Based on a photograph by Mikko Saari

An App for Revolution

Coordination problems protect the status quo. Decentralised technology can help us overcome them.

Perhaps you’ve considered switching from Facebook to an alternative social media site. But you’d only want to migrate if your friends came with you.

Maybe you’ve thought about protesting by breaking some absurd law. But if you did it alone you’d make yourself an easy target for punishment, which you’d rather avoid.

Consider citizens voting for a party other than the one they’d most like to see win, because they believe their favourite has no chance of victory. But when many people do this it can be a self-fulfilling prophecy. Perhaps their favourite could be successful if its supporters all acted together.

In a collective action problem, everyone would be better off if the whole group took a given action. But anyone who carries out the action alone will be worse off. Without a means to coordinate, the whole group is stuck in the sub-optimal situation.

We already have web coordination platforms that help people, but their domains of applicability are constrained by centralised design. If they’re used for genuinely subversive things the politically powerful could limit access or shut them down.

We’re about to enter an age of decentralised platforms that allow strangers to overcome the coordination problems shielding these powers from disruption.

Legacy coordination platforms

The defunct Pledgebank allowed organisers to specify an action, and invite others to pledge to carry it out, if the total number of pledges exceeded a predefined target.

You’re more likely to have heard of Kickstarter, which allow an organiser to collect pledges to fund their project. The funds are only transferred if the pledged amount is higher than a predefined threshold by a given cut-off date.

Twitter, though not designed for solving this kind of problem, was also pivotal for coordination during the ill-fated Arab Spring. Activists and protesters took courage and were spurred into action by a preference cascade that played out on the platform.

Online coordination platforms are still relatively new technology. Pledgebank was launched in 2005. There are many variations on the pattern still to be tried at scale. For instance, one particularly ingenious pattern, invented back in 1998 but still not widely known, is Alex Tabarrock’s Dominant Assurance Contract.

Decentralised coordination platforms (DCPs)

Even on the centralised web that we’re familiar with today, it seems likely that there’s significant room to experiment with and improve on our collection of coordination platform recipes. But with the advent of cryptocurrency networks supporting smart contracts (e.g. Ethereum, and in the future Bitcoin Cash) it’s possible to create platforms that differentiate themselves in new ways I believe will turn out to be very important.

An enlightening conversation with Justin Goro helped me figure out one way a DCP could be arranged on the Ethereum network. Here a disclaimer is in order: Any faults with the design I’m about to describe belong to me. This is a domain I’m new to, so faults can be expected. Nevertheless, I believe this design is close enough to something that could work, to be helpful for the case I’m trying to make here.

A simple DCP design

Once the system had been developed and deployed, there are three non-exclusive human roles relevant to its operation. The participants interact with smart contracts to carry out the actions described below.

• Instigator. A person creates an Initiative by specifying some time-bounded goal in writing. The goal is assumed to be something that Backers could help bring about. The Instigator also nominates one or more people to be Oracles, an uneven number.
• Backers. Backers are people who anonymously join the initiative by making an arbitrarily large or small deposit of Ether into a smart contract.
• Oracles. After the time limit of the Initiative is reached, the Oracles observe the state of the world and each render a decision about whether the Initiative succeeded or failed. The ruling is reached by a simple majority vote among the Oracles.

After the decision has been rendered, a small fee is transferred from the pool as payment to the Oracles. If the Oracles deem the Initiative successful, the remainder of the pool of funds is paid out to the Backers, divided equally between all of them. If it failed, the pool of funds is ‘burned’ (no person is able to spend the funds).

The incentives

• Everyone who has made a deposit (whether success would make them a net giver or taker of funds) is financially incentivised to take action to help realise the goal.
• People or organisations who most strongly want to see the goal realised (e.g. lobbyists) are more likely to deposit relatively high amounts, or to become initial Backers. Backers who turn out to have deposited more than average can be sure they’ll lose money on the Initiative, but their contributions provide a financial incentive for lower level Backers to join it. Which in turn makes the Initiative more likely to succeed.
• Less passionate supporters of the cause are incentivised to join by making a lower than average deposit. In the case of the goal being realised, they will have gained financially.
• Potential Oracles assess invitations based on the fee being offered, on how politically risky the initiative is (do powerful entities want it to fail?), and on how diligent they have been in separating their real world identity from their online pseudonym. They would want to avoid being targeted for intimidation or worse.
• Sympathetic outsiders can witness the (limited) social proof attached to an in-progress Initiative in the form of number of deposits, and total deposited amount. Reassured they’re not alone, this makes them more likely to act towards realising the goal — which would make the Initiative a success — whether or not these outsiders chose to make a deposit.

The differences

Kickstarter and Pledgebank hosted projects depend to some extent on social proof — prospective backers are emboldened to join by seeing large numbers of existing backers. Initially a DCP might seem to have a handicap in this area. Since DCP Backers are anonymous, it’s not possible to discern how many real world identities map to the deposits that have been secured — one Backer could make multiple deposits. The best that a prospective Backer can do is take a guess about how many real people are likely to have collectively made those deposits.

Unlike Kickstarter or Pledgebank, DCP Backers need to make a financial contribution to join an Initiative. Even though the deposit can be arbitrarily small, the barrier to participation is higher. In the case of a failed initiative, DCP backers lose any deposits they have made.

By pledging more than the average deposit, DCP Super Backers effectively become recruiters to the Initiative by strengthening the financial incentive for lower level Backers to join. This dynamic has no analogue in traditional coordination platforms.

While Kickstarter and Pledgebank judge campaigns as succeeding or failing based on the number of pledging supporters, or the amount of money pledged via their platform, a DCP instead determines failure or success by referring more directly to whether or not an outcome in the broader world is obtained. This sidesteps a risk inherent to traditional platforms — even if a Kickstarter campaign is a success, according to the platform’s website, the downstream outcome in the world it was intended to help bring about (e.g. the creation of a film, the changing of a law) can still fail to come about. Even the rewards promised to Kickstarter backers of a successful campaign aren’t a sure thing.

Kickstarter does not guarantee projects or investigate a creator’s ability to complete their project. On Kickstarter, backers (you!) ultimately decide the validity and worthiness of a project by whether they decide to fund it.

By triggering redistribution of funds to Backers when an outcome in the broader world obtains, instead of when a number field in the platform’s database reaches a predefined threshold, DCPs are better placed to incentivise and harness the creative agency of their Backers. They can help shape the world to realise the goal in unpredictable ways.

The dark side

With no ‘terms of service’, and a beyond the reach of authorities, DCPs can be used to coordinate around a huge range of goals. You and I are unlikely to approve of all of them. For instance, a decentralised coordination platform closely resembles the design of an assassination market, and could be used with the same intention.

Radical

On the other hand, the carefully curated and authority-compliant coordination platforms we see right now preclude many uses, some of which could be dramatic forces for good. In legacy political contexts this might mean acts of mass civil disobedience, or even toppling a tyrannical regime.

A Kickstarter or a Pledgebank can’t be relied on as a launch pad for initiatives which threaten the most powerful adversaries like states and privileged corporations. These adversaries could uncover identities of organisers and backers and make their lives miserable. Or even disable the service altogether.

DCPs are anonymous by design. There is no central platform recording personal details. An Instigator need not know who the backers are (and they’ll receive any reward due to them nonetheless). This setup creates a much harder target for adversaries who’d like to uncover the identity of backers.

DCPs’ decentralised nature also means that an adversary would have a much harder time shutting the service down. It’s doubtful that they’d be willing and able to pull it off.

I believe that DCPs can function as a powerful lever for undermining the world’s most dangerous and unaccountable entities, that these platforms are coming whether we like it or not, and we ought to be thinking about what that means.

Update 2018–10–05
There’s a problem with the design proposed above that Justin noticed. A successful Goal results in the combined staked funds being distributed equally over all addresses that contributed (regardless of how much they contributed). That means that a user who contributes tiny amounts to the same Goal from many different addresses secures himself a greater cut of the ‘winnings’.

As a fix to this exploit, here’s a redesigned version of what I’m calling the Conductor app:

• An Intiator specifies a Goal, a date by which the success of the Goal will be judged (test date), and a cutoff date for trading Goal tokens (trading freeze date). The Goal is described in the kind of language used to specify prediction market futures. The trading freeze date must be earlier than the test date. The initiator also specifies a trading freeze price. If the Goal token price reaches this level, trading is also frozen (this limits a person’s ability to ‘pump and dump’).
• At any time after the Goal has been specified, until the trading freeze date, Patrons can contribute ETH funds, and Stakers can buy Goal tokens (specific to the Goal they are staking towards), or sell Goal tokens back to the contract for ETH.
• Patrons are users who transfer funds to the contract which will be divided between Stakers if the Goal obtains. If the Goal is ultimately ruled not to have been realised, Patrons’ contributions are returned to them. The combined amount contributed by Patrons to a Goal is called the Patron reward.
• The Conductor smart contract automatically sets the price of Goal tokens that Stakers can buy from or sell back to the contract. The price varies according to the amount of funds currently staked on the Goal (ie. the amount of ETH currently held by the contract, sent from Stakers rather than Patrons). The greater the amount currently staked on a Goal, the higher the buy/sell ETH price of the corresponding Goal token.
• At the test date, a judge (or judges) issue a ruling on whether the Goal that was specified by the initiator was realised in the world. Perhaps Kleros’ system would be used to solicit rulings. Judges would be paid by a portion of the Patron reward.
• If, and after, the Goal has been reached (as determined by a judge or judges), all Staker Ethereum addresses receive their stakes back, as well as a share of the Patron reward reflecting address stake/total amount staked, at the time of the trading freeze.
• If the Goal is not realised, active Stakers lose their stake, (the staked ETH is burned) and Patrons’ contributions are returned to them.