Martin Dittus, so far as I know, is the originator of the handy term ‘facilitators paradox’ – the idea that anyone organising a process of collective decision making is always exercising a type of power over participants, even if their goal is the exact opposite: to be egalitarian and democratic.
This is obviously a concern since participatory design is so often couched in terms of it’s ability to transfer power to participants . I’ve come to this problem from an usual angle, since my focus is not participatory design, but designing participatory systems – in particular, mining Twitter for data that can help local governments respond to their citizen’s needs. Participatory design became relevant for me because its widely believe that participatory systems have to be designed with the community they serve to be effective and to be legitimate.
Political science, unsurprisingly, has lots to say about systems of participation, and this post is intended to layout three attempts to deal with the ‘facilitators paradox’ from within that discipline. I will also be using political theorist Steven Lukes’ conception of power to set up the problem.
In any participatory process there’s always a background set of decisions which the facilitator will unavoidably have taken, trying not to do so leads to an infinite regress. Imagine designing a playground through a participatory process, we might have a vote to decide what colour to paint the fence. But who decides the choices of colour? What about other finishes? Who gets to vote, who decides the criteria for a legitimate community member? Who chooses what we vote on, or the system by which the votes are counted? Of course, you could discuss any one of these issues, but then who decides how to organise that discussion?
The answer is clear – the only way for the facilitator to truly exercise no power over the community is to never leave their desk and never contact that community. Equally clearly, we should not abandon all ideas of participation, a course of action that is certain to disempower participants: the perfect is the enemy of the good.
Lukes addresses exactly these issues . He contends that power has three dimensions, as highlighted in the following three cases:
- Cases where “key political decisions in which the preferences of the hypothetical ruling elite run counter to any other group” (Quoting Dahl) 
- Cases where “A devotes his energies to creating or reinforcing social and political values and institutional practises that limit the scope of the political process to public consideration of only those issues which are comparatively innocuous to A” (Quoting Bachrach and Baratz) 
- “A person may exercise power over B … by influencing, shaping or determining his very wants” (Lukes himself)
Case 1: Opinions actively suppressed
This scenario could manifest itself in many subtle ways, but a caricature example might be that of a king who imposes very high taxes on his subjects, to which they strongly, publicly object but about which they can do nothing. We have clear empirical evidence that the king is acting against the interests of the people, and their opinions have never been considered. In some sense the very existence of a participatory design process mitigates against this kind of power, at least within the scope of the project. Any process that can at all be referred to as participatory is the opposite of actively ignoring collective will.
Case 2: Opinions covertly excluded from discussion
Lukes gives the real world example of two American towns, one of which votes for clean air laws to improve the environment. In another, similar town where there is a steel plant, the same laws are never passed. This is not because of Case 1 style power, the management of the steel plant do not ignore the clearly expressed views of the local people. Instead, they prevent the passage of clean air laws by preventing them for coming up for discussion or voting. Even though there is a system to pass such laws, the steel plant uses its power in the background to further its own interests. Again, here we can see empirical evidence of this type of power, for example in the comparison between the two towns.
In participatory design, we might see this kind of power as being exercised when the scope of the project is set to exclude some important issue. In the example of the designing a playground, we could imagine that it is replacing an older, much loved park that is going to be redeveloped. If the key sentiment the community wishes to express is that it wants to prevent the closure of the old park then there will be questions about the value of a participatory process that prevents this outcome.
Even within the scope of a participatory project we can imagine this kind of power at play. If the facilitator chooses what subjects are discussed they are necessarily steering the way decisions go.
Case 3: Voting against your own interests
This is the major philosophical threat to any collective decision making process, a kind of insidious, unobservable power. Lukes gives the example of an inheritance tax once considered in the United States, christened the “Death Tax” by opponents. Many people believed, incorrectly, that they would be subject to the tax and so the proposal became very unpopular. In fact it was only to be imposed on the extremely rich, and the money gathered through it would have been widely distributed throughout society. In this example it’s suggested we can see a ‘ruling elite’ influencing perceptions using emotive language so that many people expressed opinions which are against their own interests. In this case a minority interest wins despite a ‘free’ debate on the topic.
While we might all accept this type of power is widely exercised it’s on a dangerous empirical footing. How do we know that the proposed inheritance tax was really against the best interests of the majority? Perhaps many people felt taxing inheritance represents a moral issue that overrode their own desire for financial gain. Who are we to decide? If we decide that it’s acceptable to ignore people’s expressed opinions ‘in their best interest’ isn’t that a kind of dictatorship?
We could term this type of power ‘preference shaping’. The question is not so much whether preference shaping takes place, it clearly does. As Przeworski  says:
In a society in which interests are in conflict, the fact that various economic agents spend money to persuade others constitutes prima facie evidence that someone is irrational. Either those who spend money to communicate are throwing it away or these costly messages persuade others to hold beliefs that are not in their best interest.
The question is rather how to ‘unshape’ preferences, or access ‘true’ preferences.
Attempts to ensure equality across all three dimensions of power
Cases 1 and 2 concern the issue of how to collect together individuals preferences in a legitimate and effective way. There is a large literature on the difficulties involved in that process which I won’t go into here: much of it concerns voting systems and participatory design rarely hinges on voting. Cases 1 and 2 also seem to imply some kind of self-serving intent on the part of the facilitator, which we can at least hope is not so much of a problem in small scale participatory design. The third case, however, must often be present in PD. It’s frequently the essence of PD projects that experts in some particular field are working with non-experts – this seems almost by necessity to represent a power dynamic.
Attempt 1: Statistically adjusting for non-expertness
In his book The Myth of the Rational Voter  Bryan Caplan attempts – and I realise that his approach will seem dubious to many people – to calculate how opinions would change if members of the public were trained as economists. To do this he uses data from a survey (the Survey of Americans and Economists on the Economy) that solicits opinions from economists and non-economists on a range of economic issues . Unsurprisingly, the two groups give systematically different answers. This in itself is notable: many economists are extremely reluctant to countenance the idea of the third dimension of power. Economics is often conceptualised as finding the optimal way to satisfy people’s preferences, if preferences are susceptible to shaping then they might seem like an unreliable foundation on which to build.
A common defence is that although many people may have random variations in their preferences, in large numbers these random variations will cancel out. Caplan demonstrates that preferences are not (just) subject to random noise, but are systemically different between experts and non-experts.
That might be because the economists are wiser, but it could also be because economists are disproportionately male, rich or white. Fortunately, respondents in the survey also provided a wide variety detailed demographic information, so he is able to use statistical regression to remove these effects, in attempt to discover what an average person would think about issues of economic policy, if they benefited from a PhD in economics.
In this way, he is able to simulate the view of what he calls the ‘enlightened public’. Using the following scoring system, they were asked how much various factors hindered the economy:
0 – no reason at all
1 – minor reason
2- major reason
For example, subjects responded to the question “Foreign aid spending is too high”. The public at large gave this proposition an average score of 1.4 (strong agreement) while economists gave it a score of 0.1 (strong disagreement). The simulated enlightened public gave it a score of 0.3 (strong disagreement). This, Caplan asserts, indicates that all most all of the widespread antipathy towards foreign aid is connected with a lack of information. Elsewhere in the book Caplan cites a statistic that indicates 41% of Americans believe foreign aid is one of the two most significant national budget expenditures, in fact it represents only 1.2% of the budget. It seems likely that the opinion of the non-expert public is mostly misinformed, that is, that their preferences are shaped by a lack of information.
Through this method, which he repeats for a number of other questions, Caplan attempts to discover what an informed public would think, that is, a public free from the exercise of the third dimension of power over them.
One serious problem with Caplan faces is that he cannot control for the variable “kind of person who wants to become an economist”, while another is that he is extremely lucky in finding such useful dataset in the SAEE survey.
Attempt 2: Taking account of cognitive biases
In Sunstein’s Health-Health Tradeoffs  he considers difficult policy decisions around risk, and asks how those decisions could be made more sensitive to the way citizens actually perceive risk. He provides some examples of the type of tradeoffs he is considering:
Fuel economy standards, designed partly to reduce environmental risks, may make automobiles less safe, and in that way increase risks to life and health. Regulations designed to control the spread of AIDS and hepatitis among health care providers may increase the costs of health care, and thus make health care less widely available, and thus cost lives. If government bans the manufacture and use of asbestos, it may lead companies to use more dangerous substitutes.
Normally, when policy makers consider these types of risks they assume that one type of fatality is exactly the same as any other – being hit by a bus is the same as a heart attack. Sunstein compares this approach with data from Viscusi’s Book Fatal Tradeoffs . Viscusi uses a well attested method of contingent valuation  to discover how people actually think about risk. Participants are asked to state how much they think it would be appropriate to spend to prevent certain kinds of fatalities. For example on average preventing a single ‘unforeseen instant death’ is given an average value of $2 million, while preventing a death from lung cancer is valued at $4 million. Sunstein derives some patterns from these statistics, for example noting that risks arising from man-made sources are ranked as warranting higher levels of preventative expenditure compared with naturally occurring risks.
Sunstein suggests ‘Economic approaches promise to avoid some of the problems of expert valuations.’ This work has interesting symmetrical relationship to Nudge, a book he later co-authored with Richard Thaler . In Nudge the authors advocate that policy makers should consider the ‘non-rational’ behaviour of those subject to a particular policy – a detail of implementation of policy. In Sunstien’s Health-Health trade-offs he instead argues for the inclusion of similar complex behaviours in the policy objectives themselves.
Attempt 3: Deliberative polling – structured access to experts
Deliberation is thought to offer an important way both of avoiding preference shaping and also of avoiding various issues with counting votes, with continental and analytic philosophers arriving a roughly similar conclusions:
… arguments advanced by Habermas and Rawls do seem to have a common core: political choice, to be legitimate, must be the outcome of of deliberation about ends among free, equal and rational agents. 
Fishkin has a rather neat definition of deliberative polling:
Ordinary polls seek to gauge the opinions people actually hold, Deliberative Polls to gauge the opinions they would hold if they knew and thought more.The design provides random samples with information and gives them the opportunity of discussing
the issues with one another and questioning policy experts about them.
Again, the idea is to access ‘unshaped’ preferences lurking behind superficial responses. Among the benefits of deliberation are the possibilities of ‘revealing private information’ and ‘lessening or overcoming bounded rationality’ . So can we see it in effect empirically?
Yes. For example, in Disaggregating Deliberation’s Effects:An Experiment within a Deliberative Polling , Fishkin is able to demonstrate exactly the effects we might expect from access to expert information. Over a weekend, a representative sample of residents in New Haven, Connecticut were split into groups and given opportunities to hear expert testament and debate. They discussed two issues, the highly controversial expansion of a local airport, and a much less contested issue surrounding what if any sharing there should be of property-tax revenues from new commercial development.
Over a weekend each group is presented with evidence about the likely effects of various policies by a panel of experts. In plenary sessions the group are able to put questions to the experts, and there is time for discussions within, but not between, the groups.
Surveys administered before and after indicate that the groups all shift their views over the course of the weekend, seeming to support the hypothesis that participants true preferences can only come to the fore through the provision of more information and space for discussion. Larger changes were observed for the less contested tax-sharing issue, compared with the controversial airport expansion. The authors attribute this to the fact that many participants already had some information about the airport through coverage in local newspapers.
In Fishkin’s book  on deliberative polling, he goes on to discuss a number of other similar projects, and highlights two possible effects. Firstly, he looks at polarisation and groupthink, where people discussing a topic will naturally tend to more and more extreme positions. This theory is attributed to Cass Sunstein (mentioned previously). Fishkin uses statistical evidence from his research to demonstrate that this is not a powerful or widespread effect.
More relevant to participatory design, Fishkin looks at what he terms ‘domination’, where the privileged are able to make their voices louder than others. When participants selected for a participative democracy project, they are selected so that they are representative of the underlying community. But, what if attendance is not enough:
Some people, even if formally included, may not have their voices, if they speak at all, taken seriously. They may give off cues that indicate they are not well informed or not worth listening to.
Domination, which I would suggest is an example of the second dimension of power, will show up empirically in participatory democracy. In before and after surveys, it ought to be possible to see views shifting to those of the ‘privileged’ – those with more money, higher social status etc if domination is occurring. This is not widely evident: in Fishkin’s project consulting a population about airport expansion, there was no evidence of views being systematically shifted in ways that might be indicative of domination.
Alice Siu’s  research looked at five deliberative polls, breaking down the number of words spoken by participants down into demographic categories. Demographics that might be expected to exhibit domination, such as educated white males, did not have higher word counts than others – which might tend to indicate that domination did not take place.
Firstly, Lukes’ dimensions of power may be a useful framework for participatory designers seeking to reduce the impact of the ‘facilitators paradox’ and think about the practice more generally. I’m not aware of it’s use in design, however it’s widely discussed across participatory and deliberative politics, as well as in projects in developing countries .
Secondly, the work across political science indicates several ways in which participants preferences might be ‘shaped’ by the third dimension of power – either through lack of expertise, cognitive bias, or lack of information.
Thirdly, perhaps the techniques for empirically observing the various dimensions of power discussed above could be applied to benchmark various approaches to participatory design. In design features such as domination and polarisation should be observable through the same methods.
Finally, the very close, and recursive similarity between the agendas of deliberative democracy and participatory design are something that might bear further examination.
 Kensing, Finn, and Jeanette Blomberg. “Participatory design: Issues and concerns.” Computer Supported Cooperative Work (CSCW) 7.3-4 (1998): 167-185.
 Lukes, Steven. Power: A radical view. Vol. 1. Macmillan: London, 1974.
 Dahl, Robert A. “The concept of power.” Behavioral science 2.3 (1957): 201-215.
 Bachrach, Peter, and Morton S. Baratz. “Two faces of power.” American political science review 56.04 (1962): 947-952.
 Elster, Jon. Deliberative democracy. Vol. 1. Cambridge University Press, 1998. Przeworski chapter.
 Caplan, Bryan. The myth of the rational voter: Why democracies choose bad policies. Princeton University Press, 2011.
 Sunstein, Cass R. “Health-health tradeoffs.” The University of Chicago Law Review (1996): 1533-1571.
 Viscusi, W. Kip. Fatal tradeoffs. Oxford University Press, 1992.
 “Report of the NOAA panel on contingent valuation.” (1993): 4601-4614.
 Leonard, Thomas C. “Richard H. Thaler, Cass R. Sunstein, Nudge: Improving decisions about health, wealth, and happiness.” Constitutional Political Economy 19.4 (2008): 356-360.
 Elster, Jon. Deliberative democracy. Vol. 1. Cambridge University Press, 1998. Introduction.
 Elster, Jon. Deliberative democracy. Vol. 1. Cambridge University Press, 1998. Fearon chapter.
 Farrar, Cynthia, et al. “Disaggregating deliberation’s effects: An experiment within a deliberative poll.” British Journal of Political Science 40.02 (2010): 333-347.
 Fishkin, James. When the people speak: Deliberative democracy and public consultation. Oxford University Press, 2009.
 Siu, Alice. Look who’s talking: Examining social influence, opinion change, and argument quality in deliberation. ProQuest, 2009.
 Gaventa, John. “Finding the spaces for change: a power analysis.” IDS bulletin37.6 (2006): 23.