 Our next speaker is Greg Nemet, who will be speaking on the topic of technology diffusion and deployment. Greg Nemet is a professor at the University of Wisconsin-Madison's Lafellette School of Public Affairs, and the title of his talk is Learning Adoption and Scale Up. Over to you, Greg. Thank you. Thank you, Sarah. Thanks for, I mean, be part of this great panel. It's always a challenge to follow David Victor, but it's also, I always learned so much from him. He's got such a distinct perspective, so it's great to see some of that. So on barriers, I'll be talking about learning adoption and scale up. One thing I'd like to talk about is this innovation gap that some colleagues and I have identified. Sabina Fuss actually mentioned this a couple of days ago in the conference, and it motivates a project that I'll talk briefly about. So that's one thing. A second is talking about the need for accelerated models of scale up and adoption and some early ideas about the potential for those. And then three on policy support. So the main point I'd like to make with that is that we need something that I'm starting to call dynamic policy support. And the key to that is that it absolutely goes beyond research and development. Innovation is at the core of what we need on carbon dioxide removal, but that innovation involves way more than just R&D. And I'll talk a little bit about the broader set of policy enablers that we'll likely need for that. And that this project on the slide here will really evaluate. So just to give you a sense of where I'm coming from on that, this is a new project that started four weeks ago on, it's a comprehensive assessment of CDR and also solar radiation management. And here's some of the key institutions and players there. And part of the idea here, and I think this has been made over the last couple days, is that carbon dioxide removal is one of a broad set of tools that we're going to need to deal with the climate problem, including reducing demand, cleaning up energy, changing industrial processes, and then these other options that we're looking at in the project, which is removing CO2 and potentially blocking sunlight and thinking of some of the issues with those and then an adaptation as well. So getting to what we'll actually do in this project and how it connects to what we're talking about here on barriers to scale up, one thing I'd point out is that we've got, I don't know, eight work packages over the next six years and almost all of them relate in some way to what we're talking about here. So the first part is on thinking about what are the dollars per ton and then how many tons in the next few decades. Two on bottlenecks, that's pretty similar to what we're talking about here. Three on social acceptance and justice that I know Anish will be talking about in a few minutes. Learning and adoption, that's something that's kind of core to my work that I'll be really pushing forward on this and then risks and then the policy support. So let me talk about some of those blue items in the next couple of minutes. The motivation here for this project, and I think this is kind of broadly why a lot of people are here on the call this week, is that we're potentially talking about billions of tons by the 2050s. So that's on the order of tens of percent of today's emissions. As David mentioned, if you look historically at technologies, the scale up time takes a long time. It's not talking about something happens in a few years, but we always talk about decades when we talk about scale up. Third, and this is the innovation gap that the focus so far has been almost overwhelmingly on R&D, and that for we need this post-R&D innovation activities and support, and I'll talk a bit about those. So the objectives here in this project are to inform scale up, specify some of those enablers, identify the barriers, and really get it timing and pace. I think that's really important information. I think a lot of people may have it wrong in terms of how quickly some of these technologies will be at our disposal to remove substantial amounts of CO2, and that informs a lot about the other things that we need to do in terms of cleaning up our energy supply and adapting to the changes that will happen, and then some of the insights for policy there. So in terms of some of the theoretical basis that I'm talking about for this project, but I think it's helpful for anyone who's working in this area is that the technologies are new, but the ideas are not, and there's a lot that's been built up about how innovation systems and transitions work, that Frank Heel's work that David mentioned is core to that. In the energy technology innovation system work that I've been involved in, one of the reasons I put these six stages here, there may be other stages more or less than the six here, but innovation is way more than just funding research and development, and the crucial parts are downstream, and that's really what we're going to focus on in this project, and I think is really crucial to CDR working. What's also key are these feedback. So these arrows here are flows of knowledge from one stage to the other, but market acceptance informs research directions and how niche markets work out can also alter the configuration, the design of the next generation of technology. So really thinking about how the experience and the demand side is not just about reducing emissions, but it's also about improving the technology. So there's innovation involved all throughout this process, not just in terms of coming up with new patents and inventions, and that's something that comes out of the economics of innovation. And then the final point on national innovation systems that I've found really helpful to understand what's been happening in some of these complicated stories in the real world. And David mentioned it, in solar, it's multiple countries played a role there because they each had distinct capabilities that have to do with their education systems or manufacturing capabilities, how the markets work, how governments work. And so that's important to think about too is that technologies that might work well in one place might not in another and conversely for others as well. This is something we looked at a couple of years ago where we found 2,500 articles on carbon dioxide removal and coded them by the innovation stage at which they were focused. And the overwhelmingly largest area was on research and development. And these downscale stream activities that I've mentioned are really important demonstrations, scale up, identifying niche markets, how demand might work, public acceptance and the feedbacks that go back to the earlier stage were just a much smaller part. And so that's what we're talking about this innovation gap that we need much more effort on the demand side and the linkages back to innovation processes on carbon dioxide removal. And I think there's strong implications for policy of that as well. On the need for acceleration, so this is something that Sabina mentioned this as well. So just to give the short of it, if we want to direct air capture to follow the pathway of solar where the first commercial solar plant was in 1957 and the first commercial DAC plant was in Switzerland in 2017, that means we get low cost direct air capture in the 2070s and we start to have gigatons of removal by 2100. Obviously with net zero and other climate targets, that's not really going to do us very well. And so what we need is something that happens a lot faster. And so that's really what we need to think about is what would we do differently? Solar is a great success and it's making big changes to energy systems, but it happened way too slowly for it to completely serve as a model for DAC and other carbon dioxide removal technology. So key question in our project that I think a lot of people need to be working on is how can we make it go faster? What are the things that would enable this low cost, say by 2040 and then gigatons by mid-century? We've got a couple of examples of that and put it quantitatively, that's a factor for faster than what solar did. We've got some examples with vaccines. That was four times faster, about 11 months to develop a vaccine. It helped to have mRNA technology available and that's probably what we're going to need to do with CDR technologies is have things like synthetic biology or machine learning and digitalization throughout the system that make these processes go faster and the policy support as well, which was crucial for getting vaccines out so quickly. We also have some examples of scale-up happening at a relatively fast pace. These are companies that have scaled up their production of solar panels and interestingly, they're not all in China. The first one's Japan. The second one's in Germany and the last two are China. The third one on the right there is on direct air capture. Climewarps used to have a target for 1% of removal by 2025 and that was super ambitious. It was about twice as fast as the fastest solar company has scaled up. It's not out of the realm of possibility but it would be learning from this gigantic expansion that we've seen that first happened in Japan, then Germany, then in China. There are some models to learn from that approximate the speed at which we need. One of the other points that I would make on learning lessons from previous innovations is that we need different models for different technologies. Here's a mapping that's guiding some of our project here, which is that there's a certain technologies that look like solar PV. There's a technology component. There's massive iterations and there's a disruptive component as well in which we can compromise on attributes that people don't care about that much in order to make dramatic reductions in cost and some direct air capture not all has those attributes. Other areas, some of the natural solutions look more like what we've learned from how we've done fertilizer and irrigation for the hybrid seeds for the green revolution and then in part three there at the bottom, we can think of large industrial facilities and some direct air capture plants look like that, the bioengineering or bioenergy with carbon capture and sequestration looks like that and you can see that the companies that are scaling that up are looking at things like refineries and chemical plants where it's large scale and system integration is really the key where so much of the equipment is really just to protect the core processes from themselves and the pressures and temperatures and corrosion and that's a really different challenge from the other categories as well and then as I mentioned in terms of acceleration and category four on general purpose technologies, those have been crucial and if there's any chance of speeding up innovation to get to the some of the targets that we need for net zero, it's really taking advantage of things like machine learning and synthetic biology and then digitalization the connectivity that's that wasn't the case for most of the time of say solar's development that could make things go happen faster. So different models for different carbon removal options. You know this is just an example of where we go in this project is to identify some of the specific policy accelerators that would support innovation make things go faster and really catalyze some of these feedbacks from later stages in the process to earlier stages and without going through these now one thing to think about again is that R&D is certainly part of this. It's continuous and dynamic R&D so it's changing as we get further along in the problem space but that other parts of policy support are crucial as well and the middle part is about enabling knowledge to flow through the system and the bottom part is about making markets work and grow and the robustness is in part addressing David's important point about credibility as a way to catalyze investment and reduce the cost of it as well. So those are examples of where we'd like to go in some of the policy support for for different carbon removal technologies is that there is a sense in which small scale technologies could turn out to be more scalable than large ones and just to show you some of the recent evidence to support not necessarily this is true but that is a worthwhile hypothesis to think about. If you look at technology scale that's on the horizontal axis there and you know these are several like nine or ten orders of magnitude here you can see this relationship between learning rate how much costs go down with a doubling of production and scale so large scale technologies learn more slowly and small scales technologies seem to learn more quickly so small scale technology seem to improve more quickly they also get adopted more quickly and the system worked by Charlie Wilson Arnold Brubler at YASA looking at adoption times and how those also seem to go faster so lower on the y-axis at smaller scales and for both of these there's a lot of dispersion in the data there's a lot that's unexplained and there's things we'll have to account for but it is interesting to think to support this overall idea that smaller may be more scalable and this is kind of my take on looking at those data is that one of the ways that smaller seems to improve and get adopted more quickly is the numbers of iterations and if you look at how many nuclear power plants that we've built in the history of nuclear industry this is less than a thousand and we've built four billion solar panels and that's a million times more opportunities to improve to change the processes to introduce new technologies and new components and I think that is something that we may see in some cdr technologies as well okay so getting down into the policy implications and just to finish up here so if we think about taking some of these results from other technologies and some of these ideas and some of the theory and put it together I think some of the implications are that we should focus our effort and our expectations on technologies that are dynamic that are improving that have a possibility to go fast to get better to incorporate some of the general purpose technologies like senbio and computers and information technology second there's something about small unit size that learns faster and gets adopted more quickly that there's a way to do large and take advantage of the small and that's through modularity and tom mentioned that briefly in his remarks I think that's a pretty interesting component where you could have large-scale facilities but that have a lot of modularity within the components that allow for scale and iteration improvement that we've seen is such a powerful factor in these other other areas and that allows for iterative improvement and then finally that the local system integration is going to be crucial these aren't we're not just talking about devices they have to be built into the local environment and the local technological system that might include pipelines it might include secretation places and then markets in which some of those incentives are created from so local system integration is is crucial here you know and just to put this in the broader very broad picture here this is kind of a historical look at just to justify my claim that we need more than r&d is if you look at and we need more than carbon pricing if you look at the evolution of climate policy it's gotten broader in two ways one it's incorporated more goals that's the first row that I'll show there and then below there the policy solutions have gotten more comprehensive there's more parts there's more policy instruments included so if you think about going to back to the 1990s that we looked at what worked for dealing with acid rain it was putting prices on pollution carbon prices would be the solution there then we start to think about induced technological change so that's where we start to say well we need r&d too so maybe we need a polo project or Manhattan project for climate change in addition to a carbon price then the innovation systems work that's kind of come with frank heels that david mentioned has done a lot with this is to think about innovations as systems and that those need to be nurtured and supported and so that starts to think about these intermediate activities that we'd want to include as well and then finally you know in the last few years what's gotten a lot of interest is you know a much broader set of goals that includes distributional impacts like equity and justice and that also takes into account if we're talking about net zero by 2050 that speed matters it's not just about making steady incremental improvements at least cost it's about doing this all pretty quickly and creating options for the next couple of decades and to do that you need a much broader set of policies so that's kind of where i come down on this uh last two slides here on this idea of supporting carbon dioxide removal with dynamic industrial policy so some of the components of it are urgency and acceleration our goals that we've got multiple policies that they can be sequenced in a smart intentional strategic way potentially to deal with some of the political interest that david talked about that government is not kind of passively just setting a price or forming markets but that it engages more deeply in the innovation process that it's technology plus inclusivity as goals social acceptance being crucial because none of these things are going to be kind of just in the background they're going to be front and center to people's lives if we're going to do this in such a big way local learning and system integration are a big part of this and that and david mentioned this as well is that it needs to be adaptive we don't know how to do all of this we don't know which technologies to bet on there's portfolios there's different policies to try it but that we need to learn from the policies and that a policy not working is not necessarily failure or a mistake it's could be something that we learn from for the next generation of policy so all of this is a much more intrusive and engaged role for the state and it comes with risks as well and this is my last slide some of the risks that would have to be managed in this type of approach of dynamic industrial policy for carbon removal are dealing with rent seeking there would be way more opportunities for that that governments rather than the private sector is going to have to be doing more selection and picking winners which is exactly what the private sector investors try to do the government will have to start playing a role in that too information access will be improved the important for making good decisions the role of the private sector with such a big public sector role raises the possibilities for crowding out and excluding the private sector and then risk aversion which has been a big problem in the past for the public sector needs to be managed as well we need to be making big bets thinking about portfolios not penalizing ambitious bets that didn't pan out because we need to take some of these risks to make things happen well and fast so I'll stop there and yeah thanks so first question is between 1990 and current days in the last 30 years honestly the the number of large projects in development developed economies has been rather minimal so do you see this as due to industry not having the sufficient competencies or is it more not having not investing in these demos and scale so are you talking about like the slide that tom showed about ccs projects over time and how yeah there haven't been that many I mean that's been a problem because we've been talking about needing to do ccs for you know all that time or at least most of that time and yet we're not seeing the investments I mean I guess the the biggest issue is on the demand side it's it's been pretty hard to say that if you invested in this technology you know that your big get a payback on it and has to do with the credibility that David talked about and some of the markets and having some return on on that investment so that's certainly one reason but we haven't seen it but it's also something specific to such large investments I mean if we think about some of the ccs plants that have gone ahead one in Canada and a couple in the US these are billion dollar projects and that's kind of unlike what we can see in other areas that have moved ahead quickly whether it's solar or even you know early electric vehicles you didn't need to bet billions of dollars on one single installation and need to get that right to get to the next one that's such a hard hill to climb and so you know there may have been actually a better approach of iterative upscaling of doing some small projects and building them over time that that might have worked out better so yeah it's it's hard to know what would have worked better but yeah certainly that's a big issue to flag is that we should have had a lot more experience and projects to learn from than we have today it's good to see the ones that Tom showed but it's it's harder to see of how that provides us the roadmap to gigatons that a lot of those projections expect okay so behavioral change is a big part of some of the IEA models and some of the net zero scenarios so how can technology development drive behavioral change in your models well I mean it's it's certainly something that's emerging you know with the connectivity and digitalization and it's it's almost more about how people use it and how the technology gets deployed than than the technology itself I mean think about car sharing and connecting that with electric vehicles and maybe autonomous vehicles at some point I mean the behavioral part there is so huge and it's a little bit harder to predict but it becomes crucial as some of these technologies especially the small ones you know are directly connected to people's own decisions like what car they drive where they're going to live how they're going to commute those are you know much more distributed decisions than what are the next hundred power plants that we're going to build in the U.S. or in India and that's you know that's it's much harder to predict because these are not necessarily profit oriented households that are deciding on the next car it's people influenced by their neighbors or influenced by what the car looks like or what it might need them to do or what they've heard about them and so yeah the human behavior part is is crucial to the adoption part especially if we think about these technologies being small and closer to end use which I think a lot of them will be. Your slide showing where we learned faster at smaller scales was really quite interesting does this point to more of a distributed market deployment approach as needed for technologies to progress? Yeah I mean I think it it does I mean the the one caveat I put on that is it could be that the examples we've seen a lot of them are from developed economies and most of the growth in energy services will not be from currently developed countries and so it could be that those places have different configurations different national innovation systems different ways that capital moves and people make decisions and so there's a possibility that there's opportunities for large scale in different places but from what we've seen so far that the data shows pretty well is that yeah the small scale stuff goes faster and I think there's a lot of really reasonable explanations for it which have to do with their lower risk there's more iteration you can improve. The niche markets is gigantic I mean there's so many different niche markets if you've got a small technology that can grow and maybe live in that protected niche for a while and improve and get reliable and be expanded to the next niche that doesn't work so well if the niche requires a 10 billion dollar one project and so I think yeah there is something that's driving this distributed adoption that has to do with you know reducing risk in an area that has lots of risk that has to do with technology and policy and how people are going to react already and so if there's ways that you can reduce risk with smaller investments and incremental and niches I think that is probably what's going to propel that that distributed arrangement that you mentioned. Awesome great okay thank you so much.