Return to main page

Preface.   BROWN v. SMALLWOOD has been cited by several voting groups to justify their (rather opposed) views, e.g.the MN voter's alliance argues it proves IRV is unconstitutional while claims the exact opposite. Here, we try to take an unbiased look. Our view is (a) this case should have no impact on anything because its reasoning was bogus and should be regarded as invalid; (b) but if the case were taken seriously then IRV is probably also unconstitutional (and we have no doubt the 1915 Court would have invalidated IRV voting too, and the election in question would almost certainly have had the same "invalid" result using both Bucklin and IRV); (c) IRV perhaps should be regarded as unconstitutional, but if so for reasons mostly unrelated to the Court's reasoning in BROWN. Perhaps the poor reasoning was due to corruption or perhaps to the fact that very little expertise on voting systems existed in Minnesota in 1915.

Brown versus Smallwood 1915

By Abd ul-Rahman Lomax & Warren D. Smith

I examined the Minnesota Supreme Court's reasoning in Brown vs. Smallwood (1.7Mbyte pdf, 12 pages, from 1915) ruling the Bucklin voting system (then used in Duluth) "unconstitutional" under sections 1 & 6 of article 7 of the Minnesota State Constitution.

In my opinion their reasoning is corrupt. I'm not a lawyer, and there are certainly constitutional issues raised, but this decision reads like one driven by a conclusion. I'm a fan of good judicial writing and reasoning, and this is not an example. The pdf includes the election results and a sample ballot from the election under consideration.

First of all, sections 1 & 6 of article 7 of Minnesota Constitution obviously do not contradict Bucklin voting, nor indeed do they have anything to do with voting system design at all except by the furthest stretch of the imagination. The court therefore keeps informing us of what the people who wrote Minnesota's constitution thought but did not say (which they keep deducing via mental telepathy). That kind of method of argument could be used to draw any conclusion. The MN constitution in section 6 explicitly defers to the US federal constitution, but the court did not note, for example, that the initial few USA presidential elections employed Approval Voting, nor did they note that James Madison (one of the US constitution's authors, indeed often called the "father" of the constitution) had in 1824 proposed a system very much like Minnesota's Bucklin system for use to elect US presidents in the Electoral College. Evidently Madison did not think his proposal unconstitutional.

This is not to say that Brown v. Smallwood is not worth reading. It is interesting. It is noted in it that it is making an unpopular decision. This alone is interesting – Bucklin was popular!

The election also is a counterexample to the claim that Bucklin proved that "most voters would not cast second-place votes," on the grounds that they would injure the chances of their first choice. Obviously (see table below; also we have other historical Bucklin election-examples), a lot of voters did not think that way.

The Minnesota court seems to think Bucklin contradicts this version of "one man one vote" that they quoted from a Michigan Court case:

Each vote shall be of as much influence or weight in the result... as the vote of any other elector.
They also quoted from a decision in a North Dakota case, from a concurring opinion but dissenting on certain arguments in that case – it was about cumulative voting! The quote reads
"Our system of government is based upon the doctrine that the majority rules. This does not mean a majority of marks, but a majority of persons possessing the necessary qualifications, and the number of such persons is ascertained by the means of an election."
We believe Approval and Range voting, and many Condorcet systems, all obey the principles expressed in these quotes, certainly at least to a greater extent than plurality voting. All of these systems treat every voter exactly the same. In all of them, each voter can express an opinion on every possible candidate-pair, giving each voter a priori equal power to decide which one of the two will win.

In the plurality system (which the Court demanded reverting to) voters cannot express opinions about candidate-pairs, so that, e.g, a voter who thought Windom superior to Smallwood is prevented from having any influence on the question of which one wins, if they thought Louisell was the best candidate. The only voters who get to affect this decision are the ones who thought Smallwood or Windom best, which obviously can be a very biased subset of the voters, and also in principle a very small subset too. (For example, if Windom and Smallwood were big fans of capital punishment, but Louisell was not, then the point of view of the anti-death-penalty voters would be tossed in the garbage and never counted when deciding between Windom & Smallwood! Only the death-penalty fanatics would have their votes counted for this decision, and they could, if it were a 5-candidate election, be only 41% of the voters.) Bucklin at least tries to allow every voter to influence every pair, and while it does not succeed, it comes closer than plurality. It certainly is ludicrous for the court to insist Bucklin is illegal in order to demand plurality voting. In IRV, voters may think they can influence every candidate-pairoff because their vote rank-orders all candidates; but actually some voters who say A is better than B in their votes will be counted at the moment when IRV eliminates B and not A, while others, who might say A is worse than B, will be uncounted. Therefore, IRV can refuse to elect a candidate who is preferred over every rival by a large supermajority – because enough of that supermajority's votes are ignored. That seems quite unfair, capricious, and not what those voters were expecting.

To return to Brown v. Smallwood, the majority decision continues:

In the case at bar, it may be noted that the number of persons who voted were 12,313, and the number of cross marks considered on the plurality election were 18,860. It was not a voting of man against man.
So, after quoting with apparent approval the prior opinion, it then proceeds to analyze the "case at bar" based on the number of cross marks, a measure which is irrelevant to the issue. The "number of cross marks" is the total of all votes cast at any round, and this number is not the relevant quantity for the determination of the winner because many of these marks represent what become moot votes, i.e. were never used.

It's fascinating that they are asserting the principle of majority rule, but are completely content to accept a decision in an election which is not the decision of a majority, thus violating the principle enunciated. In the case at bar, there was no immediate majority winner. The results were:

Canddt1st Choice2nd ChoiceAddlChoicetotals
Louisell 992 734 402 2128
Norton 3417 1501 167 5085
Smallwood 3496 2845 240 6581
Windom 4408 604 54 5066
totals 12313 5684 863 18860

It seems obvious from these ballots that Smallwood was the best one to enthrone as "winner" and that almost certainly he was preferred over every rival by a voter majority. But the court invalidated his election!

Minnesota's variation on Bucklin required a single first choice vote, a single second choice vote, then any number of desired third Additional Choice votes. Nice, in some ways. (And our Bayesian regret studies indicate the Bucklin system is superior to plain plurality voting on average.)

The legal arguments given by the court majority were complex, and I have not had time to follow all of them. However, there was a dissenting opinion written by Judge Hallam, and it makes this simple point:

For purposes of this case it may be conceded that no voter can give more than one vote for any candidate. The legislation before us does not do this.

The city of Duluth filed a motion for rehearing before the same court. The court, in response, admitted that the interpretation the same court had (earlier) given to the Constitution would not be accepted by "five out of six competent lawyers"! So the court apparently was aware that by declaring Bucklin voting unconstitutional they were creating a new interpretation – but they merely proceeded to re-justify it, again by interpreting "vote" in a way that makes the marks important, not the intention of the voter to choose a candidate over others. It was a faultily-reasoned decision which was bull-headedly insisted upon by a majority.

From the votes in this election, it is clear that Smallwood was the equitable winner; injustice was done by reversing that result. It is practically certain that if this had been top-two runoff, Smallwood would have been elected. This election was the equivalent of collapsing top-two runoff into a single election (since the third round did not alter the result), and the court apparently would have been happy with a top-two runoff.

The decision was "conservative" in the worst sense: inventing new constitutional notions, to prevent a fair and reasonable reform.

Return to main page