Iceland’s crowdsourced constitution: an open drafting process that stalled in parliament

Iceland’s crowdsourced constitution is the draft constitution written in 2011 by a 25-member Constitutional Council after a national forum, a court-annulled election and a period of online public input, and endorsed by voters in an advisory referendum on October 20, 2012.⁠^1,2,3 The draft was never adopted: a bill based on it was blocked in the Althingi, Iceland’s parliament, in the spring of 2013, and the constitution in force remains the 1944 text.⁠⁴ The case is the best-known test of open, internet-assisted constitution-making, and it shows both what public input changed and how the legal route to adoption defeated it.

What happened

Iceland’s constitutional experiment grew out of the 2008 banking collapse. The Venice Commission of the Council of Europe described the idea of a new constitution as “a unifying project designed to restore confidence” after the crisis.⁠⁵

The law and the national forum (2010). In June 2010 the Althingi passed Act No. 90/2010 to establish a consultative Constitutional Assembly.⁠⁴ A national forum of citizens was held as part of the process; Hélène Landemore describes it as “a demographically representative minipublic of 950 quasi-randomly sampled citizens” who met for one day.⁠⁶ The Venice Commission lists the forum among the “sometimes innovative” consultation mechanisms used from 2010 onward.⁠⁵

The election and its annulment (2010 to 2011). The Constitutional Assembly election was held on November 27, 2010. A University of Iceland constitutional documentation site reports that 522 people stood as candidates, that 83,531 people took part, a participation rate of 35.95%, and that 25 delegates (15 men and 10 women) were elected.⁠⁷ The Supreme Court of Iceland ruled the election invalid on January 25, 2011.⁠^7,8 The same documentation site lists six grounds: ballots could be traced to voters, the polling booths were unacceptable, ballots could not be folded, ballot boxes were not locked, counting was closed to the public, and no candidate representatives were present at the count.⁠⁷ On February 25, 2011, the Althingi decided to appoint the 25 elected delegates to a Constitutional Council with essentially the same task.⁠⁷ Björg Thorarensen dates the appointment to March 2011.⁠⁴

Drafting (April to July 2011). The council worked from April 6, 2011.⁠⁹ According to the council, all delegates unanimously approved the bill, a prologue and 114 articles in 9 chapters, at their last meeting on July 27, 2011, and presented it to the Speaker of the Althingi on July 29.⁠¹ During drafting it posted successive drafts online and invited comment; Alexander Hudson reports that the council received 323 formal proposals and a total of 3,600 comments, and that 311 substantive proposals came from Icelanders.⁠¹⁰ Landemore describes “about 3,600 comments for a total of 360 suggestions” on 12 successive drafts.⁠⁶ The two counts differ in what they include, and both are reported here as their authors gave them.

Delay and referendum (2011 to 2012). Thorarensen writes that the draft “was not processed further for more than a year” after submission.⁠⁴ An advisory referendum was held on October 20, 2012 with six questions.⁠² Hudson reports that the original plan was to hold it alongside the June 2012 presidential election, and that a filibuster by the Independence Party and Progressive Party prevented this.⁠¹⁰

Expert review and failure (2013). The Venice Commission adopted its opinion on the draft bill on March 8 to 9, 2013.⁠⁵ Thorarensen writes that the bill “was blocked in Althingi in the spring of 2013”.⁠⁴ Hudson reports that it was not brought to a vote and failed to be attached as a last-minute amendment before parliament dissolved, and that in the April 2013 election the center-left coalition that had launched the process fared poorly.⁠¹⁰

How it worked

Iceland’s process combined several participation mechanisms, each with a different selection method and a different kind of authority.

Agenda-setting by a random-sample forum. The national forum of about 950 people produced themes for the drafters.⁠⁶ It was advisory and held for one day.⁠⁶

Drafters chosen by election, then by appointment. Candidates for the assembly stood as individuals; the ballot listed each by an identification number.⁠⁷ After the court’s ruling, the drafters’ mandate came from a parliamentary decision rather than the voided election,⁠⁷ which opponents could and did contest.

Online input to the drafters. The council used its website and social media to post drafts and receive proposals.⁠¹⁰ Hudson calls it the first constitution-making process to take public engagement online in this way.⁠¹⁰ Proposals were advisory; the council decided what to adopt.

An advisory referendum. The October 2012 vote asked whether the council’s proposals should “form the basis of a new draft Constitution” and five specific questions.⁠² It was not binding.

The legal route that was not changed. Under Article 79 of the 1944 constitution, as the Venice Commission explains, an amendment must be adopted by the Althingi, after which the Althingi is dissolved and new elections held; the newly elected Althingi must ratify the same text.⁠⁵ The Commission notes that this procedure relies “on the division of operations in time” rather than special majorities.⁠⁵ Every participatory step therefore led back to two successive parliaments.

Measured results

Iceland’s measured results are the election and referendum figures and one study of how public input shaped the text.

Election turnout. 83,531 people took part in the 2010 Constitutional Assembly election, 35.95% participation.⁠⁷

Referendum results. The Ministry of the Interior’s official notice of October 30, 2012 lists 236,911 voters on the register and 114,570 valid votes.⁠² Statistics Iceland reports participation of 115,890 voters, 48.9%, with 49.9% of men and 47.9% of women voting, and states that a majority of valid ballots answered yes on every question.⁠³

Source: Ministry of the Interior notice.⁠²

On the first question, 73,509 yes votes are about 66.9% of yes and no votes combined and about 64% of all valid votes;⁠² Hudson reports 66.9%.⁠¹⁰ and Thorarensen 64%⁠⁴ The difference is the denominator. On the church question, voters favored keeping provisions on a national church,⁠² while the council’s draft had left the matter to law.⁠¹⁰

Effect of public input on the text. Hudson analyzed the council’s records and found that “almost 10% of the proposals from the public generated a change in the draft text,” with a greater effect on rights provisions; he attributes this to “the unique, apolitical context in which the constitution was drafted”.⁠¹⁰ The article was published in Policy and Internet (doi:10.1002/poi3.167); the version consulted is the author’s uploaded copy.⁠¹⁰

Failures and criticisms

Iceland’s process drew criticism from those who opposed the draft, from those who supported it, and from neutral reviewers.

Legitimacy of the drafters. The election that was meant to give the drafters a popular mandate was annulled.⁠^7,8 Turning the voided winners into an appointed council solved the practical problem but, critics argued, cost the process democratic standing. The low election turnout of 35.95%.⁠⁷ gave critics a second argument.

Separation from parliament. Thorarensen argues that “distancing it entirely from the political forum was bound to end in failure,” and calls the expectation that four months would be enough for parliamentary review “wholly unrealistic”⁠⁴ Landemore reaches a related conclusion in her lessons: work with existing institutions, because positioning the drafters as outsiders weakened their legitimacy and alienated key actors.⁠⁶

Quality of the draft. The Venice Commission welcomed “the active involvement of citizens” but found that “numerous provisions of the Bill have been formulated in too vague and broad terms,” that the institutional system was “rather complex and marked by lack of consistency,” and that the many referendum mechanisms raised “the risk of political blockage and instability”.⁠⁵ It also noted “a risk that, if adopted, the Bill does not reach the consensus needed for it to be confirmed by the next parliament”.⁠⁵ Supporters respond that a text written in four months by non-lawyers would naturally need legal refinement, and that the Commission also praised its aims.⁠⁵

Who participated online. Hudson reports that the 311 substantive proposals came from only 204 individuals, that the nine most active submitted 24% of them, that all nine were men, as were about three quarters of all proposers, and that one Icelandic scholar estimated 80% of participants were aged 40 to 65.⁠¹⁰ Openness to anyone produced input from a narrow group.

Referendum turnout and timing. Turnout of 48.9%.⁠³ was well below the roughly 88% average turnout in Icelandic presidential and parliamentary elections that Hudson reports for 1946 to 2013, and below the 2010 and 2011 debt referendums (62.7% and 75.3%)⁠¹⁰ Opponents argued that a minority of the electorate had endorsed the draft; supporters pointed to the filibuster that kept the vote off the presidential election day.⁠¹⁰ and to polling showing support around 66%⁠¹⁰

Political timing. Hudson argues that by 2013 the economy had improved and “the sense of the urgent need for political reform had dissipated”.⁠¹⁰ A process built on a crisis mandate outlasted the crisis.

What transferred and what did not

Iceland’s methods were widely studied and partly copied; its result was not. The Venice Commission notes that the participatory dimension “has been widely praised at the international level”.⁠⁵ Hudson notes that later processes in Chile and Ireland also emphasized public engagement;⁠¹⁰ see Ireland’s citizens’ assemblies, where randomly selected assemblies fed referendums that governments chose to call.

What did not transfer, because it was never achieved, was a route from participatory drafting to adoption. The Venice Commission suggested that, if consensus was not possible, Iceland might first amend the amendment procedure itself and leave further revision to a future parliament.⁠⁵ Landemore’s third lesson makes a similar point: when there are many approval hurdles, a narrower goal such as revising key articles may be wiser.⁠⁶

Lessons for My Vote Forward

These points are interpretation and design proposals.

• Participation must connect to the body with legal authority. Iceland’s drafting was open, but adoption ran through two successive parliaments.⁠⁵ My Vote Forward’s published priorities, like Iceland’s draft, have no legal force; the framework should say clearly where any output goes and who decides. See proposals and governance.
• Report denominators that critics will use. The 64% versus 66.9% dispute.⁠^2,4,10 is a denominator dispute. Every published release should give all relevant denominators. See measurement.
• Open input is not representative input. Hudson’s finding that a few highly active men produced much of the online input⁠¹⁰ is a warning for any open platform; see representation.
• Process integrity is the foundation. The annulled election shows that technical election flaws, such as traceable ballots, can undo a democratic mandate even when no one alleges fraud⁠⁷ See secret ballot and security.
• Show what input changed. Hudson’s method of tracing proposals into text.⁠¹⁰ is a model of how to evaluate whether participation mattered; see evaluation.

Unresolved questions

• What happened to later constitutional amendment efforts in Iceland after 2013, including bills on natural resources around 2020 to 2021? Their outcomes were not verified for this page.
• Would the draft have passed if its legal route had been changed first, as the Venice Commission suggested,⁠⁵? This is a counterfactual.
• How much of the online input came from people outside the most active group, and did their proposals change the text at the same rate? Hudson’s data address authorship and subject matter⁠¹⁰ but a full demographic profile was not found.
• Did the Supreme Court’s annulment change public trust in the process, and in which direction? No survey on this was verified.
• What were the national forum’s specific outputs, and how closely did the final draft follow them? Not established in the sources consulted.