I had my Covid vaccination and I’m feeling marginally ropey today. Nothing too bad, just a mild fever and bad sleep. I may not be around for this thread much as a result.
We know that technologies start off expensive and, courtesy of learning effects, get cheaper with increasing volume. This effect, well described by Wright’s Law, is independent of economies of scale.
Theodore Wright’s chart on cost improvements...
Given that many technologies needed for hitting sustainability goals, like carbon-free cement or plant-based protein substitutes for meat, or additive manufacturing, are expensive initially, what kind of policies can we use to kickstart the exponential price decline that learning gives us?
This, in a sense, is what Tesla did. It built expensive cars until it could build them cheaply. (The Model 3 and Y are far cheaper than the Model S & X.)
Bringing down that cost curve is critical - both to hit sustainability goals but also to ensure the benefits of technology are affordable. Or is that lens too simplistic?
What policies or approaches could be enacted to accelerate the decline in the cost of these core netzero products and services?
Cheers,
Azeem
P.S. Short responses with a single point work best — and it works well when you build on each other’s ideas.
I once asked Doyne Farmer, one of the co-authors on this beautiful paper on Wright's Law https://arxiv.org/abs/1703.05979, what was the fundamental difference between the technologies where the law works and where it seems not to. It has not worked in chemicals, and it has not worked in nuclear. Interestingly, he said he thought all the other technologies were related one way or another to printing - to decomposing a process into small reconfigurable units each of which could be recombined to create variety. Weaving, metal stamping, etc are all derivatives of the printing press in this way. As of course is wafer production, solar, software, etc.
So what is the implication of this for accelerationism - identify those processes that can be decomposed in this way and focus on those; don't waste $$ on the undecomposable, like fluid processes or huge custom builds - like nuclear. Coordination through standards can often help this process along - the alphabet on the movable type printing press is a really good example of Wright-Law inducing standardisation.
Beautiful, yes. The connection with miniaturisation is really important. It also connects to energy use and the decentralisability of finished products. Decentralised products (like chips) can scale much faster than big lumpy ones, which means demand (and, hence, demand doubling) can grow faster.
When I have read pop-science articles about these, they have seemed to me to be inherently custom, 1-off projects. So very hard to get dramatic cost reductions. But maybe that is wrong - would love to know more about the details of the process.
Ahh... sounds like "small is beautiful" in action. "[decompose] a process into small reconfigurable units each of which could be recombined to create variety" and "don't waste $$ ... huge custom builds - like nuclear" (and E. F. Schumacher would agree with the latter on ethical grounds too, let alone for its lack of scalability, but that's another discussion...)
We urgently need communities of learning around specific new technologies, that translate inventions into peer-to-peer instruction for mainstream workers, in a timely and hyper-granular fashion. These will look more like social networks and Wikipedia than universities. The faster implementation at scale of new technologies will lead to steeper experience curves.
This is really important. Beyond the sales of technologies, a critical accelerator is whether people know how to make the most of it in new circumstances. Whether that is managers figuring out how it changes their business or engineers figuring out how ti implement it.
I love this. Growing in an existing field or learning how to move forward in a new profession is largely propelled by interaction with other professionals. Especially regarding trading best practices for very nuanced processes. Because startups operate in these cutting edge spaces, it's often difficult to interact meaningfully due to IP implications. Bringing in consultants and other third party assisters to problem solve across internal projects has been very useful for my teams.
The start point of sustainability being 'expensive' is only a relative assertion. Given that total global fossil fuel subsidies continue to run at $5 trillion / year (Coady et al, World Development 91 (Mar 2017) pp.11-27), the world has huge capacity to redirect that spend into green tech and innovation and immediately redefine the newer technologies as 'cheaper' post-subsidy.
Technologies are generally more expensive when they are first produced: because we don't know how to make them cheaply. That is the argument of Wright's Law and learning curves. So even if you slash fossil fuel subsidies, you don't change the fact the the next generation of solar PV (for sake of argument) will have better price performance than the current generation.
Shifting fossil fuel subsidies to cleaner technologies would help to kick-start the market. This is, of course, one of the four or five key reasons why solar energy became cheap. The German government subsidised solar for several years in the 1990s. This drove demand, which in turn drove Wrightian learning in the manufacture of solar PV, driving down its cost. Today solar PV is cheaper now than the in 1990s even without subsidies.
The other - critical - support the German government provided was to launch the "1000 Roofs" initiative at the point when solar panel take-up was still lacklustre. They specifically targeted clusters of homes down to the particular block or street level in social networks that would create a critical mass of early adopters. By 2016, those neighbourhoods had grown into a dense concentration of solar installations. Japan copied this idea and launched a "70,000 roofs" look-alike. Same neighbourhood impact on early adoption ensued. Damon Centola's book "Change: How to make big things happen" (2021) is a great read on all this and recommends policy support for peripheral groups who early adopt.
Totally agree the learning curve will apply. The fossil fuel industries meanwhile will begin to reveal itself as relatively more expensive compared to solar etc. than current economic, accounting and political standards can admit. In addition to $5tn of subsidies p.a., if development of to-be-stranded fossil fuel assets continues, around $6.74 trillion of capex will prove to have been wasted over the next 10 years. Why not spend that on green tech instead, to accelerate the learning curves?
I would argue the reason why some of the trickier clean technologies haven't scaled is because of a funding issue. Governments are good at funding R&D but not often at scaling technologies (unless they're socialist govnts that own huge chunks of industry). Venture capitalists have been successful funding low-capex technologies and getting them to M&As. But clean-tech 1.0 (such as solar, wind, batteries) scaled because large manufacturers took it on (and there were some nice subsidies). Large Chinese companies took on the task of technology improvements for solar, and Korean tech giants took on the li-ion battery world.
I would suggest we are struggling with nuclear fusion, next-gen biofuels, carbon capture etc because there aren't the huge incumbent manufacturers/industries owning these things.
This is basically, to my understanding, how pharma has scaled well. big pharma see their job as scaling technology. This is so different from the utility sector or many car makers (who mostly compile other OEM's tech).
Perhaps this is an area where government can use its long term finance powers to offer smart funding that spans the whole expected lifecycle, smoothing out short-run cost and long-run gains at either end.
For example, a specific government investment fund that seeks to estimate smoothed out returns from a specific socially-desirable innovation (e.g. in climate change or health tech) and offers pump priming finance as part of (say) innovation challenges that would absorb the cost and much of the risk on the left side of the Wright chart in return for a larger slice of the gains on the right for a defined period, after which it is gravy for the firms involved. This reminds me of the thinking some finance / insurance groups are putting into the funding of social care, for example, where we probably need new financial products that encourage long-term savings to fund potentially large social care costs at the end of peoples' (or their parents') lives. I'm no expert, but this challenge also reminds me of the conditions in shipping that necessitated Lloyds and the insurance industry - high risk, high return, and often slow to pay back ventures that are too risky on their own, but manageable in aggregate if risk is pooled and shared, only in this case it would be cheap long-term government borrowing (or printer goes brrrrr) instead of a syndicate.
Hi Lee! I'm curious what you all think of Climate KIC's Deep Transition approach -focused on a regional transition and looking to redefine what it means to have a portfolio. I know when working with climate tech companies in the States vs. Germany we (US) are lacking in mechatronic capabilities - the spatial learn by doing and tinkering skills to own operate and maintain this kind of tech let alone invent and innovate. Regional investments in learning might be a big lever to be funded alongside investment in the tech.
Hi Jen. I like the look of the programme, especially the emphasis on ecosystem enablers and a systems approach, but I don't know much in reality about the results or impact. By coincidence, this week I am overseeing final edits to a series of learning videos for emerging government / public sector leaders (not UK or US). The sections on innovative public policy are very much focused on regional / city-based initiatives that seek to create entrepreneurial ecosystems spanning the whole gamut from physical engineering to learning, finance and culture in the hope of sparking more combinatorial innovation in areas like this. I think you're right that alongside the mechatronic shiny things, we also need to stimulate the tinkering and re-mixing or re-applying of innovations, and I think learning is a key part of that. It's often not the pure tech innovation but the clever application of existing solutions to interesting problem areas that makes the breakthrough.
I think there is a model in place for this already and that there is a complete life-cycle that optimizes this curve functioning today. The problem is, most (but thankfully not all) of the actors in this value chain prioritize and optimize around metrics like valuation, IRR, and market cap rather than contribution to sustainability or carbon zero goals. A shift in focus could make a ton of difference.
We have both government and higher ed doing the research required to create innovation, and again these institutions play a role in making these innovations feasible. Here in the U.S., for example, DARPA has been instrumental in the creation of so many everyday technologies. Another example are government positive technology transfer programs that allow private companies to commercialize government funded technologies.
Then, we (globally) have a vast funding and investment mechanism in place that creates the ability for private companies to figure out how to drive down costs and commercialize a given technology. I view the problem here as one of prioritization of outcomes, and which type of innovations to fund at what level. If ecotech companies received the funding, attention, and engagement that the last social media unicorn did, we'd have the benefit of accelerated cost improvements to truly global impact level innovations.
Big investment in science and tech research. Building a sustainable economy is hugely complex and multi-faceted and we have a lot of big gaps. We need a big supply of ideas and talent flowing out of universities.
That certainly helps with creating the innovations. But getting them cheap seems to be market-demand driven as we "learn by doing". This is what Wright's Law tells us. So the question is what tools can we adopt to accelerate "learning by doing"?
Boston University James Bessen’s “Learning by Doing” book explores a bunch of that - on the education side and shows how latency in reskilling leads to latency in “shop-floor” trial and errors. So that’s one. Another one is to create communities of knowledge - hyper scale subReddit+LinkedIn Groups+... etc to get people to learn from each other. Someone needs to throw in some initial investment to do that (government, possibly; vendors, maybe). Governments could also subsidize the first implementations in exchange for the work blueprint and results, so that a central repository of input/output is accessible.
Is the non-financial award not what makes an early adopter and early adopter? The scratching of the itch to break with tradition and be technologically progressive e.g. buy the objectively inferior early models of flat screen TV not because they were better than cathod ray tube TV's but because they resonated with the buyer's core belief in "progress" (I'm ripping off Simon Sinek here...)
Before things cross the chasm users could have some level of intrinsic incentive. But it can also be that they want to harness the experience curves before anyone else, so that they can deliver services with referenceable previous experiences, which gives them competitive advantage early on.
In the past we have had abundant resource inputs, anything that can be done to ensure raw material inputs remain in supply should be a target. Less concentrated energy may bite into supply as well?
The premise of cheap may be challenged by 2 constraints. 1 All this innovation needs to happen within a C02 total output of 300-400Gt (go over this an we are toast) and 2. Energy intensity is diffused in future, see https://ukfires.org/absolute-zero/ . This is an extreme view, but we could all be facing a future where we have 40% less energy to use - this should focus our efforts.
From the perspective of carbon in concrete, I can see the temptation to milk the monopoly offered by a patent if a company has invested in novel low carbon mixes and has taken risk in their application to prove viability. But with concrete being such a huge global source of emissions, this approach seems difficult to reconcile ethically if the patent holder keeps the cost at a premium long after investments are recovered. Is a policy approach for the government to step in, buy the patent with taxpayer money and make it "open" to benefit the taxpayer? How do you set a market value in that context?
Interface design. Early adopters of leading edge technologies are at heart tinkers and putters. The automobile, radio, television, computer/phone all needed an intuitive interface before they achieved mass acceptance.
I am not sure whether this answers the question but I have been wondering whether mining crypto could be linked to net zero activities rather than carbon producing. Does anyone have a view on how this could be achieved?
I once asked Doyne Farmer, one of the co-authors on this beautiful paper on Wright's Law https://arxiv.org/abs/1703.05979, what was the fundamental difference between the technologies where the law works and where it seems not to. It has not worked in chemicals, and it has not worked in nuclear. Interestingly, he said he thought all the other technologies were related one way or another to printing - to decomposing a process into small reconfigurable units each of which could be recombined to create variety. Weaving, metal stamping, etc are all derivatives of the printing press in this way. As of course is wafer production, solar, software, etc.
So what is the implication of this for accelerationism - identify those processes that can be decomposed in this way and focus on those; don't waste $$ on the undecomposable, like fluid processes or huge custom builds - like nuclear. Coordination through standards can often help this process along - the alphabet on the movable type printing press is a really good example of Wright-Law inducing standardisation.
Beautiful, yes. The connection with miniaturisation is really important. It also connects to energy use and the decentralisability of finished products. Decentralised products (like chips) can scale much faster than big lumpy ones, which means demand (and, hence, demand doubling) can grow faster.
Very interesting. I want to see if this works for immuno-oncology and cell therapy cures which cost up to $1.8m per patient.
Depends what the drivers of the cost are, really.
When I have read pop-science articles about these, they have seemed to me to be inherently custom, 1-off projects. So very hard to get dramatic cost reductions. But maybe that is wrong - would love to know more about the details of the process.
Sounds like you need the ability to run natural experiments in the field at scale, which isn’t practical in large monolithic deployments.
Ahh... sounds like "small is beautiful" in action. "[decompose] a process into small reconfigurable units each of which could be recombined to create variety" and "don't waste $$ ... huge custom builds - like nuclear" (and E. F. Schumacher would agree with the latter on ethical grounds too, let alone for its lack of scalability, but that's another discussion...)
We urgently need communities of learning around specific new technologies, that translate inventions into peer-to-peer instruction for mainstream workers, in a timely and hyper-granular fashion. These will look more like social networks and Wikipedia than universities. The faster implementation at scale of new technologies will lead to steeper experience curves.
This is really important. Beyond the sales of technologies, a critical accelerator is whether people know how to make the most of it in new circumstances. Whether that is managers figuring out how it changes their business or engineers figuring out how ti implement it.
I love this. Growing in an existing field or learning how to move forward in a new profession is largely propelled by interaction with other professionals. Especially regarding trading best practices for very nuanced processes. Because startups operate in these cutting edge spaces, it's often difficult to interact meaningfully due to IP implications. Bringing in consultants and other third party assisters to problem solve across internal projects has been very useful for my teams.
The start point of sustainability being 'expensive' is only a relative assertion. Given that total global fossil fuel subsidies continue to run at $5 trillion / year (Coady et al, World Development 91 (Mar 2017) pp.11-27), the world has huge capacity to redirect that spend into green tech and innovation and immediately redefine the newer technologies as 'cheaper' post-subsidy.
Technologies are generally more expensive when they are first produced: because we don't know how to make them cheaply. That is the argument of Wright's Law and learning curves. So even if you slash fossil fuel subsidies, you don't change the fact the the next generation of solar PV (for sake of argument) will have better price performance than the current generation.
Shifting fossil fuel subsidies to cleaner technologies would help to kick-start the market. This is, of course, one of the four or five key reasons why solar energy became cheap. The German government subsidised solar for several years in the 1990s. This drove demand, which in turn drove Wrightian learning in the manufacture of solar PV, driving down its cost. Today solar PV is cheaper now than the in 1990s even without subsidies.
The other - critical - support the German government provided was to launch the "1000 Roofs" initiative at the point when solar panel take-up was still lacklustre. They specifically targeted clusters of homes down to the particular block or street level in social networks that would create a critical mass of early adopters. By 2016, those neighbourhoods had grown into a dense concentration of solar installations. Japan copied this idea and launched a "70,000 roofs" look-alike. Same neighbourhood impact on early adoption ensued. Damon Centola's book "Change: How to make big things happen" (2021) is a great read on all this and recommends policy support for peripheral groups who early adopt.
Totally agree the learning curve will apply. The fossil fuel industries meanwhile will begin to reveal itself as relatively more expensive compared to solar etc. than current economic, accounting and political standards can admit. In addition to $5tn of subsidies p.a., if development of to-be-stranded fossil fuel assets continues, around $6.74 trillion of capex will prove to have been wasted over the next 10 years. Why not spend that on green tech instead, to accelerate the learning curves?
This is old so the numbers will be bigger now, but it's an interesting thought bubble http://carbontracker.org/wp-content/uploads/2014/09/Unburnable-Carbon-2-Web-Version.pdf
I would argue the reason why some of the trickier clean technologies haven't scaled is because of a funding issue. Governments are good at funding R&D but not often at scaling technologies (unless they're socialist govnts that own huge chunks of industry). Venture capitalists have been successful funding low-capex technologies and getting them to M&As. But clean-tech 1.0 (such as solar, wind, batteries) scaled because large manufacturers took it on (and there were some nice subsidies). Large Chinese companies took on the task of technology improvements for solar, and Korean tech giants took on the li-ion battery world.
I would suggest we are struggling with nuclear fusion, next-gen biofuels, carbon capture etc because there aren't the huge incumbent manufacturers/industries owning these things.
This is basically, to my understanding, how pharma has scaled well. big pharma see their job as scaling technology. This is so different from the utility sector or many car makers (who mostly compile other OEM's tech).
Perhaps this is an area where government can use its long term finance powers to offer smart funding that spans the whole expected lifecycle, smoothing out short-run cost and long-run gains at either end.
Interesting - can you put some flesh on that idea?
For example, a specific government investment fund that seeks to estimate smoothed out returns from a specific socially-desirable innovation (e.g. in climate change or health tech) and offers pump priming finance as part of (say) innovation challenges that would absorb the cost and much of the risk on the left side of the Wright chart in return for a larger slice of the gains on the right for a defined period, after which it is gravy for the firms involved. This reminds me of the thinking some finance / insurance groups are putting into the funding of social care, for example, where we probably need new financial products that encourage long-term savings to fund potentially large social care costs at the end of peoples' (or their parents') lives. I'm no expert, but this challenge also reminds me of the conditions in shipping that necessitated Lloyds and the insurance industry - high risk, high return, and often slow to pay back ventures that are too risky on their own, but manageable in aggregate if risk is pooled and shared, only in this case it would be cheap long-term government borrowing (or printer goes brrrrr) instead of a syndicate.
Hi Lee! I'm curious what you all think of Climate KIC's Deep Transition approach -focused on a regional transition and looking to redefine what it means to have a portfolio. I know when working with climate tech companies in the States vs. Germany we (US) are lacking in mechatronic capabilities - the spatial learn by doing and tinkering skills to own operate and maintain this kind of tech let alone invent and innovate. Regional investments in learning might be a big lever to be funded alongside investment in the tech.
Hi Jen. I like the look of the programme, especially the emphasis on ecosystem enablers and a systems approach, but I don't know much in reality about the results or impact. By coincidence, this week I am overseeing final edits to a series of learning videos for emerging government / public sector leaders (not UK or US). The sections on innovative public policy are very much focused on regional / city-based initiatives that seek to create entrepreneurial ecosystems spanning the whole gamut from physical engineering to learning, finance and culture in the hope of sparking more combinatorial innovation in areas like this. I think you're right that alongside the mechatronic shiny things, we also need to stimulate the tinkering and re-mixing or re-applying of innovations, and I think learning is a key part of that. It's often not the pure tech innovation but the clever application of existing solutions to interesting problem areas that makes the breakthrough.
Combinatorial innovation indeed. As always you're right ahead of me :)
I think there is a model in place for this already and that there is a complete life-cycle that optimizes this curve functioning today. The problem is, most (but thankfully not all) of the actors in this value chain prioritize and optimize around metrics like valuation, IRR, and market cap rather than contribution to sustainability or carbon zero goals. A shift in focus could make a ton of difference.
We have both government and higher ed doing the research required to create innovation, and again these institutions play a role in making these innovations feasible. Here in the U.S., for example, DARPA has been instrumental in the creation of so many everyday technologies. Another example are government positive technology transfer programs that allow private companies to commercialize government funded technologies.
Then, we (globally) have a vast funding and investment mechanism in place that creates the ability for private companies to figure out how to drive down costs and commercialize a given technology. I view the problem here as one of prioritization of outcomes, and which type of innovations to fund at what level. If ecotech companies received the funding, attention, and engagement that the last social media unicorn did, we'd have the benefit of accelerated cost improvements to truly global impact level innovations.
Big investment in science and tech research. Building a sustainable economy is hugely complex and multi-faceted and we have a lot of big gaps. We need a big supply of ideas and talent flowing out of universities.
That certainly helps with creating the innovations. But getting them cheap seems to be market-demand driven as we "learn by doing". This is what Wright's Law tells us. So the question is what tools can we adopt to accelerate "learning by doing"?
Boston University James Bessen’s “Learning by Doing” book explores a bunch of that - on the education side and shows how latency in reskilling leads to latency in “shop-floor” trial and errors. So that’s one. Another one is to create communities of knowledge - hyper scale subReddit+LinkedIn Groups+... etc to get people to learn from each other. Someone needs to throw in some initial investment to do that (government, possibly; vendors, maybe). Governments could also subsidize the first implementations in exchange for the work blueprint and results, so that a central repository of input/output is accessible.
Excellent question. My question is How can early adopters be financially compensated (incentivised) for being early adopters?
In particular for products which enjoy significant learning effects and will deliver huge welfare gains (e.g. decarbonisation) more broadly.
Is the non-financial award not what makes an early adopter and early adopter? The scratching of the itch to break with tradition and be technologically progressive e.g. buy the objectively inferior early models of flat screen TV not because they were better than cathod ray tube TV's but because they resonated with the buyer's core belief in "progress" (I'm ripping off Simon Sinek here...)
Before things cross the chasm users could have some level of intrinsic incentive. But it can also be that they want to harness the experience curves before anyone else, so that they can deliver services with referenceable previous experiences, which gives them competitive advantage early on.
In the past we have had abundant resource inputs, anything that can be done to ensure raw material inputs remain in supply should be a target. Less concentrated energy may bite into supply as well?
The premise of cheap may be challenged by 2 constraints. 1 All this innovation needs to happen within a C02 total output of 300-400Gt (go over this an we are toast) and 2. Energy intensity is diffused in future, see https://ukfires.org/absolute-zero/ . This is an extreme view, but we could all be facing a future where we have 40% less energy to use - this should focus our efforts.
From the perspective of carbon in concrete, I can see the temptation to milk the monopoly offered by a patent if a company has invested in novel low carbon mixes and has taken risk in their application to prove viability. But with concrete being such a huge global source of emissions, this approach seems difficult to reconcile ethically if the patent holder keeps the cost at a premium long after investments are recovered. Is a policy approach for the government to step in, buy the patent with taxpayer money and make it "open" to benefit the taxpayer? How do you set a market value in that context?
Interface design. Early adopters of leading edge technologies are at heart tinkers and putters. The automobile, radio, television, computer/phone all needed an intuitive interface before they achieved mass acceptance.
I am not sure whether this answers the question but I have been wondering whether mining crypto could be linked to net zero activities rather than carbon producing. Does anyone have a view on how this could be achieved?
quite a lot of people thinking about that exact question - see for example https://www.climatechaincoalition.io/