Skip to content

Referendum Reform – Why, When, and How

Portland took to the polls last night and rejected everything that whiffed of controversy, while approving everything with a low profile. In the at-large race for city council, incumbent Pious Ali comfortably (and rightly) sailed to victory virtually unopposed. In District 3, the familiar candidate who carefully avoided taking any strong stances won against the outsider candidate carrying a clearly reformist platform. In the charter races, the “window dressing” proposals (1, 3, 4, 6-8) all passed easily, while the two substantive proposals (2 and 5) failed.

And then there were the citizen referenda. Subject to a toxic opposition campaign, there were five of these ballot initiatives, of which four were sponsored by the Maine branch of the Democratic Socialists of America. Question A, the only one to not be, was a sparring partner to Question B, both taking alternative routes to regulating short-term rentals. The rancorous confusion between these twins surely contributed to both of them failing. Question E, sponsored but then disavowed by the DSA after a row with organized labor, also failed. Question D, the subject of whirling debates about why people tip and whether waitstaff should continue to be legally siloed off from the rest of the workforce, also suffered from the controversy and failed.

The only DSA-sponsored proposal to pass was Question C, the unattended middle child. This proposal was by far the least discussed of the five, and I suspect this low-profile progression to the ballot box is why voters ultimately approved it.

Why was it so underdiscussed? The answer is obvious: it was by far the least clear of the five about what it even is. All five were complex, but the rest could be straightforwardly reduced to a central thrust. A and B: “Ban Airbnb!” D: “Raise the minimum wage!” E: “Get rid of cruise ships!”

C has no similar core. It has been described as “tenant protections,” but this is vague. The truth is that C is a list of related but distinct regulations for renters and landlords, all motivated by the same principles but that do not add up to a cohesive whole. This doesn’t mean that C is moderate, it isn’t, but it does mean that it’s hard to summon up the sort of indignant, partisan fire over it. This is very similar to the “green new deal” of 2020, where a broad collection of reforms (unrelated to the actual Green New Deal in D.C.) were bundled together into an illegible whole.

This vagueness, however, is a great defense, and it’s almost certainly why Question C passed and the much more straightforward Questions B and D failed. You’ll see a lot of claims that Portland rejected radicalism last night, but this isn’t true. Portland rejected clarity in favor of anodyne obscurity.

Two Knights in Combat, 1518. Albrecht Dürer.

Therein lies the issue. Many feel frustrated by referenda, not because they believe direct democracy has no place, but because the way the ballot initiative process works in Portland is prone to confusion, abuse of public trust, and “not getting what you paid for.” The initiatives that have been passing in Portland aren’t the sort of clear-cut questions of conscience that we expect from referenda. Instead, they are thickly complex webs of legislation full of precise figures and systems that should be subject to educated discourse before implementation, and able to be tweaked afterwards. The referenda which have dominated local political discourse over the past several years are 1) too complicated, 2) require too little popular support, and 3) are too rigid for too long once enacted. This process needs to be reformed, and now is the ideal time to do it.

If such a reform were being considered right after a sweeping victory of referenda, then the supporters of those referenda would justifiably accuse city government of trying to unmake the people’s will. If such a reform were being considered in the leadup to a big election, then that would be an obvious conflict of interests.

But right now, immediately following an election where referenda almost entirely failed, now is the time to act. The city government will never have a more favorable political environment than this winter and spring to make much-needed changes to the process. But, hold on, can the city council change these things by themselves?

The process for citizens’ referenda are not in the city charter, many are surprised to discover. The provisions for these ballot initiatives are found in Chapter 9 of the city’s code, generally able to be amended at any time by the city council, and only the city council. A provision (Sec. 9-36(b)) of this chapter actually prevents the referendum process from being amended by referendum. Ironic!

How would Chapter 9 be amended? Not a question with an easy answer. The first step is certain - the City Council must, by majority vote, enact the agreed-upon changes.

What comes next is uncertain. According to the current understanding of Portland's legal counsel, this amendment would need to be approved by the electorate at a municipal election. A yes/no referendum, essentially. This popular ratification isn't mandated by the Portland Charter, but rather is based on a reading of Maine's Constitution, Article 4, Part Third, §21. 

Whether or not this provision applies to merely amending, in addition to implementing, ballot initiative systems, however, isn't clear. A plain reading of the relevant law doesn't explicitly require a popular ratification. If this is so, then the City Council could amend Chapter 9 unilaterally, without reference to a ballot referendum.

But even if the Council can do this, it would still remain a prudent decision for the Council to seek popular ratification. This would ensure that the legitimacy of the amendments is universally recognized.

An earlier version of this article made a bolder claim in favor of this latter interpretation, which The Portland Townsman now retracts, while acknowledging that it may still be accurate under the law. [Updated 11/11/2022]

So, what should Portland do? It would be a horrible shame to abolish the referendum process entirely, as many cities have, but in order to protect the process, there are three problems to solve. What follows are three proposed methods for fixing it. They are not too specific, as the details should be up for debate by the people and our elected representatives, but they represent templates for how the Council can fortify our local democracy.

I. Set a sustainable number of signatures required by ballot petitions

This first one is contentious, but it’s a problem that must be solved. As it stands, Portland, with its population of about 70,000 citizens, requires would-be sponsors of referenda to gather exactly 1,500 signatures in support of their proposals. Once they do, the proposal must either be adopted (with possible amendments) by the city council, or be put on the next ballot. That’s about 2% of the city’s population. Opponents feel this number is too low, and that it allows for motivated but tiny minorities to massively influence elections. Whether or not you think it’s too low, you may also notice that it’s an exact number, 1,500, and not a figure that scales with population growth or decline. This means that, assuming Portland continues to grow, the percentage of the population required to put questions on the ballot will shrink every year.

The number of required signatures, therefore, shouldn’t be a fixed figure. It should be proportional to Portland’s electorate, growing with the city. This is typical for cities with referendum arrangements similar to ours.

Why don’t we look at cities, both in New England and around the country, and see what they do? Specifically, whether or not they allow ordinance-by-referendum, as we do, how many signatures they require a petition to attain to propose a ballot initiative, and how much that figure is proportional to their population. First, here is Portland:

So as we can see here, Portland has a population of about 68,300 citizens, it requires 1,500 signatures to put a proposal on the ballot, and that figure (1,500) is equal to about 2.2% of the total city population. Let’s look at two other cities in Maine:

Immediately we see both Lewiston and Bangor use proportional, rather than fixed, figures as a minimum for their petitions. Lewiston leans on the permissive side, requiring a number of signatures equal to just 7% of the total number of people who voted for Governor in the previous election. Bangor, meanwhile, skews strongly conservative, requiring 20% of the same figure, which ends up being more than Portland requires. As you might imagine, ballot questions are a rarity in Bangor.

For more context, let’s look at the rest of New England:

Oh. Despite the reputation as a New England tradition, dating back to the days of Puritan meetinghouses, turns out that ordinance-by-referendum is in fact quite uncommon in New England’s cities.

This actually surprised me; I had expected more cities in New England to possess some form of local referendum system. This dearth shows what we have to lose – if we want to maintain any form of referendum system, we need to make sure that it works for us, lest antipathetic elements seek to abolish it entirely.

But loathe as I am to believe it, there is more to the world than just New England. Let’s look at some other assorted cities around the country:

As you can see, cities that possess citizen-initiated ballot referenda tend to settle on a figure in the range of 5-15% of active voters. How to define an “active voter” varies, but is usually either a registered voter, or a person who voted in the last state election. Of this arbitrarily assembled list, only one other city (Cleveland) uses a fixed, not proportional, figure for this purpose.

If we were to steer a medium course, and require a number of signatures equal to, say, 10% of Portland’s registered voters, that would be about 5,000 signatures needed. But if wanted to keep the ballot accessible, we could opt for 5% (~2,500), or if we wanted to put a higher barrier against unrepresentative agitation, we could pick 15% (~7,500). Whichever figure the council settles on, this is the first step to fixing the current system.

II. Limit the breadth of content

The “green new deal” that Portland voted for in 2020 had nothing to do, officially or conceptually, with the Green New Deal being proposed in Washington DC by progressive politicians including Senator Sanders and Representative Ocasio-Cortez. Portland’s “GND” was rather a motley mixture of labor laws, density taxes, and building restrictions. When Portlanders turned out to the polls, they likely knew that the Federal-level Green New Deal invested in infrastructure, renewable energy, and jobs in public works. They probably didn’t realize that the local “GND” in front of them would primarily impact their city by punitively taxing anyone who wants to build dense housing, while rewarding suburban homeowners. How much this anti-density tax has impacted Portland’s housing shortage is yet to be definitively determined, but I think it is uncontroversial to say that the name “green new deal” did not adequately convey the referendum’s content.

The issue here is not the branding, as disingenuous as it was, but rather the rolling of pages and pages of legislative reforms across a swathe of ordinances into a single question that (often uninformed) voters are forced to either approve or deny wholesale. The average voter does not have a thorough understanding of the ins-and-outs of Portland’s Code, or U.S. constitutional law, or Maine’s body of statutes, or political economy, but they do have a set of values that they strive to live by. This makes questions like, “Should Maine recognize same-sex marriage?” or “Should Portland raise the minimum wage to $18?” good questions, because they are straightforward questions that ask the voter to act according to their principles and sense.

Should Zone R-5a in the Land Use Code be expanded to allow the set of commercial uses defined in B-1?” is a bad question to begin with, incomprehensible legalese to >90% of voters. If fifteen similarly opaque questions are rolled into a single ballot proposal, slapped with a catchy name, and presented to the voters, we’re looking at an almost inherently illegitimate outcome.

Detractors claim that restricting the breadth of ballot proposals is anti-democratic, but to the contrary, it’s usually more democratic. Consider 2022’s Question D. While a single question, it contained at least three major proposals: first, raising the minimum wage to $18, second, eliminating the legal exception for a sub-minimum wage paid to tipped employees, and third, establishing a city government organ to redress labor abuses. If these were three separate questions, voters could more accurately make their voices heard, but since they were all rolled into one, it’s hard to know who wanted what and why.

Portland’s city council needs to restrain this practice, which is already starting to become commonplace. Looking at other states, there are two primary ways by which to do this, each with their flaws:

Option A: Single-Subject Restriction

This is the most straightforward way to approach the problem, used already in many cities and states. With this method, there is a simple clause that compels ballot referenda to only concern a single subject of law. If a designated government commission determines that a referendum spans multiple subjects, they will divide it into as many separate questions as they see fit.

The problem that faces any attempt to limit the complexity of ballot referenda is this: “How exactly do you measure the ‘complexity’ of a proposal?” Single-Subject Restriction laws concede that, basically, you can’t be too technical about it. You just have to appoint a group of people with the power to break up any proposal they deem too unwieldy. While the threat of arcane litigation lurks around the corner, this is likely the most effective way to tackle the problem.

Option B: Text Length Restrictions

But there is this other way. If the Single-Subject Restriction is a highly subjective, personal method of handling things, this is a very black-and-white, objective method. Any proposed ordinance to go on the ballot needs to fit on a certain number of equally formatted pages. A character limit, essentially. Like Twitter. This has the benefit of not relying on imperfect humans to make potentially controversial decisions, just make your proposal fit on so many pieces of paper, 12pt font, standard margins, single-spaced. It’s like school again.

While the objectivity is appreciated, this method has many problems. First, short laws are rarely better laws. This would incentivize absolutist statements which fail to properly account for the nuances of the real world. And it wouldn’t necessarily fix the real problem, which is the blending together of disparate issues. A lot could be changed across a wide variety of subjects while still sticking to the page limit.


Whichever option is picked, it’s also crucial that the full information be given to voters on election day. In Portland this year, after the city council approved the referenda questions for the ballot (they had no choice) the question arose: Do we put the whole text of each proposal on the ballot? City staff were adamant: No. It would make the ballots far, far too huge. It would be expensive, messy, and ridiculous. City Council agreed, allowing for summaries to be put on the ballot instead, but these summaries were flawed. Written by the sponsors, they often failed to adequately explain to voters what the proposals were. This was especially the case with Questions A and B, two proposals regarding the same subject. Their summaries completely failed to indicate what the differences between them were, to the point that at that Council meeting, supporters and opponents of both questions asked for the Council to amend the summaries to make things more clear to voters.

The City Council declined, however, and the summaries stayed as-is. All citizens should inform themselves before voting, but we all know that many won’t. For those voters, they should have everything they need to see with these citizen referenda (that at this point only 1,500 Portlanders have expressed approval for) right there on the ballot. And if it’s too long and ridiculous? Then woe betide the sponsors of these questions, who may face an onslaught of annoyed “NO” ballots. Making ballot initiatives more straightforward and accessible to understand is the 2nd step to fixing this.

III. Shorten the period of inflexibility

As of right now, when an ordinance is enacted by referendum, it cannot be touched by our elected officials for five years. Five years is a long time. During that time, it may become imperative that, due to changing circumstances, tweaks or adjustments to the (often very intricate) legislation be made. But for five years, our Councilors’ hands are tied. This is frustrating, disempowering, and dangerous.

The reasoning behind this five-year bubble is obvious. If there was no time limit, then the people could enact an ordinance, and then the City Council could immediately repeal it. Talk about disempowering! Supporters of the current system point out that referenda can be amended by another referendum, but this would also “reset the clock” to five years, and we’re back where we started. Some have suggested that the solution would be to keep the five-year period, but allow ‘moderate amendments’ or ‘adjustments’ to be made, as long as the spirit of the law isn’t injured. This is sensible, but it’s also very prone to both over- and under-restricting change.

The best option, most likely, would be to simply shorten this period of inflexibility. Two years, perhaps? Those who trust our elected officials might be tempted to say one year, those who more strongly believe in referenda may say three, but I think it’s becoming eminently clear to all sides that five is just a ridiculously long time to be stuck with the very particular outcome of a single election. Most term lengths are shorter than that. Cutting this period down to size is the third and final key step to fixing this.


To summarize, here is an example of a simple reform to Chapter 9 which would solve the above problems (proposed new language is bolded, removed text is struck through):

The submission to the vote of the people of any proposed ordinance dealing with legislative matters on municipal affairs or of any such ordinance enacted by the city council and which has not yet gone into effect, may be accomplished by the presentation of a petition therefor to the city council in the manner hereinafter provided and signed by at least one thousand five hundred (1,500) voters a number of voters equal to at least 10% of the number of registered voters in Portland. The submission of a proposed ordinance, or amendment or repeal, in whole or in part, of an ordinance already in effect shall be hereinafter referred to as the direct initiation of legislation or “initiative.” Such initiatives must contain at most one proposal of law. If, in the opinion of the city clerk, the initiative contains multiple proposals, these will be divided by the city clerk into separate questions prior the public hearing described in 9-36(f). The submission of a petition to override any ordinance passed by the city council but which has not yet gone into effect shall be hereinafter referred to as the “people’s veto.”

Sec. 9-36(a)

An ordinance enacted by a vote of the people at an initiative or referendum election shall not be repealed or amended for a period of five (5) two (2) years from the effective date of the ordinance, except by a vote of the people, unless such ordinance shall otherwise expressly provide. After five (5) two (2) years from the effective date of the ordinance, the city council after public hearing may repeal or amend such ordinance by vote of five (5) of its members.

Sec. 9-46

This is just an example. The details can be discussed, debated, adjusted, and compromised, both among the public and our elected officials. But the current arrangement cannot hold. Its weaknesses have been exposed, and public trust in the referendum process has been foundationally shaken. To protect the referendum process, to maintain an element of town meeting democracy in the governance of our city, it must be reformed. It would be better to do so now, in the shade of the last election, with good intentions, rather than let it again threaten to tear apart our democratic community.

This article was updated on 11/10/22 to allow for ambiguity concerning the method of amending Chapter 9 while more research is done.


Ashley D. KeenanAshley is an editor of The Portland Townsman, writer on urbanism, local small business-owner, and Maine native. Her work primarily covers the national housing crisis, building sustainable and livable cities, responsible market economics, and New England culture and history. She lives in Portland with her fiancé and can be personally reached at


  1. Dan Hogan Dan Hogan

    Ashley, your analysis of, and recommendations for the citizen referendum process are spot on. You will, no doubt, hear a lot of blowback from the democratic socialist interest but, as you have concluded, better to start the process now when the existing process has failed than to wait until a faulty referendum has passed and incur subsequent accusations of sour grapes for the process.
    I’m a South Portland resident for the past 20 years but a Portland resident for the previous 35 and a downtown Portland worker for 32 of that. As such, I’ve observed the sometimes painful process of Portland’s phoenix-like rise from vacant storefronts and limited job opportunities to the sometimes qualified definition of today’s successes. I hope you keep writing about how to manage this inevitable growth in terms of transportation, housing, and open space protection. In some ways, this may be Portland’s last chance for the type of growth that prevents us from having to say in the future “I wish we had (hadn’t) done that.”

  2. Patty Skerritt Patty Skerritt

    This is an outstanding summary of the very thing this Portland citizen and taxpayer was thinking about Wednesday. How do we reduce the number of referendum questions so the process does not repeat from the election completed November 8th.

    I love everything about this and will share widely.

    Thank you for such thoughtful research.

Leave a Reply

Your email address will not be published. Required fields are marked *