We want to apply our intelligence and resources to promoting social change. To promote major change we will need to create or at least empower institutions. By institutions, we mean groups of people working toward common ends in interdependent ways. This path is not without risks.
We are dubious about schemes that try to make the world conform more closely to some particular a priori vision. So we don’t want to promote change toward a specific vision of “how things should be.” Furthermore, we have seen in many cases (some personally observed, many more described) that institutions often generate perverse incentives, and end up systematically creating effects completely at odds with their founding goals and their avowed mission. So we don’t want to empower institutions that have a substantial risk of becoming these kinds of juggernauts. We see two principles that will help us to avoid both of these problems.
• The goals of social change must be controlled by the value judgments of individual participants. Of course we will try to create an environment in which people will make considered, mature value judgments, but ultimately the choices of what is good and bad, important and unimportant, must be made by the participants.
• Good institutions must be dependent on support from their “customers” (i.e. their purported beneficiaries). Any institution that begins to work against the self-perceived interests of its customers must either reshape itself or (in the worst case) be rendered powerless.
Finally, to ensure that these two principles are effective, good institutions must be made up of fluid relationships, both internally and with their customers. Fluid internal relationships mean that the institution won’t be locked onto a dysfunctional course by its internal structure. Fluid relationships with customers mean that the customers can shift their allegiance if the institution isn’t meeting their needs, forcing it to change or become irrelevant. Given these principles the means for determining and aggregating customer values, and the means for creating and managing fluid relationships become very important.
Our interest in social change is best served by functions that weight customers by individual commitment, creativity, and effectiveness, not by individual income or wealth. Thus participation needs to be weighted in non-monetary terms. One traditional way to do this is by voting (one person, one vote), but this has many problems, including greatly inhibiting fluidity in a winner-takes-all vote, providing very little rational motivation for participation in large groups because of the very low marginal contribution of individual votes, being very susceptible to manipulation by media and subjective images, and ignoring the level of commitment, creativity and effectiveness of each individual.
We are inclined toward approaches which evoke participation based on broad self-interest, and then aggregate values based on the preferences revealed through that participation, weighted by the quality of the contribution to the success of the overall activity. Of course this is circular since the success of the activity will be judged by the revealed preferences of the participants, weighted by their contribution, but we believe that this circularity can be made non-vicious in practice. For example, Google’s page weight propagation algorithm and recent bayesian network “belief propagation” algorithms work just fine in networks with cycles. However, we must be careful to design the circularity resolution to avoid strong lock-in, which would defeat fluidity. For example an institution that becomes a mutual admiration society could shrink down to nothing while maintaining a high weighting for the participants who are “contributing” to its demise.
In addition to direct benefits to a given customer, we expect that their broad self-interest will often include indirect benefits, such as the desire for positive recognition by others and the desire to benefit others even without credit. These indirect benefits will be especially significant when the cost of pursuing them is low, and when the number of people potentially granting recognition and/or gaining benefit is high.
To implement this approach of evoking and aggregating values through self-interested contributions, institutions must have the following properties:
1) Enable participants to join or leave with low cost. This is necessary to fluidity.
2) Offer significant benefits to participants, independent of their contribution. This generates social value and helps to recruit participants.
3) Enable contributions by most participants most of the time. This supports fluidity, helps to generate value and helps to prevent circular lock-in.
4) Depend mainly on contributions that directly and indirectly generate significant benefits for the contributor. This makes the institution largely self-sustaining and helps to recruit contributors.
5) Record and aggregate revealed preferences of participants based on self-interested choices (e.g. following links, filtering, commenting, downloading, tasks completed). This ensures that expressing preferences generates a net benefit to the participant, and that recorded preferences accurately reflect participants’ perceived interests.
6) Feedback revealed preferences as relevant to other participants to guide their use of the institution. This will reduce their costs and increase their net benefits.
7) Make tradeoffs in committing scarce resources based on their anticipated contribution to satisfying the revealed preferences of participants. This will tend to keep the institution aligned with the needs of its participants.
8) Enable all participants to monitor the institution to detect mismatches between preference reporting and social reality. Any way of recording and aggregating preferences will still have failure modes, such as susceptibility to gaming, lock in or missing key participant concerns, and these must be corrected through individual review and social decisions.
Looking at these constraints, we can predict several ways that information infrastructure can enhance such institutions:
• Make participation more frictionless and reliable (joining, leaving, and authentication).
• Increase efficiency of delivering benefits to participants (reduced cost per participant).
• Make contribution more frictionless and reliable (reduced cost of making, evaluating, and integrating contributions).
• Tailor benefits more closely to the preferences of a participant (lower cost/ higher benefit evocation of preferences; lower cost and/or more precise aggregation of preferences; and more complete, relevant and easy to use feed back of preferences).
• Enhance the richness, breadth, accuracy and timeliness of monitoring, and reduce its cost (complete access to aggregate information about activity in the environment, in relatively easily understood and customizable forms).
Existing infrastructure for online activities already does well with relatively frictionless participation, and efficient delivery of benefits (e.g. page views, queries and downloads). However we can reasonably hope for substantial improvements in frictionless contribution (especially evaluating and integrating contributions), tailoring benefits to participant preferences, and enhancing self-monitoring. The infrastructure opportunities for activities that are partially or entirely offline (e.g. third places) are harder to analyze. However we can probably gain some of these benefits in many cases by augmenting offline activities with online infrastructure. However the design issues are somewhat different from purely online activities and need more thought.