Democracy on Mars 2: Current Shortcomings
Note: this is the second post in a series. The first post is here.
Today’s democratic institutions gesture towards ideals that the Martians could realize more fully. Current architectures are limited by logistical constraints from the time of their founding.
The United States serves as a good example.1 When the Constitution was written, soliciting opinions from a large group of people required either getting everyone together in a room or having each of them go to a polling place to mark a piece of paper that would be counted by hand and whose totals would then be carried to another location by foot and horse. Elections were relatively infrequent events, which was fine because the government had, by contemporary standards, relatively few decisions to make.2
Governments today make far more choices. They legislate for things like data privacy and environmental protection; provide services such as education, healthcare, and transportation infrastructure; and handle functions undreamed of in the eighteenth century, like stewardship of nuclear arms. Head counts and budgets offer a rough proxy for the scale of government activity: the U.S. federal government employs roughly a thousand times as many people as it did in the early years of the republic, outpacing population growth by more than an order of magnitude, and the federal budget has grown by a factor of more than ten thousand.3 The space of possible state actions will likely continue to increase as technology grants governments more degrees of freedom.
To keep the “popular sovereignty ratio” constant and maintain a baseline of binding oversight, this increasing volume of governmental decision-making would need to be balanced by more frequent and/or deeper opportunities for the public to guide those decisions, but the electoral metronome and format for providing input have remained mostly unchanged for centuries.4
Today, major elections in most democracies take place every four or five years.5 Given how much can change in half a decade, voters exercise remarkably infrequent control over governments that oversee large swaths of their lives. A citizen who, every cycle, dutifully fills out a ballot covering three electoral races or referendums, each of which contains two viable options, generates on average less than one bit of electoral information per year.6 That bit, furthermore, has become more diluted than original institutional designs intended as constituencies have grown: American congresspeople represent, on average, ten times the number of constituents they did in 1787; senators forty times the number; and the president around eighty.7
Part of the issue also lies with the ballot itself, which compresses detailed, nuanced preferences into a multiple-choice exercise. This enables legibility and scale, but at great cost. Voters are given a handful of choices for a restricted set of elected positions and referendums. The processes that led to these selections – agenda-setting, party machinations, etc. – lie outside most voters’ control. And once elected, politicians have little binding them to voter preferences beyond the threat of a performance review a few years down the line.8
Referendums likewise offer few guarantees about outcomes. The Brexit process demonstrates the illusory nature of the agency that ballot-based public consultation promises: leaving the E.U. was a choice with enormous consequences that hinged almost entirely on how, not just whether, the U.K. withdrew; the public was asked to initiate a procedure in which they actually had negligible involvement, and which ended up disappointing many early advocates of separation.9
The reductive nature of ballots not only leaves lots of decisions outside of the public’s control, it also means that many voters will not find any options they genuinely like. The qualities that enable a candidate to appear on a ballot (like the ability to win primaries or charm party higher-ups) may not correspond with what the public wants in a leader. Recent elections in which all major candidates were seen negatively by most voters demonstrate that, even in large democracies full of creative thinkers and thriving discourse, pre-ballot winnowing can produce an unappetizing menu.10
In short, the amount of binding oversight that democratic publics exercise over governments remains limited, despite a significant rise in the volume of government decision-making. The frequency of elections and structure of ballots grant voters limited expression relative to the scope and gravity of the decisions that governments make on their behalf.11
The expanded scope and complexity of government decision-making has also provided fodder for anti-democratic claims that it is unreasonable to expect the general public to grapple with the breadth of issues public institutions confront, and arguments that we should err towards less rather than more public consultation.12
It is true that the expansion of government remit means that completing a ballot asks a lot from participants: to vote responsibly you need to reach judgments across many policy domains, compare your preferences to candidate platforms, weigh each domain by personal salience and condense all this into a single vote for a candidate. There is a big gap between having a view on the state of affairs (feeling frustrated about inflation or worried about climate change for instance) and translating that into policies (e.g., a particular monetary strategy or carbon credit scheme). Making an informed decision on a candidate for office involves further work – looking at platforms, assessing a person’s honesty and competence, and then figuring out which trade-offs to accept: does an enlightened position on child tax credits compensate for bad foreign policy?13
Even on subjects one knows well it is challenging to develop conclusions at the requisite level of abstraction because policy levers operate through systems mediated by bureaucratic and legal machinery as well as markets, the environment, and other unpredictable elements of the real world. Motivated voters can read up, but policy debates are broad and deep, with large volumes of conflicting literature, some of which is not available to the general public. All this calculation and compression takes work. The reality is that shorter inputs can be more taxing than detailing one’s opinions in full. (As the saying goes, “If I had more time, I would have written a shorter letter.”)14
There is a mismatch between the world our institutions were designed for and the one in which they now operate.15 Today’s systems allow for only a trickle of binding public input, and as the state has expanded the popular sovereignty ratio has fallen. This imbalance makes voting more taxing and fuels anti-democratic arguments.
The U.S. is not fully representative of the democratic world, but the issues described above are pervasive: the vast majority of governments have expanded their remits over time, very few countries hold national elections more often than every few years, and almost all large-scale democratic decision-making (i.e. involving more than a few hundred people) involves multiple-choice ballots with a small number of options. There are some bright spots – Taiwan, thanks largely to the work of Audrey Tang, is probably the best example: it has launched a series of digital and in-person deliberation exercises and created new avenues for people to engage with public decision-making. Citizens assemblies have also shown promise in overcoming the constraints of the ballot, but so far almost all have been limited in scale.16 Adopting new voting mechanisms such as ranked-choice and quadratic voting can reduce dysfunction on the margin, but as long as these changes take place within the context of infrequent elections and compressive agenda-setting, their impact has a ceiling.
I’ve focused only on formal institutions, neglecting the role of softer components of democracy such as media vibrancy, local community health, government transparency, volunteer work, protest, and so on. These informal elements can go a long way towards improving accountability and patching over shortcomings at the electoral level. But formal institutions are the backbone of democratic societies – a country without regular, legally-binding public input would not be considered a democracy – and today’s were designed for another era.
Now, however, we have new tools, which the Martians could use to build something better.
Next post: New tools for popular sovereignty
I’m using the U.S. not because I think it’s the only place that matters, but because the American model has been widely mimicked, and because the framers of the Constitution took what we would today call a fairly “first principles” approach to institutional design, which means it nicely parallels the scenario faced by the hypothetical Martians. The British parliamentary model satisfies the first criterion but not the second, since it evolved more gradually. A number of more recently-founded countries, including many post-colonial states, satisfy the second, but not the first.
That said, technological constraints were not the only factors limiting popular sovereignty in 1787. Technology, for instance, didn’t prevent the expansion of voting rights to people other than white male landowners. But technology did limit design choices then in ways that it would not now.
The government also has responsibilities that don’t show up via internal headcount due to the rise in outsourcing to contractors.
I say “mostly” because of course some things have changed. Most importantly, extending voting rights beyond property-owning white males has made the United States much more democratic. Other adjustments, such as direct election of senators, the introduction of public primaries and the increased use of referendums have had an impact too, as have transparency and accountability measures, such as the notice and comment process for agency rule-making, more searching judicial review of administrative action, and increased congressional oversight of agency budgeting. Additionally, many local governments have created avenues such as public comment and referendums that enable citizens to get involved in decisions. However, some of these mechanisms have had a mixed record, coming under fire from the left for susceptibility to corporate influence and skewing in favor of the wealthy, and from others for needlessly impeding progress and exhausting voters by asking them to make detailed decisions with insufficient context.
For instance, the world’s six most populous democracies – which account for around half of the global population living under democracy – all fit this template. In the case of India, this refers to Lok Sabha elections. (Local office electoral frequency varies by state.) In the U.S., presidential elections take place every four years, Senate cycles every six years, and House elections every two. Indonesia holds elections for the People's Consultative Assembly (MPR), the People's Representative Council (DPR), and the Regional Representative Council (DPD) every four years, and most provincial, district, and municipal elections also follow five-year cycles, although there is some variance based on local regulations. Pakistan’s elections for the National Assembly and Provincial Assemblies occur every five years, but local and Senate elections take place on different timetables. Nigeria holds elections for president and the upper and lower houses of the National Assembly every four years. Brazil’s general elections (which determine the President, Vice President and National Congress members as well as State Governors and Legislatures) also occur every four years. It's worth noting that all of these countries have come under fire recently for erosion of democratic rule, and how many states worldwide one counts as democracies is a matter of opinion. For the purposes of this accounting, I have relied on Freedom House’s ranking. (I have counted “free” and “partly free” countries – as opposed to “not free” states – as democracies.) Regardless of where one draws the line of classification (at a meta level, this issue itself reflects some of the limitations of the use of language in law and policy, discussed in more detail below), the four-to-five-year frequency of elections is representative of almost all countries that hold regular elections.
In many countries, people have more frequent opportunities to vote, and the typical ballot includes many positions and referendums. However, voters often lack the context necessary to make informed choices on many of these (e.g., valorem tax exemption for timber production equipment, or the regulation of kidney-dialysis clinics) and participation accordingly tends to be lower. And even in the U.S., which presents voters with a relatively high number of choices, the historical increase in information conveyable by ballot is dwarfed by the rise in government responsibilities.
The 1790 U.S. census recorded a population of just under four million, and the 1st United States Congress featured 26 senators and 65 representatives at peak – they were not all present for the entire session. Today, the U.S. population of around 330 million is represented by 50 senators and 435 voting congresspeople.) A proposed amendment to increase the number of representatives proportionally with the population fell one state short of ratification in 1791. Had it passed, the House would now have around 6,000 members. Given that the framers explicitly disenfranchised most of the population, the intended ratio of government decisions to voters has grown even more sharply. The average American voter’s influence over government decisions has thus declined significantly over the nation’s history. I’ve used the U.S. as an example, but the phenomenon of government expansion with no matching increase in electoral representation and frequency is pervasive across the democratic world. The average Indian MP, for instance, represents four times as many people as during the country’s first elections in 1951-52.
This assumes they’re running for reelection, which they may not be. Even if they are, voter decisions about whether or not to re-elect tend to overweight recent behavior, to be easily swayed by rhetoric when interpreting track records, and to inaccurately assess factors outside of the candidate’s control such as global economic conditions.
There are plenty of other examples that demonstrate the pitfalls of reducing complex policy choices to binary ballots. Recently, this has included the 2016 Colombia Peace Referendum and Switzerland’s 2014 Referendum on Immigration.
For instance, the contests between Pedro Castillo and Keiko Fujimori (Peru, 2021), Boris Johnson and Jeremy Corbyn (U.K., 2019), Jair Bolsonaro and Fernando Haddad (Brazil, 2018), Emmerson Mnangagwa and Nelson Chamisa (Zimbabwe, 2018), Marine Le Pen and Emmanuel Macron (France, 2017) and Donald Trump and Hillary Clinton (U.S., 2016).
A useful parallel here is Ashby’s Law of Requisite Variety. Public feedback mechanisms today don’t offer nearly enough variety to guide government action. Thanks to Adam Russell for pointing this out.
These claims have a lineage dating back to Plato, so they’re hardly new. More recently, however, some thinkers have argued that the complexity of today’s decision-making and information environment has made democracy less suitable than it once was (see: Walter Lippmann, Jason Brennan and Garett Jones).
Of course, in reality few voters undertake such exhaustive assessments, but can we really ask them to, given the marginal impact of each vote? The return on such labor is poor.
This quote, or something along the same lines, has been attributed to many people over the years. The Quote Investigator website has written an interesting post on the phrase’s lineage.
This includes institutions founded recently but closely modeled on older ones, such as the slew of governments formed during the 20th century on the model of established presidential or parliamentary systems such as the U.S. and U.K. (or, in several cases, the U.S.S.R.).
By scale here I mean the number of attendees of a given session (assemblies face constraints similar to legislatures’). And I am referring specifically to exercises run by or with governments. As the next section will discuss, recent academic work that uses AI systems to run deliberative discussions at larger scales is very promising.