We lead our lives largely unaware of the immense effort required to support them. All of us grew up inside the so-called “Grid” — actually one of many interconnected regional power grids that electrify our modern world. The physical infrastructure and the regulatory intricacies required to keep the lights on: both have grown organically, piecemeal, in complex networks that nobody seems to fully understand. And yet, we must. Compared to life 150 years ago, we are all utterly dependent on the power grid, and learning how it operates — how tiny failures cause cascading crises, and how tense webs of collaborators make decisions on the way that electricity is priced and served — matters now more than ever.
Welcome to COMPLEXITY, the official podcast of the Santa Fe Institute. I’m your host, Michael Garfield, and every other week we’ll bring you with us for far-ranging conversations with our worldwide network of rigorous researchers developing new frameworks to explain the deepest mysteries of the universe.
This week on Complexity, we speak with SFI External Professor Seth Blumsack (Google Scholar page), Professor of Energy and Environmental Economics and International Affairs in EME and Director of the Center for Energy Law and Policy at Penn State. In this conversation we explore the arcane yet urgent systems that comprise the power grid and how it’s operated, reminding us that the mundane is ever a deep reservoir of questions.
If you value our research and communication efforts, please subscribe, rate and review us at Apple Podcasts, and consider making a donation — or finding other ways to engage with us — at santafe.edu/engage. You can find the complete show notes for every episode, with transcripts and links to cited works, at complexity.simplecast.com.
Thank you for listening!
Join our Facebook discussion group to meet like minds and talk about each episode.
Podcast theme music by Mitch Mignano.
Follow us on social media:
Twitter • YouTube • Facebook • Instagram • LinkedIn
Mentions and additional resources:
Topological Models and Critical Slowing down: Two Approaches to Power System Blackout Risk Analysis
by Paul Hines, Eduardo Cotilla-Sanchez, & Seth Blumsack
Do topological models provide good information about electricity infrastructure vulnerability?
by Paul Hines, Eduardo Cotilla-Sanchez, & Seth Blumsack
Can capacity markets be designed by democracy?
by Kyungjin Yoo & Seth Blumsack
The Political Complexity of Regional Electricity Policy Formation
by Kyungjin Yoo & Seth Blumsack
The Energy Transition in New Mexico: Insights from a Santa Fe Institute Workshop
by Seth Blumsack, Paul Hines, Cristopher Moore, and Jessika E. Trancik
EBF 483: Introduction to Electricity Markets
by Seth Blumsack
What’s behind $15,000 electricity bills in Texas?
by Seth Blumsack
RTOGov: Exploring Links Between Market Decision-Making Processes and Outcomes
by Kate Konschnik
Ensuring Consideration of the Public Interest in the Governance and Accountability of Regional Transmission Organizations
by Michael H. Dworkin & Rachel Aslin Goldwasser
Electricity governance and the Western energy imbalance market in the United States: The necessity of interorganizational collaboration
by Stephanie Lenhart, Natalie Nelson-Marsh, Elizabeth J. Wilson, & David Solan
Untangling the Wires in Electricity Market Planning, with Kate Konschnik
by Resources Radio
Matthew Jackson on Social & Economic Networks
Complexity Podcast 12
Elizabeth Hobson on Animal Dominance Hierarchies
Complexity Podcast 78
The Collective Computation of Reality in Nature and Society
Jessica Flack’s 2019 SFI Community Lecture
Tyler Marghetis on Breakdowns & Breakthroughs: Critical Transitions in Jazz & Mathematics
Complexity Podcast 67
Early-warning signals for critical transitions
by Marten Scheffer, Jordi Bascompte, William A. Brock, Victor Brovkin, Stephen R. Carpenter, Vasilis Dakos, Hermann Held, Egbert H. van Nes, Max Rietkerk & George Sugihara
Ricardo Hausmann & J. Doyne Farmer on Evolving Technologies & Market Ecologies (EPE 03)
Complexity Podcast 84
Tina Eliassi-Rad on Democracies as Complex Systems
Complexity Podcast 73
Mirta Galesic on Social Learning & Decision-making
Complexity Podcast 9
Signalling architectures can prevent cancer evolution
by Leonardo Oña & Michael Lachmann
The Ethics of Autonomous Vehicles with Bryant Walker Smith
Complexity Podcast 79
Image Credit: Paul Hines
Seth Blumsack (0s): The grid was not planned as sort of a single thing. Like nobody sat down one day and got out a box of crayons and started drawing things on a map and said, this is what the grid should look like. It's a system that has evolved over decades in response to things like population growth, increased demand for electricity. And it's one that was really kind of built around connecting natural resources to places where you would want to actually use electricity.
And the natural resources work. The original drivers of the grid are very different things that we think about today. This is not wind and sun. This is water for hydro-power coal, raw, natural gas or oil. And part of the story around just the physical topology of the power grid is connecting the locations of these natural resources with places where people want to use electricity.
Michael Garfield (1m 23s): We lead our lives largely unaware of the immense effort required to support them. All of us grew up inside the so-called grid, actually one of many interconnected regional power grids electrify our modern world, the physical infrastructure and the regulatory intricacies required to keep the lights on. Both have grown organically piecemeal and complex networks that nobody seems to fully understand. And yet we must. Compare to life 150 years ago, we are all utterly dependent on the power grid and learning how it operates at tiny failures cause cascading crises and how tense webs of collaborators make decisions on the way that electricity is priced and served matters now more than ever. Welcome to
Complexity
, the official podcast of the Santa Fe Institute. I'm your host, Michael Garfield, and every other week, we'll bring you with us for far ranging conversations with our worldwide network of rigorous researchers, developing new frameworks, to explain the deepest mysteries of the universe. This week on complexity, we speak with SFI External Professor, Seth Blumsack, Professor of Energy and Environmental Economics and International Affairs in EME and Director of the Center for Energy Law and Policy at Penn State.
And this conversation, we explore the arcane yet urgent systems that comprise the power grid and how it's operated, reminding us that the mundane is ever a deep reservoir of questions. If you value our research and communication efforts, please subscribe, rate, and review us on apple podcasts and consider making donation or finding other ways to engage with us at
. You can find the complete show notes for every episode with transcripts and links to cited works at
.
Thank you for listening, Seth. It's a pleasure to have you on
Complexity
podcasts.
Seth Blumsack(3m 19s): It is a pleasure to be here. Thank you so much.
Michael Garfield (3m 23s): So let's start as we often do by taking a trip through time, back to your origin story as a researcher, how you got into asking the kind of questions that you asked today in your work, how you became a scientist and then how that path brought you to SFI.
Seth Blumsack (3m 44s): All right. So the Seth Blumsack creation myth, here you go. So I was a math major in college and I went to college at Reed in Portland, Oregon, and I didn't want to leave Portland because I thought it was heaven on earth. So after graduation, I got a job as a journalist because like that's what you do after you major in math is you go and write for a living. So I got a job as a journalist working for a trade publication in downtown Oregon that covered energy stuff in the Western United States.
And this was in the mid 1990s. And so my beat, I guess, if they still use that word, so my beat was to cover the newly deregulated California electricity market. So in California, in the mid 1990s, it basically underwent this very, very large scale experiment in restructuring its electric utilities and deregulating electricity pricing and all sorts of stuff like that. So I did that for three years.
And for your listeners who don't sort of know this part of the story, in the summer of 2000, this newly created California electricity market, it basically became like a roulette wheel for all of these different energy traders and other entities that had figured out very, very clever ways to kind of manipulate this market, to drive up prices and sort of do all of this bad stuff. And so my job as a journalist was to basically talk to electricity traders every day.
And it's sort of weird to think about people that trade electricity, like people trade oil or natural gas, or like wheat or hog bellies or something like that, because electricity is a weird commodity because you can't touch it. Like you can't grab it. So, but this experiment with deregulation in California created this kind of gigantic commodity market for electricity.
And it was not very well designed. And there were a lot of loopholes in this market that clever companies like Enron is the most famous one, basically figured out how to manipulate. And so companies started figuring out how to manipulate this market. Prices in this market increased by about like three orders of magnitude, basically overnight. And while politicians and regulators were sort of busy squabbling about what to do about all of this in California, the state's largest utilities had all gone bankrupt because they were buying all of this expensive electricity and then having to resell it and really cheap prices to people who lived in California and basically like electricity had to be rationed in California. People had to line up for gasoline in the 1970s, or like a developing country where you get your two hours of electricity a day. And this is a major disaster, both economically and socially in California and in the Western U.S., it basically kind of destroyed the career of a rising star in the democratic party at the time, a guy named Gray Davis who was the governor of California and led to the election of Arnold Schwartzenegger as Governor of California.
So you can all these sort of interesting things, you can trace back to the California electricity debacle. So anyway, so my job was to talk to the traders who were playing in this market every day and figuring out sort of clever things to do, to basically suck money out of the California power grid and watching something as big as this happened and meltdown in real time, on the one hand, it was totally horrifying because when you start talking about doing things like rationing electricity in the middle of summer, I mean, this is serious.
Like people could die because of this. This is not a joke. But at the same time, it was just so fascinating to watch happen because it was just amazing to me, how so many smart people who put all of this time into designing this electricity market could get this so wrong. So it was just a really, really interesting thing to watch. And so after a few years of doing this journalist job, I was kind of felt like going to do something else.
The other cool thing about being an energy journalist in Portland, Oregon at the time was I got to hang out in Enron's trading room. And so I got to sort of see kind of firsthand all of the stuff that they were doing to suck money out of the California electricity market. So Enron really wanted to hire me because I was sort of clearly was like an analytical person. And I had basically spent the past three years figuring out the trading strategies of all of Enron's competitors. So they naturally wanted to hire me. And at the same time I'd been talking to this professor at Carnegie Mellon who wanted me to go to graduate school.
And so I talked to him about PhD programs and kind of in the spring of 2001, I sort of finally decided that I was going to leave my journalism job. So Enron offered me all the riches of the world to come work for them. And they said, Hey, we're going to give you like giant mounds of money and you get to stay in Portland which I thought was heaven on earth, or this professor from Carnegie Mellon, Lester Lave. I said, well, tell you what, I'll give you a bag of beans and rice every month as a graduate student and you have to move to Pittsburgh.
Which is a really wonderful city, but I didn't know anything about it at the time. And so this was in the spring of 2001, and this is going to sound really sappy and awful. But in the end I just felt Enron didn't understand me. So in June of 2001, I packed up my bags and moved to Pittsburgh to start a PhD program at Carnegie Mellon. And three months after that, Enron was bankrupt. So all of my friends and family at the time thought that I was nuts, but it turned out to have been sort of a very fortuitous decision.
And sort of the story of Enron is one that is really interesting because they did a lot of innovative things, but of course, they also manipulated this electricity market in California and kind of engaged in massive accounting fraud. There've been some interesting books written about Enron. One that is particularly good as called The Smartest Guys in the Room. So that's kind of how I became a scientist. And I was sort of drawn to working with this person at Carnegie Mellon because when I first met him, I thought he recognized the thing that was really kind of fundamentally flawed about kind of this whole process of creating an electricity market in California, which is that it was driven really by these kinds of purely economic principles that had been very, very successful in designing other commodity markets, but they basically forgot how the actual physical system works like how the power grid actually works. And so the market that was set up was really kind of incommensurate with like how the grid actually works and like how you actually keep the grid from falling apart. And this is where those loopholes kind of got opened up. This was kind of what drew me into being a scientist and into research was kind of really understanding, one, how this thing in California could have failed so badly.
And, two, kind of really understanding how do you sort of take as sort of a first principle, how this very complicated system works and kind of understand both kind of the opportunities that it creates to do different things sort of economically or technologically and where there are genuine constraints. So that's kind of what got me into science. And so I did this PhD at Carnegie Mellon and had a very kind of interdisciplinary flavor. And from there I moved up to Penn State and did some early work on kind of how do you sort of think about the structure of the power grid as a network and how do you think about trying to connect that structure to the performance of the grid?
And this was sort of at about the time that these papers around like kind of network resilience and kind of small world type of network models about this time that a lot of these papers were being written. And so I don't exactly know how he sort of found this stuff, but Cris Moore, was a faculty at Santa Fe Institute is still there. He and I have sort of been collaborating ever since about 10 years ago, discovered some of the stuff that I had written and sort of invited me to like a series of workshops that SFI had organized on the power grid as a complex system.
So we all agree that there are definitely complex features about the power grid. How do we think about them? And that led to an invitation to spend a sabbatical year at SFI, working at power grid stuff. And as they say, the rest is history. So that's the creation myth. What did I leave out?
Michael Garfield (13m 4s): The separation of the lower waters from the higher waters, ocean from the permanent, where plants come from, but seriously let's because you took us there very naturally. And also I want to discuss with you these kind of two different layers to this question, tone about the structure of the grid and its sort of physicality and its vulnerabilities in that respect. That's one area you've spent a lot of time in your research. And then the other area that you've already given a little bit more attention to in your backstory is this issue of the regulation in the markets that grow up around this physical, like I'm always bringing this up on the show, but Rice University Philosopher, Timothy Morton calls the hyper object of the power grid, which is this thing much like the internet or global warming or radiation or whatever, the phenomena that the objects that define our lives in this time are so often these objects that are so large that we cannot perceive them all at once.
And so how do we wrap our heads around, as we're talking about here, they're composed of multiple different kinds of things. I'm kind of jumping the gun here a little bit, but in your papers around electricity markets, even just thinking about the system of agents responsible for shaping the markets and shaping policy around this stuff are themselves so heterogeneous and composed of so many different kinds of actors. It's just a whole thing to try and even kind of understand how these things operate in the first place.
So again, let's back it up and let's just talk about the simple fact, quote unquote, that these are networks that the grid is this physical structure that's extensive in space and is full of interconnections and is thus subject to, as we'll link to relevant stuff in the show notes about this is subject to, as everybody listening to this, I'm sure is keenly aware, cascading failures, rolling blackouts. And that these are grids that I, you know, I've been thinking for years. I don't remember who it was that I first heard talk about this, but talking about the Carrington event of 1859, the solar flare that knocked out the telegraph infrastructure at the time, it made visible the Aurora Borealis from latitudes as low as Florida, where people were reading their newspaper at midnight by the Aurora.
And just the thought that an event like that were to happen in this day and age, we are so exquisitely vulnerable to the loss of the infrastructure upon which we depend in the 21st century. So I think the work that you and Paul Hines and Eduardo Cotilla-Sanchez have written a couple of papers about the typology of power grids and the risk of blackouts and how do we design these structures so that they are not as fragile in the face of both sort of random exogenous blackout events, but also directed attacks, like cyber warfare against the grid and this kind of thing.
Could you just introduce us to this and to the history of your work on this particular area?
Seth Blumsack (16m 12s): Sure. So yes, the grid is a network. It has things that you might identify as nodes. So things like power plants that inject energy into this network, there are consumption nodes that withdraw energy from the network you have, what are basically junctions. And these are typically going to be substations where a number of different things will come together or the voltage and the network will get stepped up or stepped down. And then there are also branches, which are kind of the power transmission lines that connect all of the nodes.
So there's definitely a physical topology to the power grid. And if you spend sort of a little time looking at just that physical topology, you'll see some kind of patterns emerge just from the physical topology that if you study networks you would be familiar with. So just the physical node edge connection topology of the power grid sort of looks structurally similar to kind of like a preferential attachment kind of a network.
And there's kind of creation myth story behind that structure as well which is the grid was not planned as sort of a single thing. Nobody sat down one day and got out a box of crayons and started drawing things on a map and so this is what the grid should look like. It's a system that has evolved over decades in response to things like population growth, electrification, the kind of increased demand for electricity.
And it's one that was really kind of built around connecting natural resources to places where you would want to actually use electricity. And the natural resources were kind of the original drivers of the grid are very different things that we think about today. This is not wind and sun. This is water for hydro-power. This is coal to produce electricity right later, natural gas or oil.
And kind of part of the story around just the physical topology of the power grid is connecting the locations of these natural resources with places where people want to use electricity which may not be the same place as where the resources themselves are. So that's part of the physical story. The other part of the physical story is kind of the level of interconnectedness that you see in the power grid. A lot of portions of the power grid are very, very dense networks.
And they're designed with a tremendous amount of very deliberate redundancy so that if like something bad happens to one power plan or one power line, then you can kind of reroute the electricity different ways. So that's kind of why we have kind of the physical topology that we have in the grid today. And it's very, very tempting to kind of try to draw connections between that physical topology just the note edge connections and for example, how a failure might cascade right through this network kind of things like that. And there have been papers written about kind of this structure function connection in other kinds of networks. And so sort of a very natural thing to like try to apply this framework or this concept to looking at the power grid. The trouble with that and the reason that that kind of framework doesn't necessarily work so well as we found, I think sort of goes back to something you said earlier about the grid being a many layered network. I mean, the grid is definitely an engineered system and it has its own sort of set of kind of like fundamental laws that drive behavior on that physical system. But it's very fundamentally one of these socio-technical systems because yes, it's an engineered system that obeys all of these kinds of physical laws, but you know, it's a system that is designed by people and has always been really a reflection of a number of kind of different sometimes conflicting social objectives.
So even apart from the kind of concern about the environmental impact or environmental or climate impact of electricity production, which has sort of really arisen over the past few decades, there are sort of always been these goals of wanting electricity to be cheap and wanting it to be highly reliable. And kind of the behavior on the power grid and sort of, for example, how things get rerouted, how flow on the grid gets rerouted, if you happen to lose a power line or something like that, it's not totally separated from the physical topology. If I'm at the end of the line and you take out my transmission line, then I lose power. That's pretty simple. But a lot of the behavior on the power grid is around choices that people and organizations have made over the years about what voltage do we want this line to be? How much power do we want this line to be able to carry? How much is it safe for this line to be able to carry? There are physical limits to those things?
There are upper bounds for how much power aligned to carry before bad things start happening. But what really drives those kinds of choices is the judgments of people and organizations. And so one of the sort of first things that Paul and Eduardo and I tried to do was we tried to say if you take kind of conventional descriptive network metrics like a degree distribution or centrality or something like that, is there a way to adapt these to the power grid in a way that makes sense sort of recognizing that the grid is this thing that has kind of a whole bunch of design criteria that go into it.
And that turned out to be sort of really, really tricky because there are all of these criteria, both physical and judgmental that go into designing the power grid. It makes it a very kind of high dimensional object. And if you want to describe its structure, then it's very, very difficult to kind of boil it down to like a handful of topological metrics. Some of those early papers were, I think, trying to do was to say, okay, like if we have to go beyond just simple topology, what is it that's important to look at?
What are other ways of thinking about structure that turn out to be descriptive or informative or connected to some performance measure, like reliability that we might care about. And I don't think it's a problem that we totally solved. I think we convinced ourselves that it's kind of a higher dimensional problem then I think we may have thought of going into it. I mean, if you want to think about this sort of question of, is there sort of a minimum set of network metrics that will kind of describe everything you want about the power grid and still sort of an open question.
I don't know what that is.
Michael Garfield (23m 25s): So this is where I get to jump in and make all of these sort of spurious analogies between your work and other people at SFI. Just listening to you talk about this, I want to call back to a couple of episodes that we've done in which these network metrics have featured really prominently. One is the one that we had with Matthew Jackson back in 2019 on social networks and the flows of not electrical power, but social power through society and why it is that you can transform the life conditions of a person by moving them out of one neighborhood and into a different neighborhood, rather than just donating a bunch of money to that family, or, Liz Hobson on animal dominance hierarchies.
We haven't had her on the show, but on Julie who was an SFI postdoc and studies organizations, human organizations where innovation and power are actually seated and how they flow through firms and academic groups and so on. And one of the things that comes up again, and again, to your point, I want to say, there's a quote from the second paper here that you did with Paul and Eduardo on topological models and critical slowing down that I just want to read this real quickly “Because connectivity loss does not directly account for cascading failure, it roughly predicts only the minimum size of the resulting blackout that many disturbances with small connectivity loss produce very large blackouts.” So I mean, this kind of relates to conversations I've had with people about how sometimes the most influential people in a culture are actually the ones that they're not very like prominent, and then they're like laid off in a wave of sort of casual releases and then suddenly the company doesn't work the way that you expected it to. And it's not clear to administration why that is.
And, you know, similarly that to call more towards Liz's work and also Jessica Flack’s work on thinking of animal societies as these computers, by which information is encoded in the structure of that social organization, that there are these grooming betas that don't look like they hold the most power in society, but then actually they're the ones that it's not the alpha male that's having the most offspring. It's the ones that when the behaviorists have gone home and they're rolling night vision cameras, it turns out they're the ones giving everybody else a back rub.
So this is a very loose analogy here, but I want to give you an opportunity to, again, talk about this and then basically why it is that, you know, like I have friends, one of the moderators for our Facebook group, Tim Clancy at Worchester Polytechnic who studies extremist radicalization and like violence and the proliferation of terrorism. You think about how it is that people are thinking about how to attack these systems and to just kind of extrapolate from your findings you're into talking about in a more general way, why it is that you say max traffic attacks appear to contribute the most to vulnerability, as opposed to some of these other attack vectors, which by the way, also include as I think you mentioned earlier, random failures.
So I don't know if that's the sort of a bushy kind of weird question, but please take it and iron it out for us.
Seth Blumsack (26m 39s): I mean, you're right that because just the power grid has this very, very multi-layered structure also because while it is this big interconnected thing, it is essentially kind of planned and operated by a large number of different actors. So the grid in New Mexico is connected to like every other state west of the Rockies and Canada, but there isn't like a single planner for that entire Western grid. So like the power grid in New Mexico and Arizona public service makes decisions about the power grid in Arizona and that level of kind of distributed planning.
And decision-making means that you can have these circumstances where seemingly innocuous failures like a failure that may look innocuous in New Mexico, just thinking about New Mexico ‘s grid may have much more drastic consequences for the grid in other parts of the west. And we have sort of seen this behavior in a number of different large-scale cascading failures. And this is sort of the other thing that is kind of complicated is that even kind of failure models don't work very well because, you know, you might have a failure that starts in one place like you have this sort of initiating event in one part of the grid. And then, because the way that the power redistributes itself is not just a function of typology, but also of all of these other engineering choices about capacity and voltage and all these things, the next thing that gets overloaded maybe very, very far away geographically. But the failure in one location is somehow connected to this failure in another location which is not geographically proximate, but one is very likely going to cause the other. And they're not the kinds of failures that the grid planner in New Mexico or the grid planner in Arizona is going to think about. And there have been some pretty creative attempts by people like Paul to kind of try to take that information in patterns of cascading failures and sort of use that to effectively kind of re-imagine the structure of the grid based on these failure, proximities, as opposed to physical proximity. I think that's a great example of when you're thinking about a system that is, this multi-layered that has, is complicated in both an engineering and a social sense that in one sense, it's almost best to almost throw out everything you know, and start with the, sort of the fundamental question of what is it that we want to learn. And in the paper that I worked on with Paul and Eduardo kind of on the, essentially like the information content of topological models and kind of understanding cascading failures, that's kind of what we were trying to do.
It was written almost in response to like a series of papers that sort of claim to uncover connections between topology and cascading failures and simulated power grids. It just didn't sort of make a lot of sense to us. And so the topology and failure paper, we sort of started that paper with the question of like, okay, if topology is not the good predictor, or it doesn't have a lot of information about how cascading failures are going to propagate then what information in the network does?
And so it's kind of through almost throwing out everything that you think you know and just starting sort of with the basic questions that leads to some of these really interesting discoveries about the behavior of the system. The really interesting thing about that paper in particular was it was written at a time when there was sort of a lot of attention being paid to identifying really sort of critically weak links in the power grid chain. So aside from random failures, if you were going to have some like hypothetical bad actor who was going to like attack the grid, where would they do it to cause the most damage. There are some counter-intuitive answers to that, but it's a question that we kind of started with this paper.
And one of the things that I think is really interesting and speaks to the organizational way in which the grid is regulated and run, there really hasn't sort of been this attempt to like get everybody in some hypothetical room and say, okay, all of the different actors that own and operate the different parts of the system, there really hasn't been an attempt to sort of get them in the room and say like, okay, where is the grid really vulnerable? What do we know about this from you people who have built and have been operating this thing for decades?
And there was a National Academy's report for the future of the power grid that came out several months ago that I think sort of made this point very nicely. There are certainly analytical needs to understand how the propagation of cascading failures and the vulnerability to both kind of random attacks, targeted attacks and increasingly I think sort of what we call like post spatially and temporally correlated failures that happened with natural disasters like hurricanes. There really hasn't been the meeting of the minds to sort of sit everybody down and say, where is this grid vulnerable and how do we do something about it.
Michael Garfield (32m 7s): So I want to get into the whole issue of this, getting everyone in the room and how decisions are made, because you've got a couple of really interesting papers about that, not only in terms of elucidating the complexity of that particular regulatory process, but also in the methods and the way that you actually conducted this research. But before we get there and because you're talking about this right now, I want to speak to this window, this portal, through which this research connects to so many other different things in systems.
There's a section where you're talking about critical slowing down. And as you said, Paul's interest in the sort of mathematical proximity rather than a geographic proximity to a cascading failure. And you bring up how Martin Scheffer we've shared plenty of his work on autocorrelation in complex systems and how you can see as things kind of lock into place with each other like we're about to go through some kind of climate disaster or a mass extinction or a market failure.
And we've discussed this a number of times on the show. Last year I think it was, we had Tyler Marghetis on, I just saw him at the musical event that SFI put on in Santa Fe last night and we were talking about his research on how you can see in the video recordings that he's made of jazz ensembles playing together, or mathematicians working out a problem on a chalkboard you can see in the autocorrelation of the various parts that they track in the video, when that like jazz band is about to do a major shift, and they're going to move into a different sort of segments of co- improvisation together.
Or you can see when a mathematician is about to stumble upon a moment of inspired breakthrough. And you say in this paper, you'll link this to again, climate models and species extinctions, but also like epileptic attacks. This is something that looking back in the literature on chaos theory, talking about heart rate variability and how you can see when somebody has low heart rate variability, they're at much greater risk of a heart attack. In the human body we're talking about electrical systems and failures in those systems.
It's not exactly the same thing, but just to give you an opportunity to talk about, again, like this is sort of related to this other thing that you brought up a moment ago about as we move from much smaller kind of local grids to these massive interstate regional organizations that are managing interconnected grids, you know, where it's like Texas has their own thing, but then like fast sections of the Eastern and Western United States are all wired in together. And just the last thing I'll add to this while we're on this sort of schizoid yarn ball of associations, you know, thinking years back about the I-35 bridge collapse and how that was a system that the bridge ended up falling and killing a bunch of people, because there wasn't a ton of redundancy in the design of that bridge. It was designed to be efficient.
And so when we just recently had Doyne Farmer and Ricardo Hausmann on the show, and we were talking about this in terms of the brittleness of supply chains and cascading failures and supply chains. So there's this thing that you're talking about here with a blackout vulnerability that is related to this sort of broader issue of how do we identify when a system is at risk structurally and then, when it's at risk, in terms of the behavior of that system, in the analysis of what it's actually doing over time, just giving people a bit of exposition on critical, slowing down, and then how you can use this to kind of bolster or enhance the way that we're thinking about this with respect to, for instance, you talk about going back to Portland, you talk about in August 10th, 1996, a power line sagged into the vegetation causing a second line trip on a neighboring line resulting in the loss of a small generator triggering along sequence events ending in the separation of North Americanian Western interconnection into five sub grids and the interruption of electric service to 7.5 million customers.
That sounds like a Jurassic Park. That's like the butterfly effect type thing, except of course the butterfly in Brazil doesn't cause a hurricane in China if you somehow like isolated those meteorological systems from each other. So anyway, and rant.
Seth Blumsack (36m 33s): So I think what you sort of have picked up on is the kind of like other reason that just thinking about topology is very limiting when you're looking at the power grid because yes, it's a network, there are physical connections. If you sever some of those physical connections, there may be consequences, but I mean, it's really like a very highly dynamic system. The power grid, or at least the power grid that we have now is basically a whole bunch of very like tightly coupled oscillators and each power plant that is on the grid is this rotating machine, it's a turbine. And it's a turbine that is rotating at a like very specific speed, like 60 cycles per second. That's kind of the standard in the United States and Canada. So all of these interconnected oscillators have to be rotating at the same speed and because they're all interconnected right via this physical network, they will exert forces on one another. And it can be very tricky to kind of try to like couple this topology to the dynamic system underneath it.
And I mean, in a way in the topology work and the critical slowing down work, we're actually kind of almost breaking them apart. With the topology work the question that we're sort of asking is essentially kind of a steady state question. So how will failures not just cascade, but how will they cascade in a very specific way? How will knocking out this part of the grid overload this other part of the grid? That's a very kind of traditional typology centric way to look at the failure question, but a lot of big failures ultimately have in their kind of root cause not just something happening to the physical topology, but a disruption to kind of the steady state dynamics of the system.
And so you have all of these rotating machines and each individual machine has some inertia. If you kind of cut off the force to that machine, it'll keep spinning for a while. And so the machine as a whole has some inertia. And so all of these rotating machines and all of the consumers that are attached to the gritter, it's kind of constantly exerting forces on one. And so when you have like a physical link in the system that gets cut, then that affects the ability of these rotating machines to exert forces on one another.
And some of them may speed up in some of them may slow down. And in particular, if some of those rotating machines speed up, then it can do damage to the equipment. And so engineers have these kinds of safeguards built in where if they sense that there's a machine that is rotating too fast, that machine will basically get separated from the rest of the grid basically to avoid like really expensive damage to that particular machine. But because all of these machines exert forces on one another and because if one speeds up or slows down that can essentially put force on another machine to speed up or slow down. The state of the system dynamics, what we call it, the frequency, the frequency with which all of these machines are rotating that can basically tell us whether we are getting close to a situation that is just dynamically unstable, where rather than it being sort of a small perturbation in one part of the grid that can essentially self-correct because you have a machine that slows down over here. There's one that sort of basically exerts a force on a different location to get it sort of speeding back up to be able to differentiate like that situation from a situation where you have one machine that might be slowing down.
And that's like the first domino that you push. And there just isn't enough force in the rest of the machines on the grid to be able to correct the first machine that was slowing down or speeding up too much. And so you sort of had some references to this and other kinds of dynamic systems they had sort of observed this behavior in the time series of the system leading up to some failure. This is a critical slowing down map. And so some of this dynamic data right connected to this very large blackout in the Western United States in 1996 was to the betterment of the research community and society made public.
And so we were basically able to access this data and look at the time series signature in the frequency of the Western grid, leading up to this really big event and see that there were these that not unlike these kinds of critical slowing down models, there was this kind of early warning signature that sort of short of shortly before the blackout actually happened here actually kind of showed up in the time series of the bridge frequency.
So that we thought was an interesting idea. The extent to which, how generalizable that is depends in part on kind of having a whole bunch of other such datasets to basically to be able to look at and there just isn't that much kind of real world data that is available to university researchers.
Michael Garfield (41m 53s): So thank you first of all, for making sense of all of that. And then in the time that we have left, we actually spent quite a bit of time on sort of the brick and mortar stuff, but there is this other, we've continued to allude again and again, in this call to the human dimension of all of this, the political dimension, and you did a number of papers with Kyungjin Yooon the political complexity of regional electricity policy formation and this question “Can capacity markets be designed by democracy?”
So, because I love depressing the listeners of our show with revealing the pickle, the so-called wicked problem of actually effecting change in these complex structures and processes upon which we depend it feels like a great time to get into the work that the two of you did on looking at the sort of largely closed door decision-making processes that determine how all of this stuff is actually going to unfold.
And I'd love to start actually kind of with the latter of these two papers, because just methodologically, I found it really interesting that the way that you decided to shape this research started, and not a lot of complex systems work that I'm familiar with start in this way with a set of interviews with the various people involved in these decisions. So you have a qualitative data gathering and assessment phase that precipitated actually like how you decided to form the hypothesis about that you were going to explore in this work.
And so I'd love to hear you talk a little bit about that. And then also about just to give us the taxonomy that you provide in these papers of who exactly is involved in voting, who are these various components of these members committees, who is it that's actually determining how changes are made to the grid? Because we talked about this in our preliminary call, technology is changing really fast, but the actual deployment of these technologies contingent on the canals that have already been carved by the history of law and the seams of power in society, who's actually holding the keys.
So laying out for us who were the people that you were talking to, and then how did you in talking to them, come to your own decisions about how you were going to understand who among those different people is actually of crucial importance when people are forming coalitions and so on in determining the future of the grid, who are we secretly trying to bribe?
Seth Blumsack (44m 38s): So why don't I actually start with sort of a little bit of backstory because the grid is a very arcane thing. The electricity industry is a very arcane world, and this is an extremely arcane sort of corner of an arcane world. So, so starting in the late 1990s, early 2000s, the federal energy regulatory commission or FERC was really encouraging electric utilities to pool their collective assets and not merge necessarily, but plan and operate the grid over larger geographic areas.
And so just for example, like in my State of Pennsylvania, we had like seven or eight utilities, each operating like little fiefdoms of the power grid in Pennsylvania and the federal energy regulatory commission really encouraged these utilities to pool together and to form these very new organizations called regional transmission organizations. So FERC the federal energy regulator really encouraged this partly for political reasons and also because of kind of just the way that electricity law works in the United States, like FERC could not force utilities to do this. It could give them lots of incentives to do so. California was one of these kind of adopters of the regional transmission organization model, where the utilities in California pooled their resources in this way. And in the kind of Eastern two thirds of the United States, there were a number of these other regional transmission organizations set up. At this point, these regional transmission organizations or RTOs are responsible for basically operating the power grid that delivers about two thirds of all electricity to U.S., consumers and businesses.
So areas that don't have these RTOs are largely like the us Southeast and parts of the Western U.S. So when FERC encouraged these regional transmission organizations to be formed, they didn't give these organizations a lot of very specific direction on what they should look like or how they should make decisions. But the one thing that they did say, as these new organizations were making decisions about planning the grid and operating the grid that they kind of had to do so in a way that engaged different kinds of stakeholders and I'll get to exactly what that means in a second. But one of the things that this means is that the model that we have a lot of the time in our head of like how the grid gets planned, there's like a utility that does this stuff, or there's some other organization that sort of acts almost like a power grid philosopher and does all this stuff. And with the creation of these RTOs, FERC directed them to move more towards a kind of quasi democratic model of collective decision-making about how different pieces of the grid should be planned and operate.
But again, they didn't tell these utilities exactly how to do this. They just said, go do it. And here's the goal for some kind of stakeholder engaged decision-making. And so they're different utilities or different RTOs in different parts of the country came up with very different ways of systems of collective decision-making. So one of my colleagues, Stephanie Lenhart, who's at Boise State has a really wonderful paper that talks about the differences in these systems of collective decision-making among different RTOs. The regional transmission organizations that operated in the Northeastern United States, so the one that covers Pennsylvania and some other states where I live, the one that covers New York, the one that covers New England, they sort of decided to have this very quasi democratic system where stakeholders would almost act like a mini Congress for the power grid with FERC kind of being the executive branch. And so stakeholders in these RTOs, when you needed to come up with a way to do power grid planning, they come up with a way to do power grid planning.
They debate it, they vote on it, and then they send it to FERC, the federal energy regulator for a thumbs up or a thumbs down. And in these three Northeastern RTOs, there's a lot of power kind of imbued in these stakeholders. And so who are they? FERC did not give these RTOs very much direction on who exactly should be a stakeholder. And so by and large, in many RTOs, the stakeholders are what you might refer to as market actors. So companies that own assets on the power grid, companies that buy and sell electricity companies that deliver electricity to consumers like you and I, large industrial companies like aluminum smelters or steel mills that use a lot of electricity kind of directly from the power grid.
So those are the organizations that essentially were sort of given, standing to be part of the Congress of the power grid in these areas. And so we started studying these stakeholders because we had gotten some of my colleagues and I had gotten interested in this process of technological change in the power grid. How do we get more renewables onto the power grid? How do we make the grid more reliable, all this other kind of stuff. And we realized that there were a lot of different organizations that were involved in promoting or thwarting technological change, but we weren't really sure how per se.
And so we thought, well, why don't we talk to them? And so we held all of these interviews with electric transmission companies, power generation companies, utilities, people who advocate for consumer interests, people who work for these various regional transmission organizations. And so we just talked to them about how these organizational decision processes worked and their experiences, and what came out of that were basically hypotheses that we realized were testable. We talked to power generation companies who complained that interests for consumers had way too much power in the RTO decision-making process.
And we would talk to the people who represented the interest of consumers. And they were like, all these generating companies, they have way too much power. And so you had these kinds of perceptions of power and influence that didn't line up. And at the same time, we realized that these regional transmission organizations actually produced fairly detailed data on how these stakeholders vote. So like every time there's a rule about planning or operating the power grid, again, it's debated and it's voted on and not to get to both arcane and political all at the same time, just at the fuddle and irritate you.
But the way that these RTOs kind of set up voting systems was to take different stakeholders and put them in camps. So like you're the power generator camp, you're the transmission owner camp, kind of all these different camps. And to set up what is called a weighted voting system where kind of each camp had a certain amount of overall political power. So the weighted voting system thing is kind of arcane. The most famous weighting voting system that most people have heard of is the electoral college.
So like the way that we elect our president in the United States is a weighted voting system where each state is like its own little sector. And each of the states have a different weight towards the overall election. And the voting process that these RTOs use is structurally the same as the electoral college. So each of these sort of stakeholder sectors has a certain amount of weight towards the overall voting outcome and inspired by these perceptions about the balance of power that we heard from actually talking to the stakeholders.
I mean, again, those were hypotheses that we could test using this voting data. And so we looked at the voting data and what we found was that at least within what's called the PJM RTO, which covers my State of Pennsylvania and some surrounding areas, because of how those different stakeholder sectors were established, they were established in a way that basically gives kind of what you might call consumer interests, the ability to basically veto any rule that they don't like.
So these interests are like a very, very strong coalition of no. On the other hand, the kind of other interests like the generation companies and the transmission companies, there were not as cohesive a coalition. And so there are more of them, but because of how this weighted voting system works, they collectively didn't have the same kind of power as the consumer side interests. And that's sort of what we found. I mean, this is something that we're continuing to look into because the process itself is kind of very multilayered because you have sort of all of these things that happened before you finally get to the vote and you may sort of wind up also not unlike Congress, you may wind up with a situation where like there are committees that tend to be very influential, but that influence is maybe hard to see just looking at kind of how Congress as a whole votes. But it was sort of a very interesting thing I think to discover that if you think the electoral college is a screwy way to elect a president sometimes also to be a really screwy way to design the power grid.
Michael Garfield (54m 11s): And yet, I mean, maybe I'm reading this wrong, but it seems like there's actually a note of optimism in this in just finding that in actually parsing and studying which of these sort of perceptions, both parties seem to think that the other side has far more decision power, and then finding that the consumer site is actually contrary to perhaps popular opinion about these backdoor deals and so on actually does have quite a bit of voting power in the way that these things are played out.
I mean, am I wrong in saying that at least in this particular limited case in the RTO that you're looking at, there's an argument to be made that, oh my God, I got democracy actually works here. At least this sort of weird electoral college, like quasi-democracy..
Seth Blumsack (54m 60s): So many times for many things, but it's a messy process that does not work terribly. But on the other hand, there are times when there are kind of very controversial issues that these stakeholders need to hash out and they just can't do it. And so there is the classic example, again, this is also in the PJM RTO, where basically the stakeholders were asked to consider different alternatives for a power generation planning group. This is the thing that in the paper we call the capacity market, you can think about it as a process for planning power generation.
So there was a way it was done and there were all of these proposals that were put forward to the stakeholders and the stakeholders rejected all of them, including keep the thing that we have now. And so the system really tends to get stuck when you have issues that are very, very controversial. It has also tended to have some difficulty with coming up with ways to integrate new technologies. And so not all of the RTOs have had this problem.
Some of them have. For example, the FERC, the Federal Energy Regulatory Commission has basically ordered all of the RTOs to come up with a way to integrate more storage into their systems. And some of the RTOs have been able to do this very easily. And others have had a lot more difficulty because storage is not just a new technology, but it's a very different kind of actor. So if you're a battery, you're both a producer and a consumer. Great. So who do you get to vote with?
Do you vote with the consumers? Do you vote with the producers? Are you a generator or are you a consumer? You're both right. And so some of these, what we call governance systems, have had a really fundamental challenge adapting to the emergence of new technology that doesn't fit neatly into these legacy categories.
Michael Garfield (56m 60s): There's so many attachment points here to other conversations we've had on the show. 'll just kind of spin the wheel and let you decide how to attack this question. But one of the things that comes up is there's a broader relationship between what you're saying and the problem that SFI has faced conventionally, as well as other research organizations that like looking through a question framing rather than through a disciplinary framing. When you're talking about your battery people, like on what side are you voting, that this is sort of a meta issue with the way that fundamental research is funded and then the way that it's communicated because if you're doing pioneering work, you don't fit neatly into a preexisting categorical system. Then it's like, well, these organizations that are set up to fund specifically biology work or a journal, the stories that people at SFI tell about the challenges of even publishing work in like biophysics or mathematical biology, is sort of a case study in how difficult it is to marshall attention and resources around these things for which we haven't all agreed. This is the record genre aisle on which we're going to file this album or whatever. So that's one thing. And it gets interesting when it comes to, I mean, just because, you know, we haven't really brought him up on this call yet, but when I had Bryant Walker Smith on the show to talk about a similar issue with the relationship between technological innovation and tort law in the rollout of autonomous vehicles and how, when you've got this car, that's no longer a car, really, it's a mobility platform that's basically a giant smartphone that you sit inside of, Lidar and all these things on the car. In some ways it's more of a surveillance device than it is a vehicle. And so it changes the way that power and agency are distributed across the public private divide, things that traffic cops used to be responsible for overseeing, or now the responsibility of a company like Ford or Tesla that is pushing out these vehicles and is directly overseeing the data collection around all of this stuff.
Should someone who's drunk not be able to get into a vehicle that's going to drive itself home, these bizarre questions. And I'd love to hear you talk a little bit more about how that plays out with respect to all of this, but then there's this other piece, and this is a choose your own adventure. And maybe we should kind of bring this conversation to a landing around now, but I just think again, one of the things that comes up again and again, in terms of the mystery that SFI and other complex systems research is confronting here is, and we've addressed this earlier in the call is that as these systems that we need to be thinking about are getting so vast now and the relationship between the information that we have on them versus the scale at which decisions need to be made.
There's this gap that we can't seem to get rid of. And in some cases it's actually growing and this is not an episode in our series for emergent political economies, but it touches on these questions that we've had like Tina Eliassi-Rad on the show to talk about democracy as a complex system science. And when we're thinking just another random shout out to SFI and MIT researcher, Jessica Transic who studies batteries and the electrification of our transportation infrastructure and electrical infrastructure, and in Jessica's work, there's something really hopeful about, oh, you know, if we can provide charging stations at these points, thinking again about all of this as just a network and like, how can we effectively roll this stuff out so that we can mitigate the amount of climate change that's precipitated through this ongoing use of petroleum-based vehicles and so on.
But then you get into this behavioral part that you just addressed, which is at some point, people can't actually gather enough information to make informed decisions about how to vote on a particular issue. And so ultimately it's not an empirical thing. It's like what Mirta Galesic studied in her work and we talked about with her in episode nine, which is that these are not rational decisions in the way that we typically use the word rational, people are making decisions based on social concerns about how much they want to irritate their friends and family and the people, it becomes a political matter where again, it's about the relationships that you're trying to maintain with other members of a coalition. And so it gets really nasty where we have these decisions that ultimately affect everybody. But then you have this sort of joint problem of not being able to relate to the data in a way that keeps the collective decision-making process from staying anchored in sort of rational self-concern, and it shrinks the horizon of like the way that people are thinking about making these decisions to something much more sort of petty and provincial.
So with all of that, and then thinking forward into, you've written other papers more recently that we haven't actually gotten into about the way that natural gas and electric power kind of overlay on each other and the rollout of the micro grids and more micro scale management of these systems, I'm just really curious to hear how all of this complexity lands in you and how you think it pertains to the considerations to bear in mind and the sort of destiny of this infrastructure.
Seth Blumsack (1h 2m 47s): So I think there's always been a very strong social side to electrons. I mean, it's as much a social system as it is a physical or engineered system, and certainly the work of people like Jessica Transic and laying out to where there are needs for and opportunities for technological innovation like what'll it take to get us there, I think all of that is sort of super valuable. The question that is really driving me right now is what kinds of organizational choices, almost like what kind of social structure do you need in order to make those kinds of technological requirements and innovation opportunities that people like Jessica talk about?
How do you make that happen socially and organizationally? So Jessica and Cris Moore and Paul Heinz and I sort of started to put this stuff together with the last power grid workshop we were able to hold before the pandemic hit, where we sort of used like a clean energy policy in New Mexico as kind of a jumping off point, talking about not just what kind of deliberate technology choices you need to make, but the social, organizational, political, even legal system that is going to have to make these choices.
How do you govern? And, you know, in particular, if you're interested in issues related to justice and equity, which are becoming kind of a bigger part of the sort of social portion of these conversations, how do you bring that into a system that ultimately has to make decisions about a very highly technical system like the power grid, This is a question that, I mean, I don't have the answer, but I think this is the question that is going to be driving a lot of what I do over the next several years.
And I think is being able to marry and integrate and make cohesive the kind of organizational piece of it with the technological piece of it, there's just a huge wide open research area.
Michael Garfield (1h 4m 49s): All right. Well, yeah, I definitely want to link to the New Mexico clean energy transition group and the document that you and Cris and Paul and Jessica helped prepare for that because it's a really not to just have wax about it, but it's a really beautiful thing to see how your team actually kind of put together a roadmap for how to cut through all of this complexity and how policy makers can start to think about how to this in a way that is not burdensome to society, but that satisfies the interests of all of these different stakeholders that makes new jobs at the same time that it helps mitigate our impact on climate and so on. At any rate, I also want to mention before we go, that you have an online course on some of the stuff that you did for Penn State, and that will link to the course in the show notes, but is there anything else that you want to bring up before we go here? And then I guess maybe the last question would just be, what else remains an animating big question for you in this work? Where is your attention focused or where do you intend to pursue your research and the next year or so?
Seth Blumsack (1h 6m 0s): So thank you for plugging the courses. There's actually two of them. I'll send you links to both of them. I think that the big animating question is still the one that is underlying the RTO governance research network which is a multi-institution research initiative that's funded by a number of private foundations. And it is really focused on these organizational questions and kind of the animating question there is how do governance choices, choices about how we make choices about the power grid, how do those governance choices manifest themselves in the behavior of the grid? So just as I started with interviews and wound up with all of this kind of quantitative analysis of voting behavior, how do you link those organizational choices and system performance? And once you figure out what those links look like, what does it tell you about the organizational choices that you need to make if you want the grid to perform differently, whether that means reliability, your sustainability or economics or any combination.
Michael Garfield (1h 7m 6s): Awesome. Anything else before we part, I mean, thank you so much for taking an extra-long time to hash this out here.
Seth Blumsack (1h 7m 12s): Very, very welcome. I do need to go make dinner though.
Michael Garfield (1h 7m 16s): Alright. Seth Blumsack. Thanks again so much for being on complexity.
Seth Blumsack (1h 7m 20s): Thank you so much for hosting.
Michael Garfield (1h 7m 23s): Thank you for listening. Complexities produced by the Santa Fe Institute, a nonprofit hub for complex systems science located in the high desert of New Mexico. For more information, including transcripts research links and educational resources, or to support our science and communication efforts. Visit Santafe.edu/podcast.