The problem with voting

I read, with great interest, Fred Ehrsam’s post on blockchain governance, as well as Vlad Zamfir and Vitalik Buterin’s responses.

Their debate centers around the value of “on-chain” governance (in other words: decisions enforced by protocol) and the sovereignty of people versus code.

When I read Fred’s post, however, I was struck by a different observation: that despite a supposed “Cambrian explosion” of new governance models, every idea that’s referenced utilizes the concept of voting.

If we’re coming up with wild ideas for social coordination, voting seems like an awfully familiar solution. Let’s start with a blank sheet of paper…

What does a governance system look like that doesn’t require voting at all?

I started to sketch out some ideas, which I’ll outline in the remainder of this post.

Why do we even vote?

Source: Flickr

Voting, in its modern form, is only as old as the Enlightenment, which bubbled up in the late 1700s and gave birth to the liberal democracy, with the French and American Revolutions as its poster children.

Prior to the Enlightenment, the dominant form of Western government was monarchy, where governing power was inherited or assumed by a small group of elite, who believed that most citizens couldn’t be trusted to make such decisions.

Under monarchy, citizens were largely uninterested in government affairs. The king was a distant authority who collected taxes every once in awhile. Why should they care when they didn’t have any input?

In the new era of “government for the people, by the people”, voting was introduced to give people an official stake in government, thus incentivizing them to participate.

Compared to what we had before, voting is obviously an improvement. But the problems that liberal democracy tried to solve for in the 1700s (apathy, lack of decisionmaking power) are different from the emerging social problems we need to address today.

What’s wrong with voting?

Although voting seems like an intuitive concept, there are a few major flaws that seem to be getting worse over time.

Voting is never truly representative

We assume voting is fair because it vaguely reflects some total population that we are trying to represent. It’s impossible to exactly pin down what “representative” means. (Similar demographics, interests, incomes, ideologies? All of the above?) But to use an extreme example, if the only people voting in the U.S. presidential election were residents of Biloxi, Mississippi, that would not be considered representative.

Even today, the analog gap between voters and “true” population is problematic. It’s ruthlessly exploited through tactics like gerrymandering, voter registration to encourage turnout, and voter ID laws to suppress it.

But the gap is even worse in a decentralized community, where total population is unknown and may never reveal themselves. An open source project doesn’t know who downloads their code, and they have no way of contacting them. ISIS, with its vision of an Islamic state, is more of a movement than a centralized organization.

Given this state of affairs, how do we know whom to represent, whom to protect, and whose interests are being served? Under these conditions, the concept of “representative voting” is likely impossible.

Voting is a competitive game

Voting is a zero-sum game, meaning that whomever wins does so at the expense of someone else. As a result, voting promotes competition, not cooperation. Players might coordinate as a means of gaining an edge (“if you vote for X this time, I’ll give you Y next time”), but ultimately, “winning” the vote means beating someone else.

Under competitive circumstances, we form groups — whether informal alliances or political parties — and we fight each other viciously for a win.

In the United States, the framers tried to address this problem with checks and balances. But they also made a fatal mistake. They didn’t anticipate the formation of political parties, other than to actively discourage them. So although the United States has checks and balances against the three branches of government — executive, legislative, and judicial — there are no checks on the power of a single, well-coordinated faction spanning across branches.

To reduce competition, some governments have experimented with runoff voting and similar approaches, but they have yet to take off, probably because they are too complicated (more on that later).

Voting still works in limited contexts

“Voting is a last resort. A healthy community should rarely need to vote.” — Ben Collins-Sussman and Brian W. Fitzpatrick

Voting isn’t entirely useless. It can be helpful in limited matters when the total population is small, known, and relatively homogenous — which, funnily enough, is how things looked when liberal democracy made its debut.

In 1800, there were 5.3 million Americans. When you take out slaves, women, and children, you’re left with ~1 million eligible voters, all of whom were free, male, and white. Those are the conditions under which the United States launched its grand experiment.

Hence the old complaint of, e.g. “Belgium has a great healthcare system, why doesn’t the United States?” It’s a lot easier to coordinate Belgium than the United States, or moreover a global, decentralized internet population.

Voting can also serve as a tiebreaker when consensus cannot otherwise be reached. But as a permanent fixture of government, people are incentivized to exploit it.

The veneer of objectivity

Voting is alluring for the same reason that on-chain governance is alluring. If a complete lack of governance leads to the “tyranny of structurelessness”, over-engineered systems create a “veneer of objectivity” that lets us put blind faith into something greater than ourselves — and in doing so, absolve ourselves of the very real, hard work of disagreeing and compromising with our fellow humans.

If humans are good at one thing, it is circumvention. It seems paradoxical to call decentralization “revolutionary” while also preaching government by automation. If so, what a disappointing revolution! If we ever get to the point where we blindly accept what a bunch of code tells us to do, machines will have won and humanity, in all its creative chaos, will sink into decline.

It’s undeniable, however, that our current system poses a number of problems, so if nothing else, we should ask ourselves what’s working and what isn’t, much like our 18th-century Enlightenment counterparts.

So. We have our current system, and we’ve identified some emerging problems that we need to solve for. What does that look like?

Designing for cooperation, not competition

If you’re an avid board gamer, you’ve probably come across a cooperative game or two, like Pandemic or Forbidden Island. In a cooperative game, you work with, rather than compete against, your fellow players to achieve a shared outcome. (In game theory, this is akin to a stag hunt.)

Source: Flickr

A few real-life examples of cooperative games:

  • Global warming: Figure out how to cool down the planet, or everybody dies. (It’s telling that the only “challenge” in this game comes from people who deny — i.e. defect from — the game entirely.)
  • Actual nuclear warfare is an outcome that everybody would like to avoid (although nuclear proliferation itself is a competitive game)
  • Government shutdown: In 2013, the U.S. government shut down for two weeks because nobody could agree on the budget. To restore operations, legislators had to work together.

Here are a few interesting features I’ve noticed about cooperative games:

Default mode is failure

Players are incentivized to work together because they’re trying to avoid a failed outcome. For example, in Pandemic, you’re trying to stop a disease outbreak, and in Forbidden Island, you’re trying to escape a sinking island.

Unlike voting or other competitive games, everyone agrees that the default outcome is bad, so they’re motivated to figure out how to escape their terrible fate. If one person loses, everybody loses.

Bias is towards action

A cooperative game’s incentive is never “Figure out how to do X”. That outcome is too hard, and when things are hard, people don’t want to do them, and our lazy reptilian brains invent ways to get around it (because again, humans are great at circumvention!).

By specifying the outcome beforehand— “Bad thing X will happen unless you change it”, and making it undesirable, we have no choice but to take action or lose the game.

Timeline is known

Players must feel a sense of urgency when working together to solve the problem. As a counterexample: why don’t more people care about climate change? Because nobody really knows when disaster is going to strike. Without knowing if it’s a 10-year or 1,000-year problem, people are reluctant to take action.

Cooperative governance in modern times

I’d love to see more cooperative mechanics introduced into our governance models, because I think they’re uniquely suited to our modern social challenges.

The first wave of open source communities (ex. Debian, Linux, and Apache) pioneered the use of voting systems. But among modern open source communities like Node.js and Rust, we’re seeing a shift towards cooperative governance.

Some projects use lazy consensus, where consensus is assumed to be the default state (i.e. “default to yes”). The underlying theory is that most proposals are not interesting enough to discuss. But if anyone does object, a consensus seeking process begins.

The goal of consensus seeking is to discuss concerns until no blockers remain, without coming to a vote. A vote is considered the “failure” outcome. Participants voice their concerns and listen to each other, but try to avoid blocking (or stalemating) the proposed course of action. When no blockers remain, stakeholders are said to have reached consensus.

Under consensus seeking, a proposal theoretically might have won 10–2, but if those 10 felt weakly about their support, and those 2 feel strongly opposed, the minority could still win.

Here are a few reasons why cooperative games work with, not against, today’s emerging social norms:

A veto is expensive

Under voting, it’s cheap to veto (i.e. vote against something), which has led to an anti-pattern of outrage today. Not only is it cheap to disagree, but you’re often rewarded for it (in the form of attention and/or reputation).

Cooperative mechanics flip this around, so it’s cheap to do nothing and expensive to raise an objection. That means people reserve their objections for stuff they really care about. If you object too often, other people are likely to start discounting your opinion.

You can focus on what you care about

When voting, the expectation is that everyone ought to participate. Even abstaining is a conscious decision. But in a cooperative game, you can participate as much or as little as you’d like.

It’s no secret that we’re drowning in information. As Albert Wenger suggests, today’s challenge is not finding time to actually read everything, but rather deciding where to spend your limited attention.

It’s impossible to maintain deep knowledge of every political topic of importance. Cooperative governance gives us permission to be curators, focusing on just our areas of passion and expertise.

Reputation becomes powerful

Reputation is quickly becoming more important than money. (Or, rather: perhaps money was just a historic proxy for reputation.) Cooperative games leverage the power of reputation really well, because you need reputation to capture attention. And whomever captures enough attention, gains influence.

While this may sound bleak at first, when compared to wealth, reputation strikes me as a more democratic alternative. Black Lives Matter started as a hashtag. The Women’s March started on Facebook. For all that was terrible about the U.S. presidential election, Trump broke the mold of how people become president, simply with a powerful enough megaphone.

Votes can be manipulated by propaganda and special interests. (Consider the ethics of a “voter registration drive” sponsored by a partisan group.) Reputation takes time to build, but it belongs to you.

Building cooperation into the system

I won’t go into this too much, but it’d be fun to consider how to actually build these mechanics into a system. Cooperative governance offers an elegant compromise between on- and off-chain governance: you have a failure outcome that’s enforced and “real”, but the process of consensus remains firmly in the hands of people.

Here are a few half-baked ideas for how we might implement cooperation mechanics:

  • Enforced deadlines for RFCs: For any new proposal, you could specify that it will be merged after X period, in whatever state it’s in.
  • Auto-renewals for existing policies: This, plus lazy consensus, could be particularly interesting for social welfare programs. Consider the benefit of opt-out vs. opt-in organ donor programs, or automatic enrollment for 401(k) retirement plans.
  • Expiration dates for existing policies: This policy will expire at X date, so people are incentivized to figure out what comes after.
  • Fail-deadlies: Similar to a dead man’s switch: if X changes, it will cause incredibly horrible thing Y to happen. (Remember that button they had to keep pushing on Lost? Kinda like that.)

Problems with cooperative games

Finally, no governance system is perfect, because humans are — again! — sneaky little buggers, and we like to find ways to break things. What problems might we anticipate?

  • Lack of representativeness: Some aren’t able to advocate for themselves directly (ex. children or people with severe mental disabilities)
  • Hijack the signaling: Media platforms, for example, could influence popular opinion on a topic by promoting or suppressing different types of content.
  • Oligarchy, cults of the individual: If you’re a prominent person with a strong reputation on a given topic, you’ll want to protect your influence. Also, others who want to participate but don’t yet have a reputation become overly reliant on your voice.
  • Defect from the game: People who don’t like the way things are going, and don’t feel they can influence it, will go somewhere else. So cooperative mechanics could cause more forking and fragmentation, not less. (I’m not sure this is necessarily a bad thing.)

We’re in the era of the stag hunt

Finally, please note: these are just ideas. We can’t, and shouldn’t, drag-and-drop new systems onto people without their buy-in. And it won’t be easy to transition from competitive to cooperative governance, because people will want to preserve their existing interests.

Still, if you’re starting from scratch, I hope this post provides some inspiration. Voting wasn’t designed for scale, and it’s not going to carry us through the next era of governance.

We need to encourage people to work together: not by making things harder, but with equally simple alternatives. We need to design a system that properly acknowledges reputation as the powerful asset it’s become. And we need a system that doesn’t overwhelm our attention, but rather respects our desire to focus.

And we need to do it all before society implodes. (In fact: the shared sense of doom and gloom that we’re feeling these days points to a great opportunity for a cooperative game!)