It's important to know what the five goals of the proposed top-two, primary are (from the text of Measure 65) and why they aren't satisfied by the measure. The last goal is even mathematically false.
1. All Oregon voters should have the full and equal ability, at every election, to choose those whom they believe are best suited to govern them.
Section 6 of the measure changes the requirements for access to the "top two" (called "voter choice") primary to match non-partisan primaries.
Thus "equal" ability would be true -- but not "full" by any measure.
By moving more deadlines for candidate nominations for third parties and independents to the primary date, it becomes more difficult for third parties and independents to recruit candidates who've been disenchanted by the major parties after seeing how candidates have changed their rhetoric after the primary -- and that's for parties that falsely play to another party's base. For major parties that don't play to a third party base (or to an independent with a special interest), the candidates providing the increased range of debate are essentially shut out of the general election and relegated to the primaries -- those candidates who propose new ideas almost always start out unable to win majority support against a field of establishment candidates.
Secondly, the only reason the major parties in the past have played to another party's base (for example, sometimes Democrats try to take Green votes with empty greenwashing rhetoric) is to secure their own party's nomination -- which wouldn't even happen in the open public with this proposal -- it would happen instead at their private endorsement meetings. It thus further limits the debate of non-top-two candidates to the primary election, where less voter turnout is expected. Yes, they would have the same prima facie ability, but as it turns out, it won't be the same "real" ability. Later I will show that in fact, they will have even less ability due to election anomalies exhibited in top-two elections. You can also see my separate analysis of Washington's experiment with "top two".
2. Competitive and open elections that encourage thoughtful debate and maximize participation are healthy for democracy and strengthen citizens' trust in their government.
The measure does open some elections (but not all) to independent candidates better; however it raises requirements for third parties. An additional signature drive or filing fee will be necessary for third party candidates above and beyond the existing signature drives, registration drives, and/or vote results that are already required for third (and major) parties. Moreover, third parties that qualify through the results of state-wide general elections will be eliminated from ballot access under this proposal.
It also raises requirements for major parties, to actually hold separate elections where they do "voter choice" endorsements. Such endorsements would typically be done through an extra level of representatives (people who attend their conventions, particularly at the state level). The type of choices that would get the nod of each party would then be more of those of the "apparatchik" and not the party membership. Otherwise, the state parties would have to seek more funds for a third level of plenary elections, increasing the cost of elections.
3. Citizens should be able to register and affiliate with any legal political party, or none at all, according to their beliefs and without any coercion or diminishment of their rights as voters.
Quite simply, this problem could be fixed directly by repealing the Kate-Brown-championed HB2614(2005) that changed the requirements for independents to be independents-only without a corresponding reduction in the signature requirements, or with a wholesale reduction in the signature requirements to be drastically fewer to be more, say, fewer than those needed to create a party in the first place. Nader created a "Peace" Party solely to get on the Presidential ballot, due to the huge difference in the level of difficulty. Secondly, to complete the repair, we would need to have the state pay a proportional subsidy to each third party or independent based on their registration base to fund or pay back the costs for third party primaries or party or independent elector conventions.
Alternatively, the state could run independent and third-party primaries where for the independents, none-of-the-above could be used to ensure not just anybody got to the general election.
Lastly, the "voter choice" elections it proposes doesn't even apply to all elections, for example, for President, so the "any" qualifier is simply false.
If the backers of this measure truly believed in this statement, they'd support Instant-Runoff Voting and Proportional Representation. If the Green Party registers ten percent of the population, we can't expect nine seats in the legislature because we don't have true Proportional Representation -- THAT would be cleaning up a diminishment of our rights as voters: to be properly represented in the legislature. The top-two proposal certainly does not.
4. Political parties should be able to endorse and support any qualified candidate, or none at all, according to the beliefs and choices of their members and without any compulsion or diminishment of their rights through operations of the law.
This measure would equalize the ability of parties to endorse candidates by removing the ability of parties to use the primary as a form of subsidized publicity for their candidates, currently unavailable to minor parties and independent candidates, as I've mentioned. Moreover, the measure actually implements fusion voting to the extent possible by allowing each endorsing party to endorse a candidate and have it listed. Of course, we're already able to do such endorsing as it is, as there won't be any partisan nominations at the general election, nor the primary election and form SEL430 and the voter pamphlet have dedicated space and subforms for many-to-one endorsement lists. Also, in section 5.1, it leaves open the possibility of charter-rule top-N partisan elections to remain consistent with ORS 254.065, which mandates plurality vote (requiring that the single highest or N-highest are elected), despite Article II Section 16 of the Oregon Constitution which specifically allows Proportional Representation and Preference Voting such as IRV. If one's going to modify statutes and the election process so drastically, we should fix the law to allow home rule charters to override plurality or simply repeal it altogether and replace it with IRV (and additionally Proportional Representation to really protect minority rights).
The Measure also doesn't address the issue of nominations. Why? Because it essentially bans party nominations in Oregon. Nomination rights are generally considered a form of political speech. Just allowing any member to run for office and say they are a member of the party, dilutes the party's right to protect its own identity. Nominations are more important than just endorsements.
5. A primary election process that advances the two candidates receiving the most votes to the general election ballot, and that allows every qualified voter to vote on which candidate to advance, helps to ensure the election of officials supported by a majority of the electorate, thereby promoting citizen confidence in their government.
DEMONSTRABLY FALSE
This measure does not ensure a majority of the electorate. There is simply no guarantee that all candidates in the primary who are the "top two" are approved by all the electorate. For example, I would not approve of Kate Brown or Rick Dancer if I were not nominated to the general election as a "top two" candidate. It thus does exactly the opposite: it LIMITS choices available for the general election that actually decides who is to be elected. The general election just becomes a "rubber stamp" of a majority, in the name of two minorities who voted for the top two candidates.
Mathematical qualities of a two-round plurality vote don't improve upon the spoiler problem at all and lead to an exacerbated non-monotonic behavior (or rank improvement leading to a failed election) that's the number of candidates minus two times higher than what could happen with IRV. A converse failure of the monotonicity criteria is actually the fundamental problem with plurality voting that leads to the spoiler issue -- e.g. ranking Neville over Novick leads to the nomination of Merkley.
Top two is, in mathematical terms, a degenerate form of Instant-Runoff Voting, except it doesn't have the main benefits that IRV provides in the real world -- instant counting, spoiler elimination, and majority approval.
First, it isn't "instant" -- costing taxpayers more money than needed. In Instant Runoff Voting (Preference Voting or Ranked Choice Voting), you're allowed to rank candidates in order of preference. The ballot then has all the information necessary to perform instant runoffs if needed to determine the majority winner.
Second, you'll hear claims that "top two" will solve the spoiler problem. That's demonstrably false. The "spoiler" is a term used only for candidates that are the "third or more" major-media-considered-electable candidates because plurality really only works when there's a choice between two options (identical to the IRV winner, the Condorcet winner, the approval winner, the Borda winner, the range winner).
Consider an election in inner SE Portland where there's no Republican or a weak one, a major-party Democrat, and a Green candidate (as in the case of District 42 and the Green Party's Chris Extine) -- the election is essentially between the Green and the Democrat -- a Republican entering the race would technically be the spoiler. Furthermore, to avoid such a situation, all the Democrats would have to do is run a second nearly identical candidate. All of a sudden, the Greens are called the spoiler between the two "legitimate" Democrats. Ignore the fact that at least one of the Dems is going to be promoted -- the problem is still there -- perhaps the other Dem was really "better" for the Greens and had their vote split.
Then consider a contested election between two major parties: the third candidate in the primary is still a spoiler -- if they pull enough votes to trip one of the candidates from being in the top two, the election will likely be between two extreme candidates, neither of which is guaranteed the support of the majority. The theory that "top two" supports moderate candidates is absurd.
Third, in the case of only one candidate (or option), the measure (similar in behavior to non-partisan races), calls off the election and doesn't even ask us for approval. In that case, there is no "majority vote" -- there's no opportunity even to "write-in" anything else to support the candidate's not receiving a majority of the votes. No vetting would be allowed. If the criterion of election was simply that a majority of the voters had to approve them to be elected, the election would be thrown out as not electing anybody and we'd simply hold a runoff with new candidates until somebody was elected. The measure simply doesn't account for that case in the single candidate case.
With IRV, a static election majority threshold automatically supports an implicit none-of-the-above election. The top-two solution is a ruse that moves the spoiler problem into a more cluttered and lower turnout election where the spoiler problems, in fact, are magnified. In theory, the general election will be devoid of spoilers, however, the spoiler problem doesn't, in fact, go away, because those who didn't like either nominated candidate will be "spoiling" their vote by either refusing to vote or voting a write-in (thus unable to express their real preference). It again forces people into casting a vote for the lesser of two evils, creating a false mandate. In IRV with a static-threshold or a dynamic threshold (threshold recomputed after each round to eliminate ballot spoilage) with "none of the below" ranking, they would be able to express both preferences at the same time. In Louisiana, candidates like David Duke end up performing well in elections because voters hated the "crook" more than the "Klan", though Duke only got into the general election with only a small fraction of support (1/3 or less -- in a primary too).
Combining the theory with the empirical reality, it's clear that Measure 65 is a poor and often non-solution for the problems it purports to solve and at the same time hurts diversity of viewpoints and fails to decrease election anomalies compared to either straight plurality or IRV, all in the name of bipartisanism.