Episode 137 - Bayesian COVID Tests, Topological Computation & How the Electoral College Votes
How the electoral college votes determines which party dominates the country and for how long. This issue is crucial as the effects of the COVID-19 pandemic are foreseen to last long-term. Now more than ever, we need to hear more people weigh in on their interests and the decisions that affect their lives.
For the first time, today’s show features calls and emails from The Local Maximum audience. Listener Gar Chan learned about Bayesian inference and published a manuscript on Bayes’ rule as it applies to COVID-19 testing. Another listener, Andrew, introduced Max to a topological approach to representing numbers in computer science. A third listener, Winston Ewert, had some points to make on the electoral college.
We have lots of ground to cover, so tune in!
Here are three reasons why you should listen to the full episode:
Discover how Bayesian inference affects the interpretation of COVID-19 test results.
What are unums, and how are they advantageous in computation?
Discern whether the winner-take-all rule helps or harms the electoral college.
Resources
“Bayes' theorem, COVID19, and screening tests” by Gar Chan
Gar’s LinkedIn and ResearchGate profiles
A Radical Approach to Computation with Real Numbers by Dr. John L. Gustafson
BC’s Proportional Representation Referendum by Winston Ewert on YouTube
Talk Python To Me #239 — Bayesian Foundations
Learn Bayesian Statistics, a podcast by Alex Andorra
Thinking, Fast and Slow by Daniel Kahneman
Related Episodes
Episode 133 on topology
Episode 126 on electoral systems
Episode 125 on the Electoral College
Episode 98 with Alexandra Andorra on Bayesian Stats and Election Trends
Episode 73 with Michael Kennedy of Talk Python to Me
Episode Highlights
Caller #1: Gar Chan
Gar is an emergency room doctor and a medical toxicologist.
He reached The Local Maximum when Max appeared as a guest on the Talk Python To Me podcast to talk about Bayesian foundations.
Bayes’ Influence on Diagnostic Testing
Gar says doctors need to apply a diagnostic test to an individual’s risks and priors.
As the priors accumulate, you get proper pretest and posttest probabilities.
The interpretation of test results should not only be positive or negative.
When someone reports a negative result to Gar, he tells them their risk is lower but not zero.
Other considerations include whether or not the test is the gold standard and whether or not it’s just a screening test.
Interpreting COVID-19 Test Results Using Bayes’ Theorem
In Gar’s manuscript, he asked what the sensitivity of the RNA PCR test is.
Gar found the sensitivity, or the rate of true positives relative to true positives and false negatives, ranges from 60 to 90%. It means the test can yield a lot of false negatives.
Gar is frustrated that no one talks about it. He worries about how drive-through testing centers will inform individuals who tested negative what to do next.
A screening test should have a lot of true positives and a few false positives.
Even if you test negative, you may still contract the illness if your risk is high.
On the Electoral-Vote Website
A listener named Andrew emailed Max, suggesting that he interview Electoral-Vote’s Andrew Tanenbaum and Christopher Bates.
Max has been following Electoral-Vote since 2004.
They predicted the 2004, 2008, and 2012 elections highly accurately; however, they did not have the same success with their 2016 election predictions. Listen to the full episode to find out why.
The Topology of Unums
Andrew also said Max’s episode on topology reminded him of John Gustafson’s type 2 unums. Max has never heard of unums before.
A unum is a replacement for floating-point numbers, which can be inexact.
Instead of using floating-point numbers, unums use sets of numbers to represent uncertainty and inexactitude.
It also uses the projective real number line, in which there’s only a single infinity.
Caller #2: Winston Ewert
Winston has a series of videos about the referendum in British Columbia, Canada, aiming to change the province’s electoral system from first-past-the-post to proportional representation.
In a first-past-the-post system, a small party would have a hard time winning. On the other hand, a proportional representation essentially creates a multiparty system.
Canada’s legislature has three parties, and people are pushing for proportional representation because they want to support the third party.
Winston’s Thoughts on the Electoral College
Winston thinks the fundamental problem with how the electoral college votes is that the constitution did not specify how each state should choose its electors.
The result is that the states try to maximize their influence, notably by adopting a winner-take-all rule.
The two alternatives could be the proportional method or the district method. In both methods, a state doesn’t allocate all of its votes to one candidate.
5 Powerful Quotes from This Episode
“Somebody comes, reports a negative test to me, I'm like, ‘Oh, well. We'll have to talk to them about that their risk is lower, but it's not zero.’”
“The main takeaway is that, even if you test negative, but if your risk is high, you need to presume that you may still have illness.”
“It seems like there is tremendous power in thinking of terms of sets rather than points.”
“That's why all the states, almost all, have chosen this winner-take-all rule because that maximizes their own influence. And I think that operates to the detriment of the system as a whole.”
“As long as you don't allocate all of the state's electoral votes to one candidate and you're doing it some way based on how much support they have in your state, it ends up not being a huge effect.”
Enjoy the Podcast?
Are you hungry to learn more about Bayes' rule, topology, and electoral systems? Do you want to expand your perspective further? Subscribe to this podcast to learn more about A.I., technology, and society.
Leave us a review! If you loved this episode, we want to hear from you! Help us reach more audiences to bring them fresh perspectives on society and technology.
Do you want more people to understand Bayes' rule, topology, and how the electoral college votes? You can do it by simply sharing the takeaways you've learned from this episode on social media!
You can tune in to the show on Apple Podcasts, Soundcloud, and Stitcher. If you want to get in touch, visit the website or find me on Twitter.
To expanding perspectives,
Max
Transcript
Max Sklar: You're listening to The Local Maximum Episode 137.
Time to expand your perspective. Welcome to The Local Maximum. Now here's your host, Max Sklar.
Welcome, everyone. You've reached another Local Maximum. Welcome to the show. Today is a—well, it's a very different day on the Local Maximum. Because today, we are introducing a brand new show format here on Local Maximum. It's called a call-in show. It's not exactly a new format in terms of podcasting or radio overall, but a new format for the Local Maximum.
Previously, we've had three main formats on the show. One, we've had the show solo shows where it's just me talking for—I usually plan to talk for 30 minutes, although I've been known to talk for an hour sometimes. Then, sometimes we have kind of news updates or discussion with me and my co-host, usually, Aaron. And those, I also plan to last for 30 minutes, but those definitely last for an hour sometimes. And then the third format that I have is the guest format, where I invite people who have something interesting to say onto the show, and the show is an interview. And I have a couple of, you know, kind of rarer ones where a few times we had an expert panel on, where I interview multiple people, and then one time I did kind of a recorded talk of mine. But today—today is different.
This is a new format for me. We are going to take feedback from people like you, from the listeners. I want to hear what people have to say about previous episodes and topics that we have done. So today—and so people have sent me email, and I asked anyone if they wanted to do a call-in, and they said, “What's a call-in? Is that very nerve-wracking? Is that difficult?” No, not at all. You just put something on your calendar, I send you to a website, and I have a 5 to 10-minute conversation with you. We talk about whatever we want. And you can know if you stumble over your words, we can always edit stuff out. It's not that big of a deal anyway.
So today, we have two calls and one email that I want to go over. We are covering tons of topics today. Today's topics cover Bayesian inference, COVID-19, topology, computer science, and electoral systems, the Electoral College—all rolled into one. So tons of ground. So why don't we just get started with our first ever Local Maximum call-in show?
Our first caller came to our show through episodes on Bayesian inference and actually ended up publishing a manuscript on Bayes’ rule applied to COVID-19 testing. Let's have a listen.
[FIRST CALL]
Max: You've reached the Local Maximum. Welcome to the show. Please tell us your name and what it is that you do.
Gar: My name is Gar Chan. I’m an emergency room doctor and a medical toxicologist.
Max: Okay, very cool. And so you reached out to me and a few other podcasters, and so I'm glad that you found some good information from the podcast, or you're enjoying it at least. What episode were you listening to? Or what topic were you listening to?
Gar: I started listening to the Talk Python to Me podcast, and this is an immersion into coding in Python, which I took part of about a year ago. And I took a liking to Bayes' episode that you were in, and that's where I got started with your podcast. It redirected me to your podcast.
Max: A lot of people tell me that, and you know, I was so surprised at how popular that Talk Python to Me show is. It's a good show. I tune in from time-to-time. But man, I've never had so many people—I've been on a lot of podcasts—and I've never had more people than people who say they come from that podcast.
Gar: Yeah, you know, in terms of just getting started with Python or just getting going with it, I took a course in person, but I wanted really to immerse myself into it. So I hardly understand half of what's going on, but when they hit the base topic, I said, “Oh, I have to check this guy's podcast out,” in terms of yours.
Max: Alright. Well, thanks for coming on. And I know you found the, Alex's podcast as well, right. That was the Learning Bayesian Statistics one.
Gar: Yeah.
Max: Yeah. Alright.
Gar: So I just skipped over that once he was a guest on yours, and I just skipped over just to try to immerse myself into the idea of how Bayes is active in the world of programming and machine learning.
Max: So what did you know about Bayesian inference beforehand, and what did you, what was your updated opinion on it if I should pose this question in a very Bayesian manner?
Gar: Yeah. So I think Bayes as an idea within medicine is pretty simple in terms of diagnostic testing. And I'd sort of jumped a little bit deeper into it when I read, I think, Daniel Kahneman's book, Thinking, Fast and Slow. And he mentions another author, I think his name is Gerd Gigerenzer, or something like that, who posed the question to doctors, how to think about a diagnostic test and I think it was a very simple question.
If a 30-year-old woman has a positive mammogram, what's the likelihood she has breast cancer? And 90% of doctors answered the answer incorrectly, and mostly because they don't understand what the test means alongside of risk. You know, so the lifetime risk of breast cancer, I think, is around 3%. So any given woman, on any given day, that test positive is going to have slightly higher than 3% risk of breast cancer, even with a positive test. So you need to apply the test to the individual. You need to know what their risk is or their priors. So is there a family history of breast cancer? Is there a suspicious lump? You know, things like that. What's the patient’s age?
So as you accumulate all those priors, then you have a pretest probability, and then you apply the diagnostic test to the individual. So on any given day, if you have a random person that just takes a mammogram, not knowing their risk, it doesn't adjust their risk differently because you're using what's the incidence of breast cancer on a given population. Whereas, when you take the individual's risks by themselves, then you could have a proper prior—sort of proper post-test—sort of a risk.
Max: Yeah.
Gar: That’s when I started getting into the idea of Bayes and its influence on diagnostic tests.
Max: Yeah, I feel like I've heard that example. That specific example, I've heard before, but I usually change it when I'm talking to some random medical test. I guess now it's—I guess the last six months has always been COVID-19 test or whatever. But yeah, no, that's a good example. And I always feel like when I go to the doctor, I wish I had more of a probabilistic sense of what was going on when they're running tests and doing diagnostics and stuff like that. But I always find that that information is very hard to come by in any of my particular situations—like routine annual exam and things like that—like I never get the stats on my risk level of so and so. It's always just, “Hey, we're just gonna tell you what it is,” and maybe, I understand, maybe that's the best way to talk to the public, but I wish it weren't like that.
Gar: Yeah, it's never that simple. And I hear people talk about, with test results, all the time at work, and the interpretation, other than positive-negative is never ideal in terms of what I hear. You know, somebody comes, reports a negative test to me, I'm like, “Oh, well. We'll have to talk to them about, that their risk is lower, but it's not zero,” you know. So I think there are other things that come into play. You know, is this the gold standard test? Is it just a screening test? You know, those are things that come into play when I think about Bayes and its influence on medicine.
Max: Yeah. So, okay. So you ended up writing manuscripts that you mentioned. Can you tell us what it was called, and, basically, what you were saying in there? And also like, where it was published?
Gar: Yeah. So, you know, during this COVID sort of outbreak around the world, you hear news reports, and you see the public health teams sort of running around like chickens without heads on and trying to figure out how to sort of ring-fence this event. And I just decided to say, you know, “What's the actual sensitivity of this RNA PCR test?” And wherever I look, the sensitivities or its rate of true positives relative to true positives and false negatives is not that great. And it's anywhere between high 60% to low 90, or maybe high 80%. Whenever you use a screening test like that with those statistics, you're gonna have a lot of false negatives, which is problematic because...
Max: So, wait, was that for most COVID-19 tests or a common one?
Gar: This is specifically the most common one.
Max: Okay.
Gar: Which is the RNA PCR test, that's the swab in the nose, in the throat.
Max: Okay, so that's the one I've heard about.
Gar: Yeah, so that's the most common one. So when people started saying that we need more testing, more testing, it's great that we could test everyone, but once you get the test result, the interpretation of the test result is which is the most important thing. So with that sort of level of frustration, I looked around and said, “Is anybody talking about this?” You know, I'm in Tasmania at the moment. A lot of my peers and my mentors are back in New York City area. So I wrote one of my mentors, like, “How come nobody's talking about this?” And he's older, and he's been around much longer, he says, “You know, Gar. Nobody wants to hear this right now. Everybody, this is—you’re driven. Nobody wants to talk about the accuracy of the test.”
So you see all these drive-thru testing centers, how are they going to inform these individuals when they get a negative test? What to do next? And that was my fear, like, because the false negative rate was too high for my comfort level. But that prompted me to write this manuscript, and I shopped it around in terms of, you know, try to find who would publish it, and things like that.
So eventually, the title of the manuscript is “Bayes’ theorem, COVID19, and Screening Tests,” just as a sort of a simple sort of way of viewing test, screening tests, at least in medicine, using Bayes' idea and with this current COVID-19 testing. And it's open access. So it's available through American Journal of Emergency Medicine, and the print publication is sometime in the future, but it's currently available in open access, which is great.
Max: Yeah, I think, early on in this whole thing, I think I tried to like go on Twitter and ask about the accuracy of the tests. And I basically got back like, “You know, you should feel bad for asking.” You know, I wasn’t—Twitter's not a very good place to go for this thing.
Gar: Yeah, but within that, there's a bit of truth.
Max: Yeah.
Gar: It's not great for a screening test. You know, for a screening test, you want a lot of false positives, you want all the true positives and a bit of false positives as a first line sort of test, and when you do have...
Max: Is that...
Gar: Yeah. I mean there’s a...
Max: Do you have like what are the approximate rates for that test? I think you said before, but...
Gar: Initially, in March, some of the sensitivities of the tests were about 67, maybe 70%. And more recently, I think people are saying that the sensitivity is upwards of 90%, which is still not great or sensitive.
Max: Yeah. What does that mean? What sensitivity in terms of like false positive, false negative?
Gar: So in terms of sensitivity, you get a number of true positives and false negatives, right? So ideally, the sensitivity is your true positives divided by true positives and false negatives.
Max: Gotcha. Gotcha. Okay. So, yeah, so there's...
Gar: In your hold, there's a whole lot more to it, there's specificities.
Max: Yeah.
Gar: Likelihood ratios and things like that, but just as for initial understanding of it, sensitivity is important for speed.
Max: Okay, yeah. Yeah. So, I mean, there's a lot that could be said about that in terms of, you know, what do we do now with that information? Like, what was your main takeaway?
Gar: The main takeaway is that, even if you test negative, but if your risk is high, you need to presume that you may still have illness, right? So if your pretest probability is upwards of 80 or 90% because of where you live, who you've been in contact with, just because you get a negative test—you know, the post-test probability is still elevated. It's not zero—it's not close to zero. So depending on what your prior is, a negative test in this regard will only slightly reduce your post-test probability down to about 60 or 70%, which is, you know, greater than a coin.
Max: Wow.
Gar: So you should not feel assured that you don't have the illness.
Max: Right, well, I guess, especially if you have symptoms, that's probably…You know, if you have symptoms that match, then that probably puts you probabilistically way high.
Gar: Yeah, if you have symptoms that match, if you live in an area that has a high penetration rate, if you live in close proximity with other individuals with it, yeah, that definitely puts your prior probability elevated. So with a negative test in that scenario, it's not reassuring.
Max: Yeah, and just, you know, in personal anecdotes, I feel like, I've talked to people who have gotten tested, and I feel like the people who kind of just thought they had it because they had a runny nose once, but like never had symptoms, most of them didn't have it. And then the people who like had it on the nose exactly when and how they tended to actually test positive. Just in personal anecdote. Interesting.
Gar: Yeah.
Max: Are you, so you have to answer this, but Tasmania, is that—are you there for work or research, or just other purposes?
Gar: I came here nine years ago.
Max: Oh, wow. Okay.
Gar: Looking for a change of phase. Yeah, change your phaae, and we haven't left. The lifestyle is pretty good. It's pretty laid back.
Max: Well.
Gar. Yeah.
Max: Well, so how—I don't know anything about Tasmania. Just tell me one thing about it that I should know.
Gar: Oh. My God. It's got the cleanest air in the world.
Max: I think I know where it is on the map. Oh, what's that?
Gar: It's got the cleanest air in the world.
Max: Really? Okay.
Gar: That's what I hear.
Max: That's interesting. That's interesting. That was never on my radar of places to visit, but now I've heard it. Alright. Unless you have anything else. Gar, thanks for coming on the show, and thanks for calling in today. I really appreciate it.
Gar: Oh, thank you very much.
Max: Next, I want to read and respond to an email from Andrew. I don't want to give his last name because we didn't talk about putting this on the show. But I found it very interesting, so I'm going to read this. “When I,” he wrote, and I quote, “When I listened to your episode on topology, I couldn't help but be reminded of John Gustafson's Type II unums.” And he sent a link there, which I'll paste into the show notes in localmaxradio.com/137. He continues, “which is a proposal for alternative computational system that operates on sets of continuous numbers, rather than points on the real number line. It seems like there is tremendous power in thinking of terms of sets rather than points.” And finally, he has also, with regard to polling, “I think you should ask to interview the folks behind electoral-vote.com, Andrew Tanenbaum and Christopher Bates.
You know that, well, I'm going to start with that last one ‘cause the first part’s more interesting. But electoral-vote is a website that I have been going to since at least 2004. I believe that's when they started—2004. And they were very, very accurate in the election of 2004, 2008, and 2012 because what they did was kind of a weighted average of all of the polls out there, based on their perception of the polls accuracy, and also how long ago it was. So you kind of get a sense of which states were going in which direction. Now, in 2016, they were not that accurate because the polls were not that accurate, and you could see from the results they had at the beginning of 2016. So it's kind of an interesting question whether the same thing will happen this year, and I am definitely passing the site from time-to-time. So that could be an interesting interview.
Now, these unums, I have never heard of this thing called unums before, and I was really intrigued, so I looked into it. It's a replacement for floating-point numbers. Now, for those of you who don't know, we're not in the know in terms of computer science now—computers work. When you have kind of a decimal number, and you're doing computations—like division and multiplication and square roots and all that—on decimal numbers, they are usually represented by these floating-point values, which can get very high, they can get very low, and they usually, they're in scientific notation. You know, they have a long list of decimals, and they have exponent. But you know, there's kind of a margin of error because when you multiply two things together, you might get even more decimals, and they just chop off the end of it. So there's a lot that's kind of inexact about floating-point numbers, but they're meant to represent exact points in space. Very interesting now. They're very, very efficient.
Computers can compute with these things very efficiently, and that's why they're highly used. Because they're inexact, when you want to represent something like a currency, amount owed, you don't use floating-point numbers. Even though it's a decimal, you'll just use an integer, the number of cents, because that's exact. And so there are certain points where you don't want to use them, but many, many points where you do—particularly in machine learning—you know, when you're learning weights of a logistic regression. For example, you want to learn the parameters of your model, you are often going to want to use floating-point numbers to represent them.
Now, here is an alternative to floating-point numbers that has some neat properties. Maybe you get rid of the kind of issues where you kind of have that inexactitude there, and instead, you start thinking of sets of numbers. And so I was looking at the general idea here, and basically, what these things represent is they represent—some of them represent open sets of numbers. Let's say like the open set between one and two is any number between one and two, not including one and two.
And then, it also has values that include the points between those open sets. So you'll have one that includes the point one, and you'll have the—not point one—but the point of one. Like this is exactly one, and you'll have another one that represents exactly two, and then you'll have another value that represents all the values in between. And you kind of represent uncertainty and exactitude in that manner, and I found it very interesting. I also found it interesting—so it is using a topological view of numbers because it's open sets and boundary points.
And I also found it interesting that it uses the projective real number line, which is, so there are two types of extensions to the real number line. One is the projective, which means that the question is, “What happens at infinity?” The affine real number line says, “Okay, well, if you go to negative numbers and you go higher and higher negative numbers, and lower and lower negative numbers, you know, higher magnitude—like negative a hundred, negative a thousand, negative a million, so on—you go all the way off to the end, you get to negative infinity. And then on the other side, when you get to higher and higher magnitude numbers, you get to positive infinity. So that's the affine real number line. The projective real number line says, “No, that ends up being the same number, just D∞ to be D 1/0.” And which one do you use? Well, it's kind of a question of what your application is.
I mean, so I did an episode, I want to say, I know it was in the 90s. Let me look at my archive here. I think it was episode 90, yes, 94, where I talked about numbers in terms of ratios. When you're talking about ratios, you're only talking about positive numbers. So the location of zero and infinity, you don't have to worry about the difference between negative infinity and positive infinity. But when you're dealing with the full real number line and computations on that, then you do. And so, I think I can do now a whole show on projective spaces versus affine spaces. I think it's a really interesting topic. And the fact that this uses the projective real number line means that there's a single infinity. So what they might do is they might say, “Hey, I have four values to represent. I represent zero, I represent infinity, and then I represent all of the negative numbers, and then all the positive numbers. And then I start dividing them up from there.”
And so I just think that's a really interesting way of looking at numbers. I think it's very cool to kind of take a look under the hood of the types of computation that underlies all computation and see how can we improve it a little bit. Because if you improve it a little bit, then you improve computers across the board, even though it takes probably many, many decades to upgrade everything. So I thought that was fascinating. I couldn't stop talking about it. I wanted to share that with you.
And finally, our final caller had some comments on the Electoral College and has also spoken about proposed electoral reform in his home province, British Columbia, Canada.
[SECOND CALL]
Max: Okay, you've reached the Local Maximum. Hi, Winston, how are you doing?
Winston: I'm doing good. How are you?
Max: Good. So I know that you emailed me, and you listen to the shows on the Electoral College? Do you listen to both of them or just one of them?
Winston: I listen to both of them.
Max: Oh, very cool. Thanks for listening. So I know you said you kind of had maybe a different take on it. You had a few videos that you did on YouTube that were related to stuff going on in your area, and you wanted to maybe give me your take on the Electoral College, maybe something that I missed. So what is that? Maybe you can talk about your video first?
Winston: Yeah. Okay, so the series of videos I have were related to a referendum that was going on in my province. So I live in British Columbia in Canada. And so there was a referendum there on changing our electoral system, and I made a bunch of videos about that. The referendum went down to defeat, somewhat predictably, people are always very suspicious of changes to their electoral system.
Max: What were they trying to do?
Winston: So, they wanted to introduce a form of proportional representation.
Max: Right.
Winston: So your listeners are probably familiar with the first-past-the-post electoral system, where you divide the whole country up into little districts, and the person with the most votes in each district represents that district. The difference in a proportional system would be that, instead of doing that, you have these various ways of accomplishing it. But the end result being that you give the number of seats in the parliament or legislature or Congress—whatever the equivalent is in your jurisdiction—proportional to how much support each party gets. So if a party gets 40% of the vote, then they get 40% of the seats. The big effect there is that under first-past-the-post, a small party that gets like 20% of the vote would actually have a hard time winning any seats in first-past-the-post.
Max: Sure, sure.
Winston: Because that is not enough. But under the proportional system, they end up with 20% of the seats. And so, that has some effects on the sole makeup of the political system because instead of being restricted to a two-party system, you end up typically having a multi-party system. And so the question was, should we switch over that or maintain the first-past-the-post which we'd been using?
Max: Would that have been for all of British Columbia?
Winston: Yes. So that would have been for all of British Columbia,
Max: That would have been intro—well, I guess, in Canada, the two-party system isn't as absolute or near-absolute as it is in the US. But imagine if that would have passed, then it would have been very different.
Winston: Yes. So I mean, as it is, we actually have three parties in our legislature. So we don't have—and yeah, we are far away less dramatically two-party than the United States is. And that's part of the motivation why there's pushes for proportional representation because people want to support these third parties in a way that's not happening in the US.
Max: Right, right. I feel like allowing third parties in the US has always been a big issue, but this year, it's just each side is at each other's throat. We don't even have time to talk about that.
Winston: This year, I'd really like to have a third party.
Max: Well, yeah, that may be true, but everyone's so scared of the other party. It's like—so anyway, what is your take on the Electoral College? Obviously, well, there's a lot that can be said about it. There's a lot I already said about it.
Winston: There is. Yeah. I suppose, so what I think the actual fundamental problem with the Electoral College is that the Constitution didn't specify how each state should choose its electors. They left it up to each individual state, and I think what that's led to is a situation where each state is trying to maximize their own influence by the rule they've chosen. And that's why all of the states—almost all—have chosen this winner-take-all rule because that maximizes their own influence, and I think that operates to the detriment of the system as a whole. I think the winner-take-all rule is probably the worst possible rule that you could come up with for that, and I think that's where the problem is.
A lot of people like to complain about the Electoral College giving support to small states because they've got this “one man, one vote” thing. I'm fine with giving more power to small states if that's what you want to do. But I think the bigger effect in the Electoral College is the winner-take-all rule, and I don't think that does anything useful.
Max: So, not do anything useful, or do you think it actually does anything harmful?
Winston: I think it is probably moderately harmful. I think the effects tend to cancel each other out somewhat because you've got states on multiple sides. So it isn't terribly harmful, but I think on net, it is harmful.
Max: Yeah. So what could be done like within each state? So there's obviously—I mean, you could district up the state again or do it by congressional district like some of them do. Or you could do it proportionally in each state? That might be—this five minute call, I don't want to try to think ahead as to what all the different effects that would be. But what are you thinking?
Winston: So obviously, basically, there seems to be two versions. There's the proportional or district method. There's some variations in the district method because you've got the extra two electoral votes per state. Do you sort of create more districts that don't win at the congressional districts? Or do you hand that to the overall popular vote winner or the person who wins the majority of the other districts? There’s various variations of those.
Max: All sorts of ways you can do it.
Winston: Lots of ways. But all those different methods don't tend to have a huge effect on the outcome because as long as you don't allocate all of the state's electoral votes to one candidate, and you're doing it some way based on how much support they have in your state, it ends up not being a huge effect which one of those other methods you might choose.
Max: Interesting. Yeah. So I'm trying to remember it like in my memory bank, and I'm gonna have to look this up. And after this conversation, I'll tell whether if this is true or not. But I think actually, at the beginning, states were trying to send congressional delegations, like as a slate. Like, oh, Connecticut has six representatives. The whole state votes, and like the whole six from one party can go, and I feel like that was ruled, you know, not allowed very early on.
Winston: So, from what we understand, yes, they did that, but it was never ruled not allowed. States are still allowed in principle to select, have their legislature select whoever they want to send. There's no limitation.
Max: Oh, no, for the Electoral College, yes. But for Congress, actually.
Winston: In Congress, the house has always been direct elections. But up until the 16th amendment, state legislators used to pick the senators.
Max: Right, right. No, yeah, but I actually think this could be totally wrong. Like, way in the beginning, like the first time around, some states tried to say, “Okay, let's send a whole slate of one party for our representatives.” And then that was like, you know, the Congress was like, “No, that's not going to happen.”
Winston: So I haven't heard that.
Max: If it's true, you haven't heard it because it probably—it was like very quickly stopped.
Winston: It’s like, “No, don't do that.”
Max: Yeah, exactly. So anyway, thanks for the call. I enjoy these five-minute calls, and we'll definitely have this out before the election.
Winston: Okay.
Max: Alright. You know, he brings up a really interesting point. I do want to think more about the idea of winner-take-all with the big states because if you kind of do a reductio ad absurdum here, you kind of imagine a situation where a bunch of states are always a lock, and then that lock kind of dominates, becomes a majority, which I don't think we have. I think the swing states right now are clearly in control. But if you do have states that become a lock and become a majority, then they can dominate the country for a long time. Very unhealthy.
So I do wonder if the power of states is maybe dispersed enough for these effects. Not be too great, but it is kind of a problem to look at. I mean, like imagine if something like California were so big that it were more than a majority of the country, then the Electoral College would not be too helpful, would it? So and you know, maybe this kind of reminds me of Antebellum America or After America where the Southern States didn't even let people vote for anyone other than the Democrats for the most part. And so you did have that block that kind of dominated basically until the Republican Party came along. So I do want to think about that a little more.
How do you like that call-in show? I think that was a lot of fun. If you have comments on any of these issues, or the issues that we discussed on the show, localmaxradio@gmail.com is the email. And hey, maybe we'll set up a call if people like this format, and if you'd like to do a call. Alright. Next week. I'm going to go for some more news updates with Aaron. And I have some great guests lined up for October. I know it was a little skimpy on it in August and September, but we've got some great guests lined up for October. I hope you're looking forward to it. Alright. Have a great week, everyone.
That's the show. Remember to check out the website at localmaxradio.com. If you want to contact me, the host, or ask a question that I can answer on the show, send an email to localmaxradio@gmail.com. The show is available on iTunes, SoundCloud, Stitcher, and more. If you want to keep up, remember to subscribe to the local maximum on one of these platforms and to follow my Twitter account @maxsklar. Have a great week.
