there is a lot of potential to be strategic about what mechanisms
(31:46):
you're investing in, what strategies companies pursue, what engineering approaches people take,
for example, in the field and on construction sites, and what policies are supported. So we
talked before about soft technology, I think there is room to be more deliberate about improving soft
technology by really understanding what driving good performance or not-so-good performance.
Munther (32:13):
So this is really taking me to, sort of, the whole question of understanding the
system-level problem with climate change and then the impact of technology and so forth,
and I think part of the confusion that you get in being an observer and not being a researcher
in that area is that the different articles you read give you a different sort of perspective. You
(32:38):
read about electrification for transportation and then you read something about the fact that ‘well,
it's going to be very difficult for putting that into ships and air and so forth as opposed to
land transportation’, and then you sort of walk away with the confusion that - to what
extent one technology is able to help a particular sector, what's your perspective on this? I mean,
(33:00):
obviously we can be hitting in all different technologies and we can be hitting for hydrogen
and looking for different modes of renewable, but from an economics perspective, right, there's an
investment that has to go somewhere, and over time - infinite time, maybe - we'll figure it all out,
but tomorrow where should we be putting our money and what are the factors? How do we
(33:21):
think about the system-level questions? It's just - it's daunting a little bit.
Jessika (33:25):
Yeah, you know we've all heard people say ‘well let's invest in all of the above,
just invest in everything’ but if you're faced with a decision, you have to make choices. We
always have limited time, we have limited money, we have to decide how much to allocate across
different approaches, different technologies we might be pursuing. So, you know, one very useful
(33:46):
way of thinking about this is to apply portfolio theory. We can draw some concepts from finance,
there are some differences when we look at technology portfolios, but what I always say
is that although forecasting is uncomfortable, we never know exactly what's going to happen in the
future, if we take information from data-driven forecasts and also understanding of the underlying
(34:12):
mechanisms of what can improve technology more rapidly, more slowly, which technologies improve
more slowly, more quickly, then we can do better than random guesses and making forecasts. So the
way to solve this technology portfolio problem that you're talking about out is to forecast,
although it's uncomfortable. You always want to understand and ideally forecast uncertainty
(34:37):
associated with your prediction, but you also want to understand if youcan't quantify that
uncertainty. That's going to push you to more diversification, that's going to push you closer
to that ‘all of the above’ solution, but you're never going to be right there at the ‘all of the
above’ solution and economy-wide, we now know some things are working, some things are not as
(34:59):
far along in terms of certain technologies. Like if we take shipping fuels for example, there's a
lot of uncertainty about will it be ammonia? Will it be methanol with carbon capture and storage?
What portion of ships can we electrify, etc. That's an area where you want to diversify more,
but this portfolio theory, and thinking about the problem as a technology portfolio problem I think
(35:22):
is very useful for informing our decisions. We shouldn't be aiming for perfection, but
we can do better than randomly guessing, and the benefits of that can be huge, because if you're
thinking about technologies changing by orders of magnitude sometimes over a couple decades,
and you forecast that they're not going to change at all or you didn't take into account past
(35:44):
information and informing your decisions, you will be very far off from an optimal investment. Now,
you might not have been able to perfectly forecast what happened but you could probably do better,
and maybe you’re one order of magnitude closer, the returns to that are huge and the returns
here are not just financial returns, they're of course returns in terms of societal well-being and
(36:07):
our ability to actually address climate change.Munther: But has this been done? These forecasting
and the portfolio optimization problem?
Yeah, it's happening more and more. So what I would say is,
if we look back at the forecast by major institutions and different groups that that are
(36:28):
in the business of forecasting and data analysis and also groups that rely on expert elicitation,
we've seen that many forecasts of technological change in the past have been very far off. They've
essentially not been very accurate at all, they've been wrong. But what we are starting to see now
(36:49):
is a movement toward more data-driven approaches, more systematic approaches and not purely relying
on subjective inputs from experts. You can also bring those expert inputs into this whole process,
and so we actually are seeing - Munther, it's interesting - we are seeing a movement toward
(37:10):
using these methods, these more formal methods and making forecasts and people do talk about
applying technology portfolio theory, I would say that there's still a ways to go there. I think
as humans we're often very driven to follow our intuition and even if we do the analysis,
(37:33):
we see some number it's like ‘oh no, this can't possibly…’ for whatever reason, it's
forecasting too fast technological improvement. We know there's what's called a pessimism bias
in some cases for humans in terms of technology improvement. So there can be some discomfort,
there's still a push to follow intuition. I'm not saying throw that out, but we are seeing
(37:57):
these methods, these more formal methods gain more attraction.
Liberty (38:02):
This has been fabulous, thank you so much.
Jessika (38:05):
Thank you so much.
Liberty (38:08):
Thank you for listening to this month's episode of Data Nation. You can get
more information and listen to previous episodes at our website, idss.mit.edu,
or follow us on Twitter and Instagram at MIT IDSS. If you liked this podcast,
please don't forget to leave us a review on Spotify, Apple or wherever you get your podcasts.
(38:29):
Thank you for listening to Data Nation, from the MIT Institute for Data, Systems, and Society.
Popular Podcasts
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!
The Breakfast Club
The World's Most Dangerous Morning Show, The Breakfast Club, With DJ Envy And Charlamagne Tha God!
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.