What is conscious?

Philosophers have thought up examples of entities that behave like thinking human beings, but that leave us wondering whether they are conscious. The China brain has been debated for many years. In recent months, Eric Schwitzgebel has discussed other entities, including the United States, on his blog (see posts dated 31 October 2011, 4 May 2012 and 19 June 2012):

http://schwitzsplinters.blogspot.com/

The more I think about examples like these, the more I think that we should not say there is always a fact of the matter, out there, as to whether a given entity is conscious. Rather, we should ask whether it makes sense for us (or whichever rational beings happen to be having the discussion) to regard the entity as conscious. Whether it makes sense will depend on the nature of our social interactions with the entity, our views on moral obligation to the entity, whether we see the entity as made up of smaller entities that we see as individually conscious, and lots of other things.

We must regard other human beings, and a fair number of animals high up the evolutionary scale, as conscious. But human beings can disagree as to how far down the scale to go. Martians can also disagree with human beings about the consciousness of at least some of the entities that human beings must regard as conscious, and we can disagree with Martians about the consciousness of at least some of the entities that they must regard as conscious.

(This assumes that Martians have a concept that corresponds to our concept of consciousness. They may not have. If my general approach is right, a possible reason for them not to have would be that they might well not have concepts that corresponded to our concepts of social interaction or of moral responsibility.)

Then a dispute about whether the China brain is conscious, or whether the United States is conscious, can be seen as a dispute about the relative significance of the members of two groups of indicators of consciousness. The first group, on which such entities score highly, includes the sophistication of processing and the existence of a generally consistent, yet gently mutable, character of conduct, that differs, but does not differ radically, from the characters exhibited by other, comparable, entities. The second group, on which such entities score badly, includes the personal nature of our interaction with the entities, the existence of feelings of moral responsibility towards them that are very similar to our feelings towards other human beings, and a sense that the entities have qualia of experience. (I do not mean to claim that qualia are real, only that most of us, in our everyday lives, think that they are real.)

The example that Eric Schwitzgebel cites in his post on 19 June 2012 presents a new challenge. There is an artificial body, which looks to us like a person, and behaves appropriately. But there is a China brain arrangement in the background, feeding instructions to the body, rather than a normal brain in the body. We are presented with a single body, with which we can interact as we would with a person. So this example scores highly on personal interaction, and might easily come to score highly on being regarded as an object of moral responsibility. The one thing about which we would still worry would be the qualia (or whatever our views on the human mind allowed along the lines of qualia).

Another interesting example is the David character in the film A.I. This is an artificial child, the capacity of which to display love towards the human being who acts as its mother can be switched on, but cannot then be switched off. Once this capacity had been switched on, and the "love" had developed, could the mother argue that the creature was just a machine, to which she had no moral responsibility? I rather think that it would depend on how the programming was done. If the intelligent processing of data from the child's environment went on deep inside, but it was only near the surface, in a separate module, that appropriate behaviour was generated, then the mother would have less of a moral obligation than if the intelligent processing and the generation of behaviour were fully integrated. I have not worked this out properly, but if there is something in this idea, and if considerations of moral responsibility are relevant to the attribution of consciousness, then the details of implementation of processing could matter to the attribution of consciousness.

I have cross-posted these thoughts, with minor amendments to allow for the context as a comment on Eric Schwitzgebel's post of 19 June 2012, on his blog at:


http://schwitzsplinters.blogspot.com/2012/06/chinese-room-persona.html

Cash and transparency

This week, a study of the extent to which different think tanks disclose their donors was published. It is available at:

http://whofundsyou.org/

Those that got high ratings were no doubt pleased. At least one that got a low rating did not accept the presumption that transparency was a virtue, and argued that donors had a perfect right to privacy, as can be seen here:

http://blogs.telegraph.co.uk/finance/timworstall/100018107/it-doesnt-matter-who-funds-think-tanks-but-if-it-did-left-wing-ones-would-do-particularly-badly/

Meanwhile, in the United States, there has been considerable concern about the use of supposedly independent Political Action Committees, or PACs, to circumvent limits on politicians' campaign spending. And some of the big corporate donors to political campaigns don't even want to have their names disclosed:

http://www.politico.com/news/stories/0512/76919.html

So what would be a sensible position, given the tension between:

(a) the prima facie right of each person to decide whether or not to disclose his or her spending (on anything, not just on politics) and whether or not to disclose his or her political views; and

(b) the need to do what we can to prevent the corruption of the political process by those of the rich and powerful - I hope a fairly small proportion - who try to corrupt it?

The kind of corruption I have in mind is the twisting of legislation and government administration to suit the private interests of those who spend money to get certain politicians elected, or to lobby the politicians who get into power. It amounts to corruption because legislation and government administration are imposed on all of us, without the freedom to opt in or out: they should therefore be in the interests of all of us, not in the interests of a few. The spending of lots of money in the marketplace, promoting the production of the goods and services that the rich happen to like, does not amount to corruption, because we are free to participate, or not to participate, in any given market, and because the production of some goods and services, at high prices, does not prevent the production of others, at lower prices.

I think that the answer depends on the current state of the polity.

If we have a healthy, free, democratic polity, or one that has only wandered a little way from that ideal, then disclosure will help to keep it healthy. I would therefore favour full disclosure in the UK, whether or not the donees are political parties. Some think tanks say that they are not party political. Such claims are often true. But they still seek to change legislation, and if they may be promoting the interests of their funders, whether because the funders ask them to or because they decide their policies first and then naturally attract the funders who agree with them, that should be disclosed. When changes to legislation or to government administration are being advocated, we need to be aware of possible selfish motives, so that we can appraise the arguments being put forward with an appropriate degree of scepticism.

An important counterpart to this is total transparency on the side of government. All papers related to the conduct of government should be freely available, except when national security would be put at risk. That should help to prevent corruption in the reverse direction, for example when a local council might refuse planning permission for new business premises because the proprietor was known to support political views that were at variance with those of councillors.

Given those conditions, I do not see the right to privacy as carrying much weight. It is not even clear to me that we do have a right to privacy against anyone except an intrusive government. (I find Article 8 of the European Convention acceptable only as a right against the state: if it is a right against the press, we can say farewell to a free press. But that is another argument.)

If, on the other hand, there is a repressive government, secrecy may be essential in order to have a chance against the authorities. But in that position, opposition movements would probably be breaking the state's (unjust) laws anyway, and the state would inspect bank accounts, whether or not it was authorised to do so. Laws on privacy of funding would then be neither here nor there.

On keeping one's distance from those who lack epistemic virtues

Suppose that X works with Y in some business in which knowledge is important, and in which people need to draw sensible conclusions from evidence, and to recognize and suppress wishful thinking when the conclusions are not what they might expect or like. X reckons that Y exhibits the appropriate epistemic virtues, and is therefore a good colleague to have.

Now suppose that X finds that in some unrelated area of life, Y holds a belief that X thinks no reasonable person could hold, if that person were confronted with evidence that is plainly available to Y, and that Y could plainly grasp and understand how to use.

Should X be less happy about working with Y? I think not. Y's performance at work would be evidence that what X saw as Y's lack of epistemic virtue in the unrelated field had not infected Y's work. And while it would be a bit much to ask X to acknowledge that he or she might be wrong about the unrelated matter, it would not be unreasonable to ask X to acknowledge that his or her perception of Y's lack of epistemic virtue might be mistaken. Y's reasoning processes would not be likely to be fully transparent to X.

X could respond to this point by saying that the reasoning processes did not matter. Y's belief was so manifestly absurd that Y should have said, "I must be wrong here, now I should try to find the error in my reasoning". On that basis, Y would be guilty of one specific epistemic vice, a failure to recognize manifest absurdity. But even then, could X be sure that Y suffered from that vice? Perhaps the process of reasoning had itself led Y to change his or her view of what was absurd.

Now let us change the example. X is considering whether to work with a think tank, T, on some project. X is impressed with T's work in the relevant field. But X also knows that T's official views, in unrelated fields, are quite as bad as Y's conclusions in an unrelated field. They are not just mistaken. X cannot see how any rational person, confronted with the widely available and easily understood evidence, could reach those conclusions.

Should this deter X from working with T? There might be a risk to X's reputation, if he or she were seen to be working with an organization that X's peers might well regard as crazy, but we shall set that to one side, and concentrate on the question that arose as between X and Y. Would it be appropriate for X to think there was a serious risk that what X perceived as T's lack of epistemic virtue would infect work in the area of the proposed joint project? (There is a side issue as to whether institutions, as opposed to individuals, can have or lack epistemic virtues.)

It would not be hard to say yes, the risk should be taken more seriously in the case of X and T than in the case of X and Y. If an institution adopted crazy views, that would be likely to reflect the views of more than one person. There might be only one person formulating views on the topics in question, but he or she would be answerable to the institution's management. The management would therefore have a general outlook that allowed the views to be published, whether an outlook that staff should not be controlled, or an outlook that included sharing the crazy views. And that management outlook might very well infect the recruitment and the management of those who would work on the proposed joint project. People who lack epistemic virtues may well associate with, recruit, and encourage other people who also lack those virtues.

Such a conclusion would have an interesting implication. The conclusion would suggest that epistemic vice could spread more easily from one area of thought to another in a group of people than within a single person, despite the fact that a single person seems to be much more closely integrated than a group of people. That is not, however, absurd. In a group of people, propositions are expressed by some and are consciously considered by others. That stage of conscious consideration may given the propositions more power to influence behaviour than if they were merely present in a single cortex, encoded in a form that did not even look particularly propositional, and were occasionally and dreamily considered by the subject.

Examinations, earthquakes and the Euro

There is a well-known paradox of the surprise examination. A teacher tells some students that there will be an examination one morning next week (Monday to Friday), but that they will not know until the morning of the examination that it is on that day.

The students reason that it cannot be on Friday, because then they could work out the day by Thursday evening. But if they know it cannot be on Friday, it cannot be on Thursday either, because they would be able to work out the day by Wednesday evening. They continue to reason on these lines, and conclude that the examination cannot be held at all, under the conditions stated by the teacher.

It is important that the timespan over which the students would be at risk has an end-point, Friday, backwards from which the students can reason. If the days at risk went on for ever, they could not reason as they do. (It is an interesting question, whether the end-point needs to be a determinate one, up to which they are at risk. I think that is not necessary. That is, it would suffice for them to be able to identify a specific date, a finite time in the future, beyond which the period of risk had definitely ended. It would not matter whether they were definitely at risk up to that date.)

It is also important that the examination would definitely be held within the finite timespan. If the teacher only said that there might be an examination next week, and that if there were, it would be held on a day that the students could not predict before that day, the students could come to school each day, unsure of whether there would be an examination that day.

It is also important that the students would definitely be unable to identify the day of the examination before that day. If they were told only that they might not be able to identify the day in advance, then Friday would be a possible day. Then it could not be eliminated, and the other days would also be possible days.

We may note that there is no requirement for the teacher to have decided, at the time of the announcement to the students, the day of the examination. The paradox would arise if the teacher planned to pick a day on impulse, as the week wore on. The teacher, wanting to meet the conditions, would know that he or she could not leave the examination until Friday, and would also know that the students knew this. The teacher would therefore know that he or she could not leave the examination until Thursday, and would also know that the students knew this. This chain of reasoning in the teacher's mind would lead the teacher to the same conclusion as the students. The examination could not be held at all, under the stated conditions, even though the teacher's initial proposal of a surprise examination looked perfectly reasonable.

We can also see that human intention only matters to make the story plausible, not to generate the paradox once we have the story, by considering a predicted earthquake.

Having studied the ways in which stresses have built up, we may conclude that there will be an earthquake at a given location on some day between now and the end of 31 December 2100, that it will be the only earthquake there before 2500 (because stresses will be relieved by the earthquake and will take a while to build up again), and that we will not know, until the day of the earthquake, that it is on that day.

Just to make the problem as similar as possible as that of the examination, let us assume that our methods give us no way of telling when the probability of an earthquake within the next year, or the next month, or any other future time period shorter than the time to 31 December 2100, is rising (apart from the rise due to the fact that as time passes, there are fewer days left to go). We cannot see it coming, and each remaining day in the given time period will always seem to us as likely as any other remaining day.

The lack of human agency makes the story implausible. There is no-one out to keep us in ignorance. But if we can overlook that, we can see that given the conditions, the earthquake could not be on the last day, because we would know by the end of 30 December 2100 that it would be on that day. Therefore it could not be on the penultimate day, and the paradox would arise all over again.

One approach would be to say that while the students, or the potential victims of the earthquake, could not say that they were in some unidentified member or other of a set of possible worlds (the examination-set being the world in which the examination is on Monday, the world in which it is on Tuesday, and so on to Friday), they could say that they were in some superposition of worlds, and that this superposition would be collapsed into a particular world by the teacher's announcement, or by the first movement of the tectonic plates. Analogies with quantum mechanics may be fun, but I fear that they can also degenerate into mere hand-waving. I shall not pursue this approach here.

Instead, we can improve our understanding by looking at a feature of the mathematical structure of the problem. Not only is there a finite timespan of risk. That timespan contains a finite number of risk-points, the individual days. For each such risk-point, there is an immediately preceding risk-point, the day before. That allows the paradox to be generated. To take the example of the examination, there is a last risk-point, Friday. The examination cannot be held on that day. Therefore, the immediately preceding risk-point, Thursday, effectively becomes the last risk-point. But the examination cannot be held at the last risk-point, so Thursday is eliminated, Wednesday effectively becomes the last risk-point, and the reasoning is repeated.

Suppose instead that the teacher had said that an examination of two hours would start at some unexpected moment within school hours, say 0900 to 1700, on some day within the next week. The examination would have to start by 1500, but there would still be an infinite number of moments in each day at which it could start. (We shall assume that time is infinitely divisible.) Then the students' reasoning would be blocked.

It would be blocked because moments of time are densely ordered. That is, between any two moments in order of time, there is another moment. Between 0901 and 0902, there is 0901 + 30 seconds. Between 0901 and 0901 + 30 seconds, there is 0901 + 15 seconds. However finely we chop up time, even into microseconds, there will always be more moments in between the ones that we have already identified. And each moment would be a risk-point, if the teacher had announced that the examination would start at some unexpected moment.

The dense ordering would be enough to block the students' reasoning. Their reasoning relied on deleting the last risk-point, and moving back to the immediately preceding risk-point. But if the risk-points were densely ordered, there would be no immediately preceding risk-point, to which they could move. Any preceding risk-point they identified would not be the immediate predecessor of the point they had deleted, because there would be another risk-point, later than that one but earlier than the point they had deleted.

Thus a densely ordered stretch of risk-points blocks the chain of reasoning. It is not even necessary for the whole stretch of risk-points to be densely ordered. (Indeed, it is not densely ordered in the case of the examination. There is, for example, no risk-point between the one at 1500 on Monday and the one at 0900 on Tuesday.) All that is necessary is for there to be some densely ordered stretch of risk-points before the last risk-point. If there is a stretch reaching back from that last point that is not densely ordered, the reasoning of the paradox can be used to eliminate all of the risk-points in that stretch, but once we have eliminated the risk-point at the end of the last densely ordered stretch, the reasoning is blocked.

(The last densely ordered stretch might be open at its end, that is, it might not include its end point. It might, for example, be all moments from 0900 on Thursday up to, but not including, 1200 on Thursday. In that case, there would be no final risk-point of the stretch to be eliminated. One could only work back as far as eliminating the risk-point that came first in order of time after the last densely ordered stretch.)

Another feature of the arrangement, which would arise outside densely ordered stretches of risk-points, but which would not arise within any densely ordered stretch, is that there would be an interval between each risk-point (other than the final one) and a risk-point that came later, which was itself empty of risk-points, and within which the students could come to appreciate that only the later risk-points were still possibilities. This matters because such appreciation would be the mechanism, by virtue of which it would be impossible for the students to have the promised ignorance of which risk-point was the one at which the examination would occur. In the interval between the penultimate and the final risk-points, satisfaction of the condition of ignorance would rule out the final risk-point.

My thoughts turned to the paradox because of Greece's likely exit from the Euro. It is essential for currency reforms to be kept secret until they take effect. Otherwise the withdrawal of funds, and speculation, will lead to a bigger crisis than the one that the reform is intended to resolve. Witness, for example, the highly secretive preparations for the introduction of the Deutsche Mark in 1948, first the Konklave von Rothwesten, at which the reform was worked out, and then the delivery of banknotes, in advance of the announcement, in Operation Bird Dog. Moreover, currency reforms take place on specific days, not at specific moments (except, possibly, midnight at the start of some day.) So is there scope for a paradox of the surprise currency reform?

There is one fact that prevents such a paradox from arising, and it is a fact that politicians happily exploit, although perhaps not out of a conscious desire to avoid the trap of this paradox. The relevant fact is that there is no known end-point to the period of risk. Even if one is certain that Greece will leave the Euro, and however bad things get, there is always the possibility that some new fix will tide Greece over for a little bit longer.

Alternatively, people might become certain that the fixes would run out, and that Greece would leave the Euro within a period with an end-point that was already known. But once people became certain of that, the markets would not sit around like students, worrying whether the examination would be today, tomorrow or the next day. They would respond immediately, as if the examination had just started, with only a very limited softening of their reaction in recognition of the fact that the exit might still be delayed by a few months.

The accreditation of qualifications

The market in qualifications is growing, and getting more open. I wonder whether modern trends will have only the expected and desired results, or some unexpected or undesired ones too.

For a long time, we have had universities, school examination boards, and professional institutes. Some of these entities are given special powers by the state, for example the exclusive right to award qualifications that are called degrees, or to confer qualifications that entitle people to represent others in court.

Now, we see a profusion of new bodies that offer qualifications, some for areas of work that did not exist a few decades ago, such as information technology, and some for long-standing trades. More interestingly, there are now some businesses that are detached from those subject-specific bodies, but that offer accreditation services. Googling on the phrase "accreditation services" will turn up several examples.

One expected and desired result of this proliferation of qualifications is that there are now plenty of focused and assessed courses available. This should increase the level of skills of the population, and indeed the general educational level of the population, more effectively than a less-focused exhortation to read books. It is also good if people seek out focused courses in in the humanities, and in the natural and social sciences considered in their own right (rather than in relation to their practical application). And plenty of courses like that are now available, some of them provided free of charge by universities. But the push to undertake such courses is not going to come from people's employment.

Another result is that there is no longer a single clear significance of accreditation. There is no single framework of standards, within which the endorsers of qualifications work. To some extent, it was ever thus. Universities and professional bodies worked separately. But when there are commercial providers of accreditation, we must ask whether they are as strict as they should be, or whether their commercial interests affect their work. Strictness requires more than consistency. One could be consistently generous in granting accreditation, but that would make accreditation valueless.

Many of the accreditors are at pains to assure us of their ethical stances and independence. But self-certification of that nature should carry no weight. Even non-profit bodies cannot be assumed to be immune to untoward influences. There may be no shareholders to seek profits, but there are still employees who seek their continued employment. That requires keeping existing customers and, just to be on the safe side, acquiring new ones.

So where there is a free market in accreditation, the mere fact that a qualification has been accredited proves little or nothing. Anyone who sought to establish the worth of a given accreditation would have to find out which body had conferred it, how that body worked, and in particular, how many qualifications that body refused to accredit. No prestige would automatically attach to qualifications that were accredited in the free market. Having said that, the process of accreditation could still be useful to those who ran qualifications, because it would give them some external comment on their work.

Is this result desirable? The choice is between the following.

1. A single framework of standards. I do not favour this option. It would be cumbersome to ensure compliance across the full range of accreditations of qualifications. Mindless bureaucrats and users of management-speak would devise forms that would be tedious and the completion of which would prove little. And the standards would probably be inappropriate to some accreditations.

2. A free market in accreditation. This is the option we have accidentally chosen, and I think it is better than option 1., although it has disadvantages, as already noted.

3. No accreditation for the great majority of qualifications. I suspect this would be just as good as option 2. It is not clear to me that either the people who take courses, or the people who offer jobs to those who may hold certain qualifications or who send their employees on courses, gain much useful information from accreditation in the free market. It must be up to individual students and employers to decide whether they wish to choose accredited qualifications, when there are non-accredited (and quite possibly cheaper) alternatives. But they should not think that non-accredited qualifications would necessarily be worth any less than accredited alternatives.

Privacy and injunctions

On 27 March 2012, the House of Lords and House of Commons Joint Committee on Privacy and Injunctions published a report, "Privacy and injunctions". The report and the published evidence (published on 16 March 2012) are available from this page:

http://www.parliament.uk/business/committees/committees-a-z/joint-select/privacy-and-superinjunctions/

(My comments here are very critical of the report, so it is worth noting that not every member of the Committee agreed to everything in the report. Some were greater friends of freedom than others. A record of voting on particular parts of the text is given on pages 88 to 103 of the report.)

The Committee really does not understand the importance of free expression. There are warm words about its importance, but the Committee regards a right to privacy as being of equal importance. On page 5, we find:

"The fundamental right to freedom of expression lies at the heart of this debate. The right to privacy is equally important. It is universal and can only be breached if there is a public interest in doing so."

Equality of importance of these rights simply cannot work, if the latter means a right of privacy as against publication in the media, rather than a right to keep one's affairs private from the state (and a right not to be the object of state snooping is arguably that right that Article 8 of the European Convention really confers). Free expression, and privacy as against media publication, are bound to come into conflict, all the time. And free expression must be accorded the higher status, if we are not to become an utterly oppressed people.

One reason to accord the right to privacy the lower status is that if someone does something shameful and it appears in the newspapers, he or she has only himself or herself to blame. Another reason is that those who are most likely to take advantage of a right to privacy are those who already exercise significant influence through their wealth, fame or political position. The reality of such people's personalities and actions jolly well should be open to scrutiny.

The Committee's attitude to Twitter is ambiguous. There are warm words about its role in the Arab Spring in paragraph 92, but then a certain pleasure is taken in the fact that Twitter users can be identified and action can then be taken against them (paragraph 100). The Committee hesitates over the use of country-specific blocking of tweets (paragraph 108), but seems to think that disclosure of the fact of blocking would make blocking acceptable.

The degree of oppression that the Committee calmly contemplates can be seen in a couple of further passages, in paragraphs 109 and 115.

Paragraph 109: "We recommend that, when granting an injunction, courts should be proactive in directing the claimant to serve notice on internet content platforms, such as Twitter and Facebook. Beyond that, claimants in privacy cases should make full use of notice and take-down procedures operated by responsible internet service and social media providers, who should also seek to disseminate best practice and discourage illegality amongst users and other providers."

Extract from paragraph 115: "Google and other search engines should take steps to ensure that their websites are not used as vehicles to breach the law and should actively develop and use such technology. We recommend that if legislation is necessary to require them to do so it should be introduced."

The message of paragraph 109 is that it is responsible to kowtow to the judges who dictate what we can write, and what we can read, and whose reasons we cannot inspect because, by the nature of the case, they have to act in secret. And it would be quite wrong for anyone to stand up and say "It is a very great evil for any arm of the state secretly to regulate expression, so I reject that regulation".

The Committee's belief that we should kowtow is set out on page 5: "The rule of law in protecting the right to privacy should be upheld by all. If a judge has made a decision, based on hearing the full evidence in a case, that decision should be respected by those who have not heard all the evidence". The judge might be better informed than the rest of us about the circumstances of the particular case, and about the relevant precedents. There would be no reason to think that the judge had an ethically respectable position on freedom.

It might be different if the proceedings were public. Then it would be reasonable to say "I disagree, but I can see that the decision is one that one could reasonably reach, and that, plus my respect for the rule of law, is enough to lead me to respect the decision". But it would be an abnegation of one's own rationality meekly to accept a judge's deicision, when one had no idea what the grounds for the decision were. The fact that privacy is the point of the proceedings, so that they have to be secret, makes no difference to this point. If the only way to enforce a right is to ask us to take secret justice on trust, the right should not be enforced.

Paragraph 115 says that search engines, in whatever country they are based, should apply British law to hide information from British users. I hope they ignore this request, even if legislation to put it into effect is enacted. In particular, American search engines have the noble First Amendment to uphold. There is no reason why people or companies in nations with greater freedom of expression should respect the laws which govern that freedom of nations with lesser freedom of expression, even if the freedom is only a little bit less.

Finally, in chapter 5, the Committee proposes a press regulator, the jurisdiction of which would extend to all major publishers of newspapers, whether they liked it or not. That would amount to the requirement to obtain a licence to publish, however it was dressed up as non-statutory or semi-statutory. It is, like much else in the report, an oppressive proposal.

My overall conclusion is that a right to privacy (as against media publication, rather than as against state snooping) can no longer be enforced, save by oppressive means. That means we must give up trying to enforce the right, and rejoice in our right to free expression instead.

Summer time and referenda

It is time for our regular twice-yearly debate over what the UK's time zone (or zones) should be, and over whether the clocks should be changed twice a year.

A piece on NDR Radio this morning explained what has been going on in Russia. Since the move to summer time in March 2011, the clocks have not been changed in Russia, on the orders of President Medvedev. Summer time lasts all year. He was reported as thinking that repeated change disturbed the biorhythms of people and animals, particularly milking cows. (The claim should clearly be restricted to animals that have some interaction with people. Bears deep in the taiga are unlikely to read or hear that the clocks have changed. But many animals could be affected, including wild animals that get used to crossing roads by a certain light, and suddenly find that at the appropriate time by the Sun, the roads are busy with rush-hour commuters instead of empty.) It seems that some political parties want to go back to a twice-yearly change, and that Vladimir Putin is likely to implement this.

The claim that biorhythms are disturbed by a twice-yearly change smacks of pseudo-science, although I acknowledge that the alignment of the light with the clock does strike me as odd for a couple of days after each change. If there is to be sensible argument about this question, it is likely to be about such things as energy consumption, accident rates, the moods of people when their hours of work limit their enjoyment of sunshine, and the needs of businesses to have time zones as closely aligned as possible with those of their trading partners. The latter point came up in connection with Russia. Businesses in Moscow find it awkward to be three hours ahead of Germany, rather than two, in the winter.

A noteworthy feature of debates over time zones and whether to change the clocks twice a year is that the factors to consider are both numerous and disparate. There is no obvious way to weigh them all in a single balance. Only the most convinced utilitarian could think that there was.

Another feature is that one of the factors is how people feel. Psychologists can advise us on how access to daylight influences our moods, but the relevant notion of how people feel is broader than that. It involves sentimental attachment to traditions and to habits of life, and the special joy that some feel at witnessing the dawn and others at witnessing the dusk.

A third feature is that how people feel is central to the process of weighing the factors. Some people think that accident rates are the most important thing, others that energy-saving matters most, and others that business efficiency is of the first importance.

All of this leads me to conclude that the question of how we should set our clocks would be an ideal question to put to a referendum. The relevant features of the question are these.

1. We need a single answer. We do not need to agree on hours of work. Different people can work at different times. But we do need to be able to say "meet me at noon", or "the train leaves at 1832", and all mean the same thing.

2. While there are experts on different aspects of the question, there is no expert on how to combine the conclusions on specific aspects to arrive at an overall answer.

3. If we were to hold a referendum, we could draw on the wisdom of crowds. Lots of non-experts, voting independently, can produce just as good an answer as a committee of experts, and sometimes a better answer. The difficulty would be to get people to vote independently. Clever propaganda, crafted be special interest groups and disseminated through the media, would persuade people to try to sway their friends.

There would be a challenging question of how to divide the territory into areas, the inhabitants of each of which would vote for conclusions that would apply within their areas. For example, should Scotland, England, Wales and Northern Ireland hold separate votes?

On the one hand, it would seem right to divide the territory into the smallest areas that could sensibly have their own time zones, so that local circumstances could be reflected in decisions. On the other hand, one of the considerations that may influence people is the desire to have the same time zone as prevails in neighbouring areas. Someone, every 15 degrees of longitude or so (or 30 degrees with two-hour jumps), must live near a time zone boundary, but each of us would rather it was someone else. That desire would lead people not to take so much account of local circumstances. Differences of latitude can give rise to the same problem. One of the objections to the UK's aligning its time zone with France and Germany (UTC + 1 in the winter and UTC + 2 in the summer) is that the Sun would not rise until very late in Scotland in winter. One solution would be for Scotland to have its own time zone, but life around the border would then get more complicated.

Perhaps the answer is to have two or three rounds of voting. After the first round, people could see how much tendency to difference there was, and could change their votes accordingly in the second round if differences mattered enough to them, and if they saw that the only likely way to remove the difference was for them to change. Just for once, tactical voting could be a good thing.