Influence Without Powers

The Corruption Perceptions Index, published by Transparency International since 1995, carries not a gram of legal force. It obliges no one, compels nothing and entails no consequences. Yet governments hire consultants in order to rise in it, and announce a rise as an achievement. Credit rating agencies are private companies whose opinion of a state's solvency moves the cost of its borrowing by billions.

The lesson. A number published systematically acquires an influence that no one granted it. The mechanism is regularity, comparability and citation by third parties.

The World Bank published the Doing Business ranking, which assessed the conditions for doing business in a hundred and ninety economies. Its influence was enormous: states rewrote legislation in order to rise in it. After irregularities in the 2018 and 2020 reports were reported inside the bank in June 2020, publication was suspended and the review was entrusted to an outside law firm. The investigation established that pressure had been put on the team that prepared the index to alter the indicators for particular countries. On 16 September 2021 the bank announced the discontinuation of the ranking. Not a suspension - a discontinuation. The bank did not restore it: its successor, B-READY, was built anew, under a different name and with a different methodology.

The lesson, the reverse of the first and more important. The whole capital of such a tool is trust in the procedure. It is not spent gradually; it is lost entirely and at one stroke.

The IPCC has no powers at all: it decides nothing, obliges no one and allocates not a penny. Yet its assessment reports set the frame of every climate negotiation - the argument runs inside its conclusions, not outside them.

Where does the influence come from? Not from the authority of the scientists: authoritative scientists are many, and their opinions diverge. The influence comes from the procedure of assembly: open review, published comments and the replies to them, recorded disagreements, line-by-line approval of the final summary with governments taking part. There is nothing to contest, because the process is visible in full, including the places where there is no agreement.

The lesson. What makes something impossible to ignore is not the content of the utterance but the transparency of the procedure by which it was obtained.

The standards on which the internet runs are set by the IETF and the W3C - associations constituted by no state, acting voluntarily and holding not the slightest powers. The IETF's working principle, formulated in 1992, runs: "rough consensus and running code". The world economy stands on those standards. No one asked anyone for permission.

The lesson. If what you produce is needed by everyone and is obtainable nowhere else, status comes afterwards and by itself.

Ireland twice changed its constitution as a result of the work of bodies that held no powers whatever.

The Constitutional Convention of 2012-2014 recommended allowing same-sex marriage - at the referendum of 22 May 2015 this was adopted by 62 per cent of those voting. The Citizens' Assembly, which parliament constituted in July 2016 and which concluded its work in the spring of 2018, recommended repealing the Eighth Amendment - at the referendum of 25 May 2018, 66.4 per cent voted for repeal on a turnout of 64.1 per cent.

The compositions were selected differently, and the difference is instructive. Of the hundred members of the Convention, sixty-six were citizens chosen by lot, thirty-three were serving politicians, and the hundredth was an appointed chairman. The Citizens' Assembly went further: its ninety-nine participants were selected by lot, every one of them, there were no politicians in the composition at all, and the chair was appointed separately. In both cases the participants studied the question for several months, listened to specialists and to both sides, discussed publicly and voted.

The lesson. Legitimacy is conferred not by election but by a transparent procedure of selection and open discussion. People selected at random, having got to grips with the question, turned out to be more convincing to society than elected politicians.

The Russell Tribunal on Vietnam of 1966-1967 and the Permanent Peoples' Tribunal established twelve years later delivered carefully prepared findings on the gravest questions. Their composition was self-appointed, their representativeness in no way established, their powers absent. Morally they are respected. Politically they are ignored completely, and have been for decades.

The lesson. Unless it can be verified who is counted, any procedure, even a flawless one, remains the private opinion of a group of respected people.

Put the six cases together and you get a list of requirements that depends on nobody's sympathies.

First. The tool produces what is needed and what is not obtainable anywhere else. Not an "important opinion" but data or a procedure whose absence gets in the way of someone's work. Second. The procedure is visible in full - not only the result but how it was obtained. Third. It works regularly and comparably: a single measurement is news, a series is movement, and movement has to be reckoned with. Fourth. All sides use it alike: a tool useful to one side is a weapon, and it is rejected along with that side. Fifth. Who is counted can be verified: it is known and confirmable that living unique people stand behind the measurement, each of whom spoke once and willingly.

The first four properties are not invented - they are adhered to, and are therefore open to anyone who accepts the discipline. The fifth is of a different nature: it cannot be adhered to by intention, it can only be built.

There is a second objection, stronger than "you have no powers": polling millions of unprepared people on a complex question is no better than polling a thousand unprepared people; it is merely larger. The objection is fair. Scale by itself confers no quality of judgment, and a large number of ill-considered answers is a large number of ill-considered answers.

The closest to a solution is the deliberative polling that James Fishkin has conducted with Stanford University since 1988: a random sample, the study of vetted materials, discussion in small groups, questions to specialists of opposing views, and a repeat measurement. More than a hundred and fifty such exercises have been carried out in over fifty jurisdictions. There is selection by lot there, and preparation, and measurement before and after. Two things are missing: scale, and a way to verify who is counted.

The second closest case is the Global Assembly of 2021: a hundred people selected by lot from across the world so that the composition matched the population of the planet by age, sex, origin and income; sixty-eight hours of work over eleven weeks; the final declaration read out at the climate conference in Glasgow. This is, by all appearances, the upper limit of depth reached to date - and it stops at a hundred people.

From this the fork is visible at which all such attempts stand. Citizens' assemblies have depth without scale: a hundred people have got to grips with the question, but they cannot speak on behalf of society. Referendums have scale without depth: millions have voted without getting to grips with the question, often on a wording composed by those in power. What prevents the one from being joined to the other is exactly what was said above: the impossibility of selecting people honestly and counting them honestly at a distance.

Hence too the practical value of a verifiable registry of unique people for those who run such assemblies: the weakest place in any of them is where to get a list such that no one will contest the randomness of sampling from it. The registry solves that problem, and solves it neutrally: selection by lot has no position and is equally fit for any side.

What follows from all this for the design of a measurement, and why the fifth property turned out to be decisive, is in the document Civic Voice.

• The Transparency International Corruption Perceptions Index - methodology
• The discontinuation of the Doing Business ranking by the World Bank, September 2021 - statement
• Business Ready (B-READY), the successor to Doing Business with a new methodology, first report 3 October 2024 - World Bank press release; B-READY FAQ
• The procedure for preparing and approving the reports of the Intergovernmental Panel on Climate Change - IPCC procedures
• The working principles of the Internet Engineering Task Force, including the formula "rough consensus and running code" - RFC 7282
• The Citizens' Assembly of Ireland of 2016-2018 - assembly materials; the Constitutional Convention of 2012-2014 - convention archive. Dates and results of the referendums: 22 May 2015 (same-sex marriage), 25 May 2018 (the Thirty-sixth Amendment, by which the Eighth was repealed, 66.4 per cent on a turnout of 64.1 per cent)

Return to the document Civic Voice