Implications of Supreme Court decision on ‘faithless electors’


Yesterday, the US Supreme Court issued a unanimous ruling of some significance. To understand why, you need to look at the truly weird system that the US uses to elect its presidents. So buckle up for a trip through that maze.

The first thing to appreciate is that the president is not elected by the majority (or plurality) vote of all the people in the country. While voters in an election do cast their votes for a specific presidential candidate, what they are really doing is electing members to an abstract entity called the electoral college and it is these electoral college members who vote for the president sometime after the presidential election is held, in a process that no one pays any attention to because it is assumed that they will vote according to the results of the presidential election so there should be no surprise.

Each state is entitled to a certain number of electoral college votes based on the following formula: one vote for each member of the House of Representatives from that state (which is roughly proportional to the number of voters in that state, with a minimum of one) plus two votes (for the two senators that each state gets). Since there are a total of 435 members of the House of Representatives and 100 senators, that adds up to 535 in total. Washington DC is not a state and thus has no congressional members but for the purposes of presidential elections it is treated as one and is allocated three electoral college votes. Thus there are 538 votes in all and to become president, you need 270 of those.

Each state gets to decide how to allocate its electoral college votes. Most of them choose the winner-take-all policy where every member of the electoral college from a state is expected to vote for the presidential candidate who got the majority vote in that state. But a few states use different formulas that can split the votes.

One interesting thing is that few people have ever even heard of the actual human beings who make up the electoral college because they are viewed as mere automatons who must do the bidding of the state. But they are not automatons and have gone rogue on occasion and voted against expectations. Such people are called ‘faithless electors’. This is not a common occurrence but does happen.

Three presidential electors in Washington state, for example, voted for Colin Powell in 2016 rather than Hillary Clinton and one voted for anti-Keystone XL pipeline protester Faith Spotted Eagle. A $1,000 fine was upheld by the state Supreme Court.

In Colorado, the legal outcome was different when Micheal Baca sought to vote for John Kasich instead of Clinton.

Baca’s vote was rejected and he was removed and replaced with a substitute who voted for Clinton. Baca was referred for potential perjury prosecution, although no charges were filed. He filed suit, and ultimately won when the 10th US Circuit Court of Appeals held that while the state does have the power to appoint electors, that does not extend to the power to remove them.

Washington state Attorney General Robert Ferguson told the justices that since the creation of the Electoral College, there have only been 165 faithless electors representing less than 1% of the Electoral College votes cast for president. Of those, 71 changed their vote in 1872 and 1912 because the candidate they pledged their vote for died.

“The scattered examples that remain have been largely symbolic gestures with no chance of impacting results,” Ferguson said, adding that “over the last century, no elector for a winning presidential candidate has switched votes to the losing candidate.”

There is always the possibility that in a very close election, one or a few electoral college members may go rogue and throw the election to someone not entitled to it according to the rules in place at the time of the election. The issue that the Supreme Court ruled on yesterday was whether electoral college members had to vote for the candidates that their state rules required them to or whether they could vote for whomever they liked. This had been ambiguous since there is nothing in the constitution that requires them to follow the state dictates. Justice Elena Kagan, writing for the unanimous court, said in her opinion that states can force its electoral college members to vote according to the state’s rules and punish faithless electors.

While that takes care of that problem, there is another issue where the significance of this decision is great and that is what I want to focus on.

There is a great deal of dissatisfaction with the electoral college system because it seems so undemocratic. One of the main problems with it that is well known is that someone can become president while losing the popular votes and this has happened twice, the most recent being in 2016 when Hillary Clinton got about three million votes more than Trump, and earlier in 2000 when Al Gore got more votes that George W. Bush. This is because each campaign calculates which states they are sure to get, which states they are sure to lose, and targets just those so-called ‘swing states’ that they need to win which, coupled with the votes from the sure states, will put them over 270. Hence candidates only compete in the swing states and largely ignore the issues that concern voters in the sure-win or sure-lose states. Yesterday’s Supreme Court decision puts some teeth into this compact since faithless electors cannot mess it up by deciding that they want to vote for the person who got the majority in their own state, not the nation.

But since the electoral college is written into the constitution, getting rid of it will be hard, since that would require a constitutional amendment. But there is a clever work-around that I wrote about nearly a decade ago and is what is known as the National Popular Vote Interstate Compact. According to this, states that dislike the current electoral college system can vote to assign all their electoral college votes to whichever candidate gets the most votes nationally. Who wins that state becomes immaterial. If states with at least 270 electoral college votes agree to be part of the NPVIC, then the compact gets put into action and the winner will be the majority vote getter. Since the constitution gives each state the right to decide how to apportion its votes, choosing this formula would not require a constitutional amendment, though you can be sure it will be challenged.

Currently 15 states with a total of 196 electoral college votes have agreed to be part of the NPVIC so there is still some way what to go. But there are votes on this issue pending in another 10 states that have 120 electoral college votes. Since the current system gives an advantage to Republican-controlled states, the national Republican leadership may try to thwart such moves. But on the other hand, Republican strongholds, just like Democratic strongholds, may be fed up with being taken for granted by their respective parties and their concerns being ignored in presidential elections and some may be willing to sign on. The biggest obstacles to the change will likely be the swing states who may not want to give up on all the attention they get in presidential elections.

Thus ends another tutorial on the incredibly complicated election system in the US. Hope you enjoyed the ride!

Comments

  1. sqlrob says

    The popular vs. electoral has happened at least three times, not twice. Also in 1876.

    Disclaimer: I grew up in the town Tilden’s from.

  2. flex says

    All the USSC did was affirm that the state laws which deal with electors are enforceable by the state.

    There three types of state laws which have been used:
    1. Impose a fine on a faithless elector (Typically outdated in level, $1,000 is not an uncommon fine).
    2. Force an elector to vote for the winner of the popular vote (or split according to the split of the popular vote).
    3. Remove an elector and replace the elector who will vote based on the popular vote (or split according to the popular vote).

    Not all states have enacted laws to punish a faithless elector or force an elector to vote in a certain manner.

    Neither New York or Texas have any laws restricting electors from voting in any way they choose.

    For all that, faithless electors are pretty rare. Electors are generally high ranking party members and not following the party line will likely mean the faithless elector is kicked out of the party. Electors who have devoted their career to a party generally are not willing to throw it away all that easily.

    But who knows what the future will bring.

  3. Janicot says

    I notice that there also isn’t any mention here of the skew of electoral power away from the popular vote towards smaller population-wise states since even a tiny state has the same number of senators as the largest one.
    Again, I wouldn’t expect that to often affect the results, but it is (to me at least) another annoying complication associated with the electoral college system.

  4. flex says

    @Marcus Ranum, #3,

    … another legacy of racism

    I’m not certain that this is a legacy of racism directly. Certainly racism had a large impact on how the electoral college was originally expected to operate, but the original compromise by the drafters of the Constitution was not directly linked to racism. It was, however, deliberately elitist. (Of course the 3/5th compromise introduced racism into the electoral college from the onset, but the concept of the electoral college was not directly racist.)

    The original idea was that state legislatures would select electors, wise men and true (HA!), who would meet at a designated location, deliberate among themselves, and select the best man for the job of president according to their deliberations. There was no requirement for the public to get involved at all. In fact, many of them feared public participation as they thought it would lead to a popularity contest.

    There was no overt racial discrimination (there certainly was sexual discrimination), but there was clearly some expectations that the electors would be white men of property who would maintain the status quo, and select people who would enrich them, rather than select anyone who didn’t belong to their same gentrified notion of fitness for public office. So no black men, no poor men, no men of obvious Irish, Spanish, or Italian descent, no Jews, no Muslims, no Buddhists, no Catholics, etc. Only W.A.S.P. men need apply. (Woman? No one would even consider women fit for public office. /snark.)

    The House of Representative and the Senate are now elected by popular vote. But there is no federal requirement that electors be selected by popular vote. There are federal requirements about how many electors each state gets, but it is the state legislatures who decide, according to the Federal Constitution, how those electors are selected.

    There is nothing in the Federal law which prevents a state legislature, like Florida, to choose to select the electors directly. People say that it was the USSC who decided Bush vs Gore in 2000, but the reality is that the USSC had no choice. The Florida state legislature decided that the electors selected by the Republican party would be the electors sent to deliberate in the electoral college. Each state legislature has the power to do that under the federal constitution. If you want to blame anyone for Bush vs Gore, blame the Florida legislature for nullifying the popular vote.

    These types of shenanigans can happen again. Even if enough states adopt the NPVIC, there is nothing in federal law which would prevent a state legislature from deciding that in a specific election they were not going to abide by the NPVIC. Hopefully the entire state legislature doing so would be thrown out of office, but the damage would have been done. We need to eliminate the electoral college completely.

  5. Ken Baker says

    If the electors are legally constrained to vote as the laws of the state require, in other words if the electoral votes are arithmetically determined by the state’s popular vote, there is no need for actual humans to fill that role.
    Not only that, but there is no need for the number of electoral votes per state to be an integer. If it’s a decimal it can be weighted to more accurately reflect the population of the state. That way each voter in Wyoming doesn’t get four votes to a Californian’s one.

  6. DrVanNostrand says

    @ Ken Baker

    Constitutionally, it does have to be an integer. An amendment would be necessary to change that, at which point I don’t see why you wouldn’t just scrap the whole system and go with the popular vote.

  7. Mano Singham says

    Ken Baker @#7,

    Yes, you should not normally need people except that the constitution seems to require it. Also, in a situation where there is a tie or if there are more than two candidates and none of them gets to 270, then negotiations will be needed. I believe that this is one of the reasons why this system was created.

  8. brucegee1962 says

    Actually, the entire system makes perfect sense if you start with one basic assumption: that nobody should ever vote for anyone whom they have not met personally. Local landowners vote for one of their peers to represent them in the statehouse, and then the gents in the statehouse select one of their peers to go to the capitol and get to know the presidential candidates. It would be entirely reasonable in a world where transportation was difficult, media was wildly unreliable (and many didn’t bother to read much), and political parties didn’t exist. Well, at least the part about the media is still true.
    It’s just that nowadays, it might as well be a system designed for Martians — that world is gone for good.

  9. robert79 says

    @7 “there is no need for the number of electoral votes per state to be an integer”

    I’m tempted to vote for any politician who knows the difference between an integer, a rational, and a real number.

  10. Pierce R. Butler says

    robert79 @ # 11: I’m tempted to vote for any politician who knows the difference between an integer, a rational, and a real number.

    Which makes me glad that, e.g., William Shockley, Charles Murray, and Sam Harris have not run for office in any area where you have suffrage.

    I am in process of finishing up Edward J. Larson’s A Magnificent Catastrophe: The Tumultous Election of 1800, America’s First Presidential Campaign. (The three presidential elections before 1800 were shoo-ins for the elite faction which eventually became the Federalists, once they had enough opposition to force them into becoming a party.) The story has enough twists and turns to defy quick summary, but one could argue the crucial turning point came in April of that year, when Senator Aaron Burr’s coterie finagled a majority in the New York state legislature. Because the legislature picked the Electors, that meant all the state’s e-votes would go for the Democratic Republicans (Thomas Jefferson’s sort-of party); it also meant the state legislative elections hinged entirely on candidates’ loyalty to the Burr or Alexander Hamilton factions, and had practically nothing to do with issues facing New York as a state. Likewise, most other states’ legislative elections that year depended on presidential politicking, not local concerns.

  11. consciousness razor says

    Mano:

    Each state is entitled to a certain number of electoral college votes based on the following formula: one vote for each member of the House of Representatives from that state (which is roughly proportional to the number of voters in that state, with a minimum of one) plus two votes (for the two senators that each state gets).

    It’s really not proportional – not even roughly. The 2010 Census published this “Congressional Apportionment” document. Bottom line, here’s what it would be if the 2010 census population per congresscritter were (roughly) proportional: 309,183,463 / 435 = 710,766.5816

    (If the population isn’t divisible by 435, as it usually isn’t, then a sensible approximation isn’t too far off, if you just round to the nearest whole number, as I do later.)

    But they don’t use this simple formula – literally a ratio/proportion of the two relevant numbers – because Congress came up with something more complicated (from p. 6):

    Step 1: Automatically assign the first 50 seats.
    First, each state is assigned one congressional seat, as provided by the Constitution. Then, in the following steps, the method of equal proportions allocates the remaining 385 congressional seats among the 50 states, according to their apportionment populations.
    Step 2: Calculate a list of priority values.
    A “priority value” is based on a state’s apportionment population and the number of its next potential seat. More specifically, the formula for a priority value (PV) equals the state’s apportionment population divided by the geometric mean of its current (n-1) and next (n) potential seat number:
    PV(n) = State Apportionment Population / sqrt[n*(n-1)]
    Because every state automatically receives its first seat, priority values start with each state’s second seat. The maximum number of priority values ever needed for each state would account for the hypothetical situation in which one state is so large it receives all of the final 385 seats that remain after the first 50 are automatically assigned.

    That last sentence suggests how they presumably try to justify this. If there were a state with such a large population that it totally dwarfed every other state, the idea is to “correct” it by using a formula like this, so that such things won’t happen. Because (despite branding it as a “method of equal proportions”) they don’t really want the number of seats in the states to be proportional to their populations, no matter what the distribution of people happens to be. They want it to be more like the Senate, although constitutionally it’s not supposed to be.

    Skipping over a bit of discussion about calculating in step 2, with an example….

    Step 3: Assign the remaining seats in ranked order.
    After all of the states’ priority values have been calculated, a combined list of priority values from every state is ranked in descending order. The state with the largest priority value in the list is given the 51st seat (because the first 50 seats are automatically assigned); then the state with [the] second largest priority value is given the 52nd seat. This process is continued for each consecutively descending priority value until the last (435th) seat has been filled. The state composition of the reapportioned House of Representatives is then complete.

    Like I said above, 710,767 is (roughly) the average population per district in 2010. However, the seat representing the largest number of people is Montana with 994,416 people (i.e., the entire state). The one with the fewest is in Rhode Island with 526,283 people (one of its two districts). So, the Montana district has about 140% of the average population, while the Rhode Island one has 74%.

    In summary: an overly complicated set of calculations, which give terrible results that aren’t proportional to anyone or anything, even though they’re supposed to be and are very often claimed to be. What’s not to like? Sorry, wrong question … what is there to like?

  12. Mano Singham says

    cr @#13,
    Thanks for that clarification! It is even weirder than I thought, and that is saying something.

  13. xohjoh2n says

    @13 I thought that formula looked familiar, like the one the UK used to use for assigning MEP seats in the European elections, so after a bit of googling…

    The old UK/MEP method is called the D’Hondt method. This just happens to be the same as the Jefferson method which *used* to be used for US Congressional apportionment.

    The current US method is slightly different, and called the Huntington-Hill method. The main difference is the divisor: D’Hondt uses (n+1), HH uses sqrt(n*(n+1)). They are still quite similar though, members of a whole family of methods that vary on the divisor. Several others of which have also been used by the US over time…

    …anyway there appear to be two basic ways to apply these methods. Either choose a divisor, D, such that pop/D is close to an integer and the sum of those integers is equal to the total number of seats. That gives you the total number of seats per state up front in one go. Or you use that priority value/iterative method that gives you a seat at a time, and stop when you’ve got enough -- note that this way doesn’t take the total number of seats as input, you just keep going as long as you need. I don’t know why you’d choose the latter if you already know the number of seats to start with -- unless you need a “fair” way to provide a seniority order to the seats or something. That could be possible I guess. I guess the fact that that formulation *visibly* cannot be subject to the Alabama paradox (increasing the total number of seats decreases an individual state’s allocation) *because* the seat number isn’t an input might be an important consideration: even if the one-step method also isn’t subject to the paradox, it doesn’t as obviously *look* like it can’t be. (The pre-Jefferson method, the Hamilton method, *was* subject to the Alabama paradox, hence the name…)

    The HH divisor apparently minimises the percentage difference between constituency sizes. Which sounds like it might be a reasonable goal given that perfect proportionality is unattainable and there is a minimum seat numer constraint (of one).

    (So, err, you’re summary isn’t entirely fair. The “overly complicated” setup appears to be down to needing seat-at-a-time selection for whatever reason, because there is a much simpler one-step method that produces the same results. And perfect proportionality is not possible given the unavoidable constraints of a) at least one seat per state, b) massive disparity in population sizes per state and c) a practically limited total House size.)

    https://en.wikipedia.org/wiki/D%27Hondt_method#Jefferson_and_D'Hondt
    https://en.wikipedia.org/wiki/Highest_averages_method
    https://www.coconino.edu/resources/files/pdfs/academics/arts-and-sciences/MAT142/Chapter_9_Apportionment.pdf

  14. consciousness razor says

    So, err, you’re summary isn’t entirely fair.

    Well, what do you expect? There is no way to be entirely fair. Isn’t that the lesson here? (Only kidding.)

    Wiki describes the Balinski-Young theorem, according to which no apportionment system with three or more parties can avoid all of these “problems” (mind the scare quotes):

    — It avoids violations of the quota rule: Each of the parties gets one of the two numbers closest to its fair share of seats. For example, if a party’s fair share is 7.34 seats, it must get either 7 or 8 seats to avoid a violation; any other number will violate the rule.
    — It does not have the Alabama paradox: If the total number of seats is increased, no party’s number of seats decreases.
    — It does not have the population paradox: If party A gets more votes and party B gets fewer votes, no seat will be transferred from A to B.

    Now, first of all, I don’t care if the Alabama and population paradoxes are “counterintuitive.” If that’s what makes them seem strange or paradoxical to some people, then fine … I can’t argue with that. But that’s not enough to justify a claim that they are therefore unfair.

    The Alabama paradox just seems like a non-problem, and I really don’t get why anybody would care.

    The population paradox, as it’s phrased in the excerpt above, could be misunderstood. So let me quote what was stated earlier on the page when the concept was introduced:

    The population paradox is a counterintuitive result of some procedures for apportionment. When two states have populations increasing at different rates, a small state with rapid growth can lose a legislative seat to a big state with slower growth.

    This doesn’t seem strange to me either, much less unfair.

    I also think that growth rates themselves aren’t especially relevant in this context. We want to count people as fairly as we can, because people are the types of things which have voting rights and want political representation and so forth, while population growth rates are not those types of things. So, yes, I’m definitely okay with the fact that a state with 100 people growing by 50% (to 150) would “lose” when it’s up against a state with 100,000 that grew by only 0.051% (to 100,051). The second state gained one more person than the first, which is what matters here.

    You said this:

    The HH divisor apparently minimises the percentage difference between constituency sizes. Which sounds like it might be a reasonable goal given that perfect proportionality is unattainable and there is a minimum seat numer constraint (of one).

    It might sound like it, and of course we can’t (practically) have “perfect proportionality.” But here’s where I’ll get to the point:

    A method may be free of both the Alabama paradox and the population paradox. These methods are divisor methods,[4] and Huntington-Hill, the method currently used to apportion House of Representatives seats, is one of them. However, these methods will necessarily fail to always follow quota in other circumstances.

    Breaking the quota rule is a deal-breaker for me. If the number for a state should be either an integer Z or Z+1 — it “should be” because the actual proportion is some non-integer rational between Z or Z+1 — but if you have chosen a method that must result in it having some other value outside of that range, then you don’t have a good method which is actually closer to the correct (“fair share”) value than other methods that lack this feature/bug. If the only thing you can say against the others is that they display these paradoxes, which are much less problematic, then it isn’t a convincing argument, and I don’t have a good reason to think we should prefer that method.

    I think that if the whole point is to get an integer that’s near a value between 137 and 138 (let’s say), then you certainly don’t need to pick some other integer that’s even farther away than either of those. It might be the higher one or the lower one, depending on how exactly the problem is being approached, but it had better not be anything else.

  15. xohjoh2n says

    @17: unless, (I’d say of course), the only way to get that is for some state for whom Z==0 and Z+1==1 to get 0 seats, ie no representation whatsoever. *Infinitely less* than any other (non-zero) state. Or, a *really* big percentage difference in constituency size. I think *that* is what skews some other states outside of the Z,Z+1 range. These methods are designed to get Z,Z+1 *after* the other constraints are met…

    I dunno, maybe if you considered it as “1 seat, then all *other* seats assigned proportionally”…

  16. xohjoh2n says

    @17: also, being *seen* to be fair is probably at least as much of a consideration as anything else.

    (Remember that whatever rule is used has to be passed by the very same politicians that are subject to it, and they are most likely not great mathematicians or statisticians, though they are most likely able to understand a chart provided by one of their advisors that shows whether they *personally* would get voted in or out under the new rules…)

    Quota rule: that just *looks* fair to anyone looking at it. Violate it without a damn good excuse and risk revolution!

    Alabama rule: if the total number of seats goes up, but our population and everyone else’s stays the same, how can our representation go *down*? (You can try arguing that your representation was already above par and this is just getting you closer to the mean, but that is going to be a *really* hard sell. Better to construct a system where the question does not arise.)

    Population rule: for this to even be a problem I think you have to have mis-stated the process mathematics involved. If you just sat still, or even decreased population, and another state grew lots, you would reasonably expect your share of representation to go down. And with a fixed number of seats, that means an absolute loss of seats. Relative growths of disparate populations is just a side-screen for the real issue. (And yes, if half your population decides to move or dies, your relative representation is going to go down. If you stay still, but others go up, same thing. If you go up, but not as fast as others, *same thing*.)

  17. consciousness razor says

    I dunno, maybe if you considered it as “1 seat, then all *other* seats assigned proportionally”

    But that would mean you just calculate the arithmetic mean using only the 385 remaining seats, rather than 435. That doesn’t sound so bad to me, but that’s not what we’re actually doing.

    Alabama rule: if the total number of seats goes up, but our population and everyone else’s stays the same, how can our representation go *down*? (You can try arguing that your representation was already above par and this is just getting you closer to the mean, but that is going to be a *really* hard sell. Better to construct a system where the question does not arise.)

    Well, the number of seats isn’t increasing or decreasing. Hypothetically, it could in the future; and if it did, we could set it to a new value that won’t have that property. (There may be some other values which do, but not all possible values.) You don’t need to alter the method itself to accommodate this; just divide people into groups of some other size which is more convenient for our purposes. There’s nothing mathematical behind the current choice of 435, for instance — the value is totally arbitrary and can on a whim be set to any number we want it to be, but that’s not how we should think of the procedure for determining apportionment.

    But anyway, even if somebody considers it a problem (rather than “this gets us closer to the actual mean, which is a good thing and not a bad thing”), it’s only a hypothetical one. It has no real-world practical consequences for the time being, since the number of seats isn’t changing. In contrast, fixing the way congressional apportionment works so it’s fairer, even if not “perfectly fair” which we can’t really achieve, does have practical consequences that people should care about. (So does getting rid of the electoral college altogether, along with making various other worthwhile changes.)

    If you go up, but not as fast as others, *same thing*.

    As I said, I don’t care about how fast a state grows or shrinks. (Even if you do, that’s changing constantly, and it’s pretty wacky to be so fixated on what’s calculated every 10 years by the census bureau.)

    I just want people (in states as well as territories) to have representation that is as close to equal as possible. How the population changes over time just doesn’t have anything to do with it.

  18. xohjoh2n says

    @20

    I just want people (in states as well as territories) to have representation that is as close to equal as possible.

    And presumably that means you don’t think the current system is “as close to equal as possible” (considering just the sub-issue of seat apportionment and ignoring the issue of representation of territories/DC), therefore some other system is possible that would be more equal.

    Without having to come up with that system itself, could you even give an example seat count that would result in “more equal” representation, in the example you already quoted?

    …because I suspect you may end up there in a situation where you can’t actually do that without breaking some other constraint such as “integer number of representatives” or “representative represents a specific non-overlapping group of constituents, who are residents of a single state” or “total number of representatives is such that a floor vote takes less than 6 months to complete”. Once you’re in the position of fixing the US by making it not the US any more and turning it into a different country, I think you’re so far into “never gonna be able to sell this” that I really don’t see the point…

    How the population changes over time just doesn’t have anything to do with it.

    Hmm. I *really* don’t understand how a system which doesn’t change after a relative change in population can possibly offer “close to equal” representation both before and after that population change…

  19. consciousness razor says

    Without having to come up with that system itself, could you even give an example seat count that would result in “more equal” representation, in the example you already quoted?

    …because I suspect you may end up there in a situation where you can’t actually do that without breaking some other constraint such as “integer number of representatives” or “representative represents a specific non-overlapping group of constituents, who are residents of a single state” or “total number of representatives is such that a floor vote takes less than 6 months to complete”.

    I’m sure you understand that it’s extremely easy to get a better approximation by simply increasing the number of districts, because that number does not logically/mathematically need to be fixed (although it is determined by law, so our legislators would need to actually legislate, which they do on occasion).

    It would not need to increase to such an extent that a floor vote takes an exceptionally long time. That would require an extremely large number of districts/seats, while even one or two hundred more would not make a big difference.

    The smallest state in 2010 (the census population) was Wyoming with 568,300 people. Since that’s the smallest and will need to have at least one representative, you can think of every other state as some multiple of this (not necessarily an integer multiple, obviously) which must therefore be greater than 1. So there will be no concern whatsoever that somebody will end up with 0 representatives. That means there’s no need for some extra step at the very beginning, in which we first allocate (non-proportionally) 1 representative to every state, and then we start using a completely different formula to figure out the rest.

    309,183,463 (the total population) / 568,300 (the smallest subgroup) = 544.0497

    If there were something in the neighborhood of 544 representatives (for the 2010 case), that would not be an unreasonable number, and the extra 109 representatives would not be such a bad thing like you suggest.

    I bet it would be more in line with what other countries do, in terms of how large their legislative districts are. Note that nearly every country is quite a bit smaller than the US, and I’m also sure they haven’t all decided to keep the number of districts fixed for more than a century. That is another reason to think we should probably increase it.

    Also, people tend to like having more localized districts, not ones where a larger number of people (spread out over some larger, probably gerrymandered, district) have a single rep, even though they may not be demographically very similar. They may have a different set of interests, because they aren’t really living in the same circumstances (e.g., urban vs. rural areas and whatnot). Looking at it globally rather than locally, what I’ll call the political importance of each individual district is reduced, because the whole is being divided into more parts. That’s generally good, because one with a particularly nasty rep can’t have as much of a negative effect on what happens to the whole country.

    Actually, Huntington’s original paper is pretty good at explaining why it does what it does

    He just asserts without argument that the Alabama paradox is something we must avoid. Maybe this was just a hot political issue at the time that seemed to need no further argument; but like I’ve tried to explain before, I really don’t get where that’s coming from at all.

    The paper does nicely prove certain mathematical facts about the various methods he examines, but he’s not even trying to make sense of a question like how many representatives we should have, in order to obtain a certain degree of accuracy with respect to what he calls the “quota.” He’s dealing with a different set of questions about different calculations that could be done, which may be important, but we don’t have to think that’s the only stuff there is to care about.

  20. consciousness razor says

    If there were something in the neighborhood of 544 representatives (for the 2010 case), that would not be an unreasonable number, and the extra 109 representatives would not be such a bad thing like you suggest.

    Although I’m primarily concerned with the House here, not the electoral college, because I think the EC should simply be abolished, I should also point out that an increase like this has the effect of reducing the importance of the Senate in the EC count.

    There would still be only 100 Senators (plus 2 more “Senators” and 1 “Representative” for the purposes of the EC which represent Washington, D.C.). With the current total of 538, the Senate part makes up 102/538 = 18.96% of the total electors. That’s not proportional in any sense at all — it depends only on the number of states plus D.C. (x2), not the number of people.

    If there were instead 544 Representatives, that means the EC would have 544+100+3 = 647 electors. Then, the Senate part, which is still 102, would make up only 102/647 = 15.77% of the total electors. That’s obviously better in terms of making the EC represent the population more closely. If it were 0%, because the Senate wasn’t a factor at all, even better. Or ideally, there would be no EC, because we would just count the popular vote like normal people.

  21. xohjoh2n says

    I’m sure you understand that it’s extremely easy to get a better approximation by simply increasing the number of districts, because that number does not logically/mathematically need to be fixed (although it is determined by law, so our legislators would need to actually legislate, which they do on occasion).

    Absolutely, though that isn’t quite as simple as you make out. For having increased the number once, increasing it again will again increase fairness, so who is to say where to stop. Once we reach a number *you* are happy with, what is your answer to the person who says “no, that’s *still* not fair enough for me, we need to go further”. (Or the one who was happy to increase a bit, but thinks you gone just that *little* bit too far.) The unwieldiness of proceedings like the overall fairness of allocation has no sharp edge, so I don’t think you could construct a solid argument for any one number against its neighbour, but someone does have to choose a specific number and stick to it. 435, like it or not, is the number they have chosen for now.

    But you still have the argument that even given a specific number of seats, they could be chosen more fairly than the current method.

    (Firstly your 2010 census numbers appear to be a little off. If I use wikipedia’s table I get 308,144,382 / 563,626 = 546.7. I split the difference between that and yours and call is 545. I could run a whole series of different numbers but I just can’t be bothered…)

    Something I noticed in playing with the numbers, which I think is fairly obvious when you think about it: any unfairness tends to be loaded fairly strongly on the low-population states. You never need to worry about California being particularly hard done by, since the large seat allocation divides down any error in their allocation versus their fair quota. In particular the sharpest transition is from the highest population state to still get an allocation of just 1 seat, to the lowest population to get 2 seats. That’s not a full global view of how the unfairness distributes across each state, but illustrative I think.

    If I redo HH on 545 seats, I get a min/max population/seat at that boundary of 407,090 / 710,231.

    If on the other hand I follow your suggestion and try to assign “more proportionally” (basically ROUND(statepop/minpop, 0) which is my best guess at something you might mean), that range becomes 448,967 / 814,180. (So the bottom end is about 10% worse represented, but it was already more overrepresented than that. But the top end which was already underrepresented gets 14% worse representation still under “your” system.)

    I think your intuition about what is or results in fairness could be failing you there.

    He just asserts without argument that the Alabama paradox is something we must avoid. Maybe this was just a hot political issue at the time that seemed to need no further argument; but like I’ve tried to explain before, I really don’t get where that’s coming from at all.

    You’ve asserted without argument that “close to equal” representation is good. Most people would agree, I think, but we don’t *have* to. We could decide that Californians are just better than anyone else and give them half the seats up front before allocating to anything else…

    Digging around a bit further I think the condition that the Alabama paradox violates is called Resource monotonicity, decribed as “if there are more resources to share, then all agents should be weakly better off; no agent should lose from the increase in resources”. Now that sounds quite reasonable to me, and apparently to enough others to make avoiding it a consideration in setting the law. I hope you can see why Alabama themselves would be put out: after all if you add just one more seat they cannot expect that it would go to them over all other states, but why on earth is another state now given a seat that had *already been theirs*?

    And contrary to what I said above (“You can try arguing that your representation was already above par and this is just getting you closer to the mean”), the wiki page for AP has an example that shows that the reallocation can happen to a state already underrepresented and make them even *more* unrepresented -- that *surely* you cannot agree to since it goes against the principle of equal representation which you do agree to.

    https://en.wikipedia.org/wiki/Resource_monotonicity
    https://en.wikipedia.org/wiki/Apportionment_paradox#Alabama_paradox

  22. consciousness razor says

    The unwieldiness of proceedings like the overall fairness of allocation has no sharp edge, so I don’t think you could construct a solid argument for any one number against its neighbour, but someone does have to choose a specific number and stick to it. 435, like it or not, is the number they have chosen for now.

    I’m not arguing that there is any sharp boundary where “fairness” begins or ends. I claim that this is an arbitrary choice, and that we can make things more fair with a different choice (one of many in a fairly large range of choices, about which I’m fairly indifferent).

    I also realize that states will typically grow in the 10 years between each census, so what we could reasonably expect to get is something that will work reasonably well over that entire period (but not perfectly well or to extremely high precision at any given moment). A certain level of precision is simply of the question because the census is only so accurate and only happens so often — meaning it’s an illusion to think that you would be able to improve on this by increasing the number of reps more and more (until the whole population is reached).

    The number “they have chosen for now” is the one we’ve been stuck with for over a century, despite having a much larger population and despite changing numerous other things about how our elections work. The idea that we need to stick with it, just because and “like it or not,” is too dogmatically conservative for me.

    Firstly your 2010 census numbers appear to be a little off. If I use wikipedia’s table I get 308,144,382 / 563,626 = 546.7.

    I used the numbers from the census report linked in comment #13. According to that table, the resident population of WY is 563,626, but they also need to add the overseas population which is 4,674 for WY, giving a total of 568,300. I guess the wiki table you checked doesn’t take the overseas populations of the states into account.

    Digging around a bit further I think the condition that the Alabama paradox violates is called Resource monotonicity, decribed as “if there are more resources to share, then all agents should be weakly better off; no agent should lose from the increase in resources”. Now that sounds quite reasonable to me, and apparently to enough others to make avoiding it a consideration in setting the law.

    Maybe if you have a “resource” like drops of water or whatever, so the bits are small enough that the total can be treated like a continuous quantity without too much of a problem. Being off by one of those tiny bits may not even be noticed at all, much less would it make a big difference in anyone’s life.

    However, it’s a different story if it needs to be a fairly small integer that everyone is sharing. And several hundred representatives is definitely small relative to the number of drops of water on Earth (or any body of water large enough for many people to share, so that you’d begin to have a meaningful economic/political issue). Failing to appreciate this difference in two kinds of “resources” doesn’t sound reasonable to me.

    Also, since the number of reps can be set at will to be any number we want, it’s really not like drops of water, which are just natural objects existing in some amount independent of what we’re able or willing to choose.

  23. xohjoh2n says

    Failing to appreciate this difference in two kinds of “resources” doesn’t sound reasonable to me.

    I’m not sure there is such a difference in this case. (In particular, if the resource is so finely dividable as to make a fair apportionment trivial and precise, RM hardly enters into the matter.)

    Adding a new seat and having a state thus lose an existing seat *clearly* makes that state worse off than before.

    The fact that adding another seat *might* (but might not, you might need to add several…) give them that seat back cannot remedy the fact that the method is now proved to treat them unfairly, and will be assumed to be doing so at all times even when those effects are not visible.

  24. consciousness razor says

    I’m not sure there is such a difference in this case. (In particular, if the resource is so finely dividable as to make a fair apportionment trivial and precise, RM hardly enters into the matter.)

    That’s why it isn’t safe to make the sometimes appropriate assumption that doing it is such a mathematically trivial thing in practice, which will have the kind of familiar and intuitive properties that people expect and can take for granted much of the time. When it isn’t like that, because it isn’t a fine division and you need to deal with a small integer instead, then you shouldn’t have to think about this “resource” in all the same ways.

    You’re saying such concerns may not come up often/typically, in the other type of case (with a legit resource like water, instead of a number we can set to anything we want), which may be true, but that doesn’t mean everything about the analogy carries over to the case we’re actually interested in. That’s just the sort of thing which makes the analogy a problematic one; it’s not a coherent reason to dismiss the problems that the analogy has.

    The fact that adding another seat *might* (but might not, you might need to add several…) give them that seat back cannot remedy the fact that the method is now proved to treat them unfairly,

    No such thing was proven. Doing so can give an allocation that is more fair, not less. And that means the overall distribution is such that the apportionment method gives something closer to actual ratio of each part to the whole.

    If you have a sequence that approximates pi, for example, it merely converges to pi in the limit. It isn’t equal to pi short of that, and nobody does (or should) pretend that it is. It’s an approximation; we should be fully aware of that and be satisfied with it when such an approximation is an appropriate thing to do. Anyway, there’s no requirement that the values need to monotonically increase or decrease when more terms are added. As you add more terms — or analogously, add more seats to represent a population to get a better fraction — it will in fact get closer to the actual value of pi. That’s just how it works, and there’s nothing wrong or imprecise or unfair about that. It’s a particular way of getting more precision or an outcome that is more fair. And it’s usually not even too difficult to understand why it works the way it does. In any case, it does work.

    But to me, it sounds absurd that we would violate the quota rule instead, in order to appease people who think that we ought to be avoiding the Alabama paradox at all costs. You even said above that this risks political revolution, which for some reason hasn’t deterred you from defending the current approach, but I still don’t see a “damn good excuse” like you said there should be.