We are pleased to announce the winner of the 2019 Foresight Edition of the Incentive Prize on Incentives:

 

A Simple Secure Coordination Platform for Collective Action

I propose a simple, streamlined, and secure platform for enabling coordinated collective action
and for revealing hidden but shared preferences. Its core function would be to collect verified
“signatures” supporting a particular statement, which would only be publicly revealed upon the
satisfaction of some criterion such as a threshold number of signatories.
Many social problems can be understood as coordination problems, in which a large
group shares a preference that could easily be satisfied by collective action, but for which there
is a significant barrier or disincentive to individual or small-group action. These barriers could
range from minor financial cost or fear of revealing an unconventional opinion all the way to
major legal, social, political, or physical retaliation.
There is increased interest of late for mechanisms that bridge the gap by allowing people
to commit to a particular action that is triggered by sufficient commitment. Kickstarter is an
excellent example, in which commitment is made via payment to the platform, and is (partially)
refunded in the case of unsuccessful projects. This idea has also gained great currency among
proponents of blockchain technologies and “smart contracts.”
But while many collective-action problems may require credible commitment in terms of
providing real resources or actual actions, there are many important cases in which just the
commitment to have one’s identity revealed could be enough to address coordination problems.
As a few broad examples:

1. Political parties can force individual politicians to “hew the party line” against even a
rather popular position by implicitly threatening retaliation; this leads to partisan,
polarized politics. Taking a public position is an action for a politician, so coordinated
revelation of shared but secret policy preferences could allow for cooperation across
party lines with the risk spread over many.
2. Whistleblowers could share risk across a much larger set of individuals, where the
accusation could be made anonymously (and should garner little credibility), but those
willing to provide testimony or information backing up the accusation could have their
identities revealed only as a large group, and thus grant real credibility to the accusation.
3. A petition advocating for a controversial position (for example in an authoritarian political
environment) could be made public only upon a sufficient number of signatures, or
possibly when enough sufficiently prominent people sign on.

Many, many other use cases are likely.
The proposed system would be quite simple in initial form, but with high-security
guarantees. Via a simple and easy-to-use interface, users would list a Statement and a
Triggering Condition. Any user with a verified identity (associated with a token on an
authentication system) could then “sign” a given Statement. Their signing is cryptographically
stored in a second system so that it is impossible to read the list of tokens, and the fact that
signing took place is provided (with no identifying information) to a third system. If and when, on

the basis of information supplied to the third system, the Triggering Condition is met, the third
system provides a cryptographic key that can be combined with data from the user
authentication system and the signatory database to generate a list of users who signed that
statement, which is then made public. Using this (or a similar) system, users can have a high
degree of confidence that their identities will be made public only in the event of the Triggering
Condition they signed up for. After a certain pre-defined time interval, if the Triggering Condition
is not met, the encrypted signatory data would be destroyed.
This system would, preferably, be open-sourced and run by a nonprofit at low cost (it
seems unlikely that there would be a good business model, and a business could create
conflicts-of-interest). The development cost/effort would be modest given that the functionality
is simple and the system should employ largely “off-the-shelf” ingredients. (There are many
ways — for example on blockchain — one could envision such a system; the idea here is to
focus on simplicity, ease-of-use and sign-up, and robustness). It could be tested first in relatively
small groups and lower-stakes settings, then grow to larger groups and higher stakes. There
are many expansions that could build upon this simple system, allowing for example more
complex triggering conditions or commitments.
The proposed system is not risk-free. No security is absolute, and there would always
be some risk of system compromise of one sort or another. Moreover, it may not always be
good to reveal hidden preferences; if widely adopted this system could enable collective action
of many types, not all of which would necessarily end up being good! However, I hold a strong
belief that on average, allowing more coordinated action is likely to be broadly beneficial.

Anthony Aguirre
Future of Life Institute
Metaculus
anthony@futureoflife.org