Warning! Voting reform trainwreck approaching – need to act now to avoid the problem

The people behind the National Popular Vote (NPV) are well-intentioned voting reformers. And so are the voting reformers advocating better voting methods (e.g. range voting) to replace our present very-flawed "plurality" voting system. Indeed, these two kinds of people often are the same people. For example, Rob Richie, a prime mover behind the "Instant Runoff Voting" advocacy group fairvote.org, is listed as a co-author of the national popular vote book.

However, despite their good intentions, these reformers are about to deal themselves a devastating self-inflicted blow because these two reforms, unless careful attention is paid to the matter (which it apparently hasn't) will conflict and may permanently destroy each other.

Review: The Goal of NPV

The NPV goal is to get a large number of US states, via an "interstate compact," to agree to award all their electoral votes to the nationwide popular vote winner. The compact only activates once the number of participating states is large enough to constitute a majority of electoral votes. The archaic "electoral college" system is then bypassed and the only thing that matters becomes the nation's voters as a whole. "Swing states" become an unmourned thing of the past. This causes these main benefits:
  1. Presidential candidates no longer are motivated to focus nearly 100% of their attention to (and unconscionably divert the Nation's resources toward) a few "swing states"; instead they are motivated to cater to the entire country.
  2. All voters nationwide become equally worthy, as opposed to the present situation where voters in non-swing states are viewed as worthless "sure things" and only voters in the few swing states are catered to, rather like medieval satraps.
  3. Pure popular vote should select superior winners on average to the electoral college "chunking" system, which basically just throws "noise" into the winner-choosing algorithm.
  4. Devastating near-tie crises (such as in Florida 2000) become much less likely (we crudely estimate 10-30 times less likely) because only a countrywide near-tie causes an election-swinging crisis, as opposed to one in any of a large number of states.

The Problem

But what is the "national popular vote winner"? The proposed NPV-compact legislation we've seen nowhere defines the term! Now you might say "it doesn't need to be defined because its meaning is obvious." Wrong.

The difficulty is if some state, thanks to the efforts of voting reformers, adopts a different voting system than plurality voting, such as instant runoff voting, approval voting, range voting, or a Condorcet system. Then what?

Several massive problems then immediately erupt.

  1. The entire compact becomes undefined and any state has an immediate excuse to drop out, dissolve the compact, or act in a different manner from the other states in the compact (or argue others should too). Of course, states with motivation to do that will essentially-always exist. The problem is if, due to foolish design of the NPV plan, they also will always be fully within their rights to do so at any time. And note that it is not ok to leave the other-voting-systems issue "for later" figuring that can just be negotiated after the compact has got going rather than before. Wrong: this omission will devastate the compact, effectively preventing it from being a compact, unless settled beforehand. And note, a compact agreed on by, say, 30 states is virtually impossible to change because you'd need to get all participating states, unanimously, to ratify the changes. It'd be writ in stone.
  2. Because a compact is not useful if any state has a motive and opportunity at any time to drop out instantly.
  3. Hello supreme court litigation!
  4. A rogue state (whether a member of the compact or not) could adopt a new voting system intentionally exactly to try to cause that pathology and to throw a presidential election.
  5. A state could adopt a new improved voting system entirely innocently to improve the world, and then be accused of having done so with evil ulterior motives.
  6. Improved voting systems (at least for presidential voting) could therefore be discouraged, destroying voting reform in the USA. (And objectively, improved voting systems matter much more in terms of their democracy-improving effect, than NPV matters.)
  7. On the other hand, the whole NPV compact could be effectively prevented, destroyed, or undercut, by the "opposed" goal of voting system reform.

Our point is, any way you look at it, voting system reform is converted into a house divided against itself, unable to stand, and wildly shooting itself in the foot; and also into a political football.

What is the solution?

It sounds like the solution is simply to insert into the NPV compact, now, language indicating exactly how to deal with other voting systems. That way all signatory states know exactly what they are getting into ahead of time, supreme court litigation is unnecessary, and "voting reform" is no longer an "evil plot" but rather "a known, good, predictable, and predicted, development."

But writing that language is not trivial. Different kinds of voting systems in different states simply were not designed to be agglomerated to yield one overall "popular vote winner." The electoral college itself can be regarded as a way to do such combining in a sensible manner – but that is precisely what the NPV compact is trying to get rid of!

A number of principles need to be adopted in the writing of such language:

  1. Better voting systems have to be encouraged, or at least not discouraged, by this language, to prevent voting reform from shooting itself in the foot.
  2. We have to avoid big "discontinuities" in the method of computing popular vote count. That is, somebody's single vote should not be able to alter the "popular vote totals" by more than a small amount. (If a single vote could alter the popular vote by a million, then we'd re-create exactly the sort of "swing state" phenomenon that was the goal of the NPV compact to abolish.)
  3. Responsiveness: Each vote should have a nonzero effect on the "popular vote totals."
  4. We have to assure that the winner in a state, always gets the most popular votes and can get (at most) the number of voters in that state worth of popular votes,
  5. We have to assure each candidate in a state gets (at least) zero.

Here is a set of counting rules which obey these principles. (Actually there is a way that the proposed rules for IRV and Condorcet could disobey the discontinuity or responsiveness principles, but those exceptions are unlikely to arise and even if they do are likely to be small in their effect and only likely to affect non-winner candidates.) We recommend these be inserted into the NPV compact's language immediately:

Plurality voting:
Each plurality vote counts 1 point toward the named-candidate's popular vote.
Approval voting:
Each approval counts 1 point toward the approved-candidate's popular vote.
Range voting:
If a candidate's average range score is S (in an 0-to-R range) and there are V voters, then the candidate gets V·S/R points.
IRV (Instant Runoff) voting (and other runoffs):
First step: Count only the votes in the final, 2-candidate-only, IRV round (after all "transfers"). That is, say the final round is "Winn" versus "Secd." Give Winn one point for each final-round ballot for Winn, and Secd one point for each final-round ballot for Secd. Second step: Now, for each candidate "Latr" other than Secd and Winn, award Latr 1 point for every ballot on which he is that voter's topmost-choice, but with the exception that no Latr-candidate is allowed to get more points than Secd gets (and Latr also is not allowed to get more than Winn gets).
Condorcet voting systems:
Define the "second place finisher" as follows. If the "Smith Set" (minimal nonempty set of candidates, each of whom pairwise defeats every nonmember) consists of a single candidate "Winn," he (of course) wins and the second place finisher is whoever would have won if Winn had been deleted from all ballots. If it consists of more than one candidate, then the second place finisher is whichever member of the Smith set is pairwise-defeated by the winner Winn, but by the smallest margin. Now proceed similarly to IRV above. That is, call the first and second placers "Winn" and "Secd." Award Winn 1 point for every ballot ranking Winn ahead of Secd; award Secd 1 point for every ballot ranking Secd ahead of Winn; and for every candidate "Latr" other than Secd and Winn, award Latr 1 point for every ballot ranking Latr topmost, but with the exception that no Latr-candidate is allowed to get more points than Secd gets (and Latr also is not allowed to get more than Winn gets).
Other voting systems:
Signatories agree not to adopt voting systems outside of the above classes until and unless the compact itself is redone to allow them (and re-ratified). The compact-states agree to create a single binding committee that will decide how to handle any additional voting methods or unexpected developments if they ever arise; and advance warning to the committee of any change should be made at least 1 year before use.

Note: with IRV and Condorcet, it is best if (whenever more than one compact-state uses the same voting system) all the votes in all those states be agglomerated into one mass then counted to determine the IRV or Condorcet winner of the resulting super-state. One reason for that is that IRV is a non-additive voting system in which X can win in both state #1 and state #2, but if we combine the votes from both states then Y wins. With additive systems like range, approval, and plurality, this issue is irrelevant because the same results always happen whether you agglomerate or not.

Another (also very important) set of problems and their solution

Problem 1: Suppose the countrywide popular vote is for A, but nearly all (or anyhow some) of the states in the Compact prefer B. In that case those Compact-states would, by obeying their own rules to make A win, be acting against their own interests!

Problem 2: The compact will have no effect until and unless enough states join it to get a clinching majority.

Problem 3: Why should a state S join such a Compact?

So there is no motivation for any state to join this compact – joining can hurt it but can never help it, whereas by not joining it still gets to "free ride" and gain all the advantages it could have gained by joining!
We suggest to the reader, that if states have no motivation to join, then they won't join! And then (by problem 2) the Compact will never have any effect! Result: NPV's whole proposal is a dead dog.
Don't believe this? OK, here's why CA governor Schwarzennegger said he vetoed CA's inclusion in the Compact: "I appreciate the intent of this measure to make California more relevant in the presidential campaign, but I cannot support doing it by giving all our electoral votes to the candidate that a majority of Californians did not support." (You'd think after the biggest state in the USA dodged the compact for precisely the reason we said, the NPVers might listen to us and adopt the simple fix we propose. Just in the interests of their own survival. You'd think.)

Problem 4: If a state outside the compact adopts a new and weird voting system (not in the list above), it can still unilaterally destroy the compact!

Simple Solution to all 4 problems: Make the compact cast a bloc vote for the popular vote winner among the compact's member-states' combined population only, and make it do so no matter how many (or few) states have joined the compact. This way, nonmembers cannot free-ride. Hence states are motivated to join the compact. Further, the compact, even well before it has a clinching majority will often be able to throw the presidential election due to the power of its solid bloc vote. Even with only a few states joining the Initial Compact, it will be an 800-pound gorilla that presidential candidates will pay immense attention to. Nonmember "shirker states" who fail to unite with that gorilla, will do so at their peril; presidents are not likely to pay much attention to the shirkers. Hence the motivation to join, will be major.

What will happen to the NPV compact?

If our suggestions are adopted, the NPV compact will be like the "Borg." It will, once it gets rolling, be very hard to stop, since all will be motivated to join it. Indeed, to make a historical analogy: the initial adoption by the states of "winner take all" electoral voting (which is just like the NPV compact idea, only writ smaller) grew in a similar fashion – nobody could afford not to do it that way and hence become irrelevant.

(Serious students: please also see the Deeper analysis of compact-growth process and worries.)

But there is one way it could quite plausibly be stopped. The US constitution says:

"No State shall, without the Consent of Congress, [stuff about war omitted], enter into any Agreement or Compact with another State, or with a foreign Power ... "

They explicitly mention "another State" and "foreign Power," so clearly the first is meant to mean a State in the US and not as a generic foreign power/nation/state.

I.e, states are allowed to have compacts, but only with the"'Consent of Congress." There was a ruling that for minor agreements between States, Congress' consent can be assumed if Congress doesn't take explicit action against it. But this seems to imply that Congress has veto power over compacts – and quite possibly they would be moved to exercise it in the case of NPV!

That would get us into a delicate "state's rights" issue. Although the above quote seems to indicate clear veto power, it also is quite arguable that congress cannot veto this because congress does not have the power to tell states how to assign their electoral votes.

We do not know what would happen to that constitutional battle if fought. But let us say this. Even if such a "veto" does occur and it is sustained – shooting down NPV – then NPV could still have played the useful role of bringing voting-methods issues to public attention. That possibility makes it doubly important to implement our recommendations!


Return to main page