Friday, August 29, 2008

Why less is more in government
[This is the pre-edited version of my latest Muse for Nature’s online news.]

In committees and organizations, work expands to fill the time available while growth brings inefficiency. It’s worth trying to figure out why.

Arguments about the admission of new member states to the European Union have become highly charged since Russia sent tanks into Georgia, which harbours EU aspirations. But there may be another reason to view these wannabe nations cautiously, according to two recent preprints [1,2]. It claims that decision-making bodies may not be able to exceed about 20 members without detriment to their efficiency.

Already the EU, as well as its executive branch the European Commission, has 27 members, well in excess of the putative inefficiency threshold. And negotiations in Brussels have become notorious for their bureaucratic wrangling and inertia. The Treaty of Lisbon, which proposes various reforms in an attempt to streamline the EU’s workings, implicitly recognizes the overcrowding problem by proposing a reduction in the number of Commissioners to 18. But as if to prove the point, Ireland rejected it in June.

It’s not hard to pinpoint the problem with large committees. The bigger the group, the more factious it is liable to be, and it gets ever harder to reach a consensus. This has doubtless been recognized since time immemorial, but it was first stated explicitly in the 1950s by the British historian C. Northcote Parkinson. He pointed out how the executive governing bodies in Britain since the Middle Ages, called cabinets since the early seventeenth century, tended always to expand in inverse proportion to their ability to get anything done.

Parkinson showed that British councils and cabinets since 1257 seemed to go through a natural ‘life cycle’: they grew until they exceeded a membership of about 20, at which point they were replaced by a new body that eventually suffered the same fate. Parkinson proposed that this threshold be called the ‘coefficient of inefficiency’.

Stefan Thurner and colleagues at the Medical University of Vienna have attempted to put Parkinson’s anecdotal observations on a solid theoretical footing [1,2]. Cabinets are now a feature of governments worldwide, and Thurner and colleagues find that most of those from 197 countries have between 13 and 20 members. What’s more, the bigger the cabinet, the less well it seems to govern the country, as measured for example by an index called the Human Development Indicator, used by the United Nations Development Programme and which takes into account such factors as life expectancy, literacy and gross domestic product.

Thurner and colleagues have tried to understand where this critical mass of 20 comes from by using a mathematical model of decision-making in small groups [1]. They assume that each member may influence the decisions of a certain number of others, so that they form a complex social network. Each adopts the majority opinion of those to whom they are connected provided that this majority exceeds a certain threshold.

For a range of model parameters, a consensus is always possible for less than 10 members – with the exception of 8. Above this number, consensus becomes progressively harder to achieve. And the number of ways a ‘dissensus’ may arise expands significantly beyond about 19-21, in line with Parkinson’s observations.

Why are eight-member cabinets anomalous? This looks like a mere numerical quirk of the model chosen, but it’s curious that no eightfold cabinets appeared in the authors’ global survey. Historically, only one such cabinet seems to have been identified: the Committee of State of the British king Charles I, whose Parliament rebelled and eventually executed him.

Now the Austrian researchers have extended their analysis of Parkinson’s ideas to the one for which he is best known: Parkinson’s Law, which states that work expands to fill the time available [2]. This provided the title of the 1957 book in which Parkinson’s essays on governance and efficiency were collected.

Parkinson regarded his Law as a corollary of the inevitable expansion of bureaucracies. Drawing on his experience as a British civil servant, he pointed out that officials aim to expand their own mini-empires by gathering a cohort of subordinates. But these simply make work for each other, dwelling over minutiae that a person lacking such underlings would have sensibly prioritized and abbreviated. Dare I point out that Nature’s editorial staff numbered about 13 when I joined 20 years ago, and now numbers something like 33 – yet the editors are no less overworked now than we were then, even though the journal is basically the same size.

Parkinson’s explanation for this effect focused on the issue of promotion, which is in effect what happens to someone who acquires subordinates. His solution to the curse of Parkinson’s Law and the formation of over-sized, inefficient organizations is to engineer a suitable retirement strategy such that promotion remains feasible for all.

With promotion, he suggested, individuals progress from responsibility to distinction, dignity and wisdom (although finally succumbing to obstruction). Without it, the progression is instead from frustration to jealousy to resignation and oblivion, with a steady decrease in efficiency. This has become known as the ‘Prince Charles Syndrome’, after the British septuagenarian monarch-in-waiting who seems increasingly desperate to find a meaningful role in public life.

Thurner and colleagues have couched these ideas in mathematical terms by modelling organizations as a throughflow of staff, and they find that as long as promotion prospects can be sufficiently maintained, exponential growth can be avoided. This means adjusting the retirement age accordingly. With the right choice (which Parkinson called the ‘pension point’), the efficiency of all members can be maximized.

Of course, precise numbers in this sort of modelling should be taken with a pinch of salt. And even when they seem to generate the right qualitative trends, it doesn’t necessarily follow that they do so for the right reasons. Yet correlations like those spotted by Parkinson, and now fleshed out by Thurner and colleagues, do seem to be telling us that there are natural laws of social organization that we ignore at our peril. The secretary-general of NATO has just made positive noises about Georgia’s wish for membership. This may or may not be politically expedient; but with NATO membership currently at a bloated 26, he had better at least recognize what the consequences might be for the organization’s ability to function.


1. Klimek, P. et al. Preprint
2. Klimek, P. et al. Preprint

Friday, August 08, 2008

Crime and punishment in the lab
[This is the uncut version of my latest Muse article for Nature’s online news.]

Before we ask whether scientific conduct is dealt with harshly enough, we need to be clear about what punishment is meant to achieve.

Is science too soft on its miscreants? That could be read as the implication of a study published in Science, which shows that 43 percent of a small sample of scientists found guilty of misconduct remained employed subsequently in academia, and half of them continued to turn out a paper a year [1].

Scientists have been doing a lot of hand-wringing recently about misconduct in their ranks. A commentary in Nature [2] proposed that many such incidents go unreported, and suggested ways to improve that woeful state of affairs, such as adopting a ‘zero-tolerance culture’. This prompted several respondents to maintain that matters are even worse, for example because junior researchers see senior colleagues benefiting from ‘calculated, cautious dishonesty’ or because some countries lack regulatory bodies to police ethical breaches [3-5].

All this dismay is justified to the extent that misconduct potentially tarnishes the whole community, damaging the credibility of science in the eyes of the public. Whether the integrity of the scientific literature suffers seriously is less clear Рthe more important the false claim, the more likely it is to be uncovered quickly as others scrutinize the results or fail to reproduce them. This has been the case, for example, with the high-profile scandals and controversies over the work of Jan Hendrik Sch̦n in nanotechnology, Hwang Woo-suk in cloning and Rusi Taleyarkhan in bench-top nuclear fusion.

But the discussion needs to move beyond these expressions of stern disapproval. For one thing, it isn’t clear what ‘zero tolerance’ should mean when misconduct is such a grey area. Everyone can agree that fabrication of data is beyond the pale; but as a study three years ago revealed [6], huge numbers of scientists routinely engage in practices that are questionable without being blatantly improper: using another’s ideas without credit, say, or overlooking others’ use of flawed data. Papers that inflate their apparent novelty by failing to acknowledge the extent of previous research are tiresomely common.

And it is remarkable how many austere calls for penalizing scientific misconduct omit any indication of what such penalties are meant to achieve. Such a situation is inconceivable in conventional criminology. Although there is no consensus on the objectives of a penal system – the relative weights that should be accorded to punishment, public protection, deterrence and rehabilitation – these are at least universally recognized as the components of the debate. In comparison, discussions of scientific misconduct seem all too often to stop at the primitive notion that it is a bad thing.

For example, the US Office of Research Integrity (ORI) provides ample explanation of its commendable procedures for handling allegations of misconduct, while the Office of Science and Technology Policy outlines the responsibilities of federal agencies and research institutions to conduct their own investigations. But where is the discussion of desired outcomes, beyond establishing the facts in a fair, efficient and transparent way?

This is why Redman and Merz’s study is useful. As they say, ‘little is known about the consequences of being found guilty of misconduct’. The common presumption, they say, is that such a verdict effectively spells the end of the perpetrator’s career.

Their conclusions, based on studies of 43 individuals deemed guilty by the ORI between 1994 and 2001, reveal a quite different picture. Of the 28 scientists Redman and Merz could trace, 10 were still working in academic positions. Those who agreed to be interviewed – just 7 of the 28 – were publishing an average 1.3 papers a year, while 19 of the 37 for which publication data were available published at least a paper a year.

Is this good or bad? Redman and Merz feel that the opportunity for redemption is important, not just from a liberal but also a pragmatic perspective. ‘The fact that some of these people retain useful scientific careers is sensible, given that they are trained as scientists’, says Merz. ‘They just slipped up in some fundamental way, and many can rebuild a scientific career or at least use the skills they developed as scientists.’ Besides, he adds, everyone they spoke to ‘paid a substantial price’. All reported financial and personal hardships, and some became physically ill.

But on another level, says Merz, these data ‘could be seen as undermining the deterrent effect of the perception that punishment is banishment, from academia, at least.’ Does the punishment fit the crime?

The scientific community has so far lacked much enthusiasm for confronting these questions – perhaps because misconduct, while a trait found in all fields of human activity, is felt to be uniquely embarrassing to an enterprise that considers itself in pursuit of objective truths. But the time has surely come to face the issue, ideally with more data to hand. In formulating civic penal policy, for example, one would like to know how the severity of sentencing affects crime rates (which might indicate the effectiveness of deterrence), and how different prison regimes (punitive versus educative, say) influence recidivism. And one needs to have a view on whether sanctions such as imprisonment are primarily for the sake of public protection or to mete out punishment.

The same sorts of considerations apply with scientific misconduct, because the result otherwise has a dangerously ad hoc flavour. Just a week ago, the South Korean national committee on bioethics rejected an application by Hwang Woo-suk to resume research on stem cells. Why? Because ‘he engaged in unethical and wrongful acts in the past’, according to one source. But that’s not a reason, it is simply a statement of fact. Does the committee fear that Hwang would do it again (despite the intense scrutiny that would be given to his every move)? Do they think he hasn’t been sufficiently punished yet? Or perhaps that approval would have raised doubts about the rigour of the country’s bioethics procedures? Each of these reasons might be defensible – but there’s no telling which, if any, applies.

One reason why it matters is that by all accounts Hwang is an extremely capable scientist. If he and others like him are to be excluded from making further contributions to their fields because of past transgressions, we need to be clear about why that is being done. We need a rational debate on the motivations and objectives of a scientific penal code.


1. Redman, B. K. & Merz, J. F., Science 321, 775 (2008).
2. Titus, S. L. et al., Nature 453, 980-982 (2008).
3. Bosch, X. Nature 454, 574 (2008).
4. Feder, N. & Stewart, W. W. Nature 454, 574 (2008).
5. Nussenzveig, P. A. & Funchal, Z. Nature 454, 574 (2008).
6. Martinson, B. C. et al., Nature 435, 737-738 (2008).

Tuesday, August 05, 2008

Who is Karl Neder?

These people tend to define themselves by what they don’t like, which is usually much the same: relativity, the Big Bang. Einstein. Especially Einstein, poor fellow.

In my novel The Sun and Moon Corrupted, where these words appear, I sought to convey the fact that the group of individuals who scientists would call cranks, and who submit their ideas with tenacious insistence and persistence to journals such as Nature, have remarkably similar characteristics and obsessions. They tend to express themselves in much the same manner, exemplified in my book by the letters of the fictional Hungarian physicist Karl Neder. And their egocentricity knows no bounds.

I realised that, if I was right in this characterization, it would not be long at all before some of these people became convinced that Karl Neder is based on them. (The fact is that he is indeed loosely based on a real person, but there are reasons why I can be very confident that this person will never identify the fact.)

And so it comes to pass. The first person to cry ‘It’s me!’ seems to be one Pentcho Valev . I do not know who Valev is, but it seems I once (more than once?) had the task of rejecting a paper he submitted to Nature. I remember more than you might imagine about the decisions I made while an editor at Nature, and by no means always because the memory is pleasant. But I fear that Valev rings no bells at all. Nonetheless, says Valev, there are “Too many coincidences: Bulgaria + thermodynamics + Einstein + desperately trying to publish (in Nature) + Phillip [sic] Ball is Nature’s editor at that time and mercilessly rejects all my papers. Yes most probably I am at least part of this Karl Neder. Bravo Phillip Ball! Some may say it is unethical for you to make money by describing the plight of your victims but don't believe them: there is nothing unethical in Einstein zombie world.” (If it is any consolation, Mr Valev, the notion that this book has brought me "fortune" provokes hollow laughter.)

Ah, but this is all so unnervingly close to the terms in which Karl Neder expresses himself (which mimic those of his real-life model). In fact, Valev seems first to have identified ‘his’ voice from a quote from the book in a review in the Telegraph:
‘Actually, what [Neder] says is: "PERPETUUM MOBILE IS CONSTRUCTED BY ME!!!!!!!!!"; his voluminous correspondence being littered with blood-curdling Igorisms of this sort.’
Even I would not have dreamt up the scenario in which Mr Valev is apparently saying to himself “Blood-curdling Igorisms? But that’s exactly like me, damn it!” (Or rather, “LIKE ME!!!!!!!!!”)

Valev continues: “If Philip Ball as Nature’s editor had not fought so successfully against crazy Eastern Europe anti-relativists, those cranks could have turned gold into silver and so the very foundation of Western culture would have been destroyed” – and he quotes from a piece I wrote in which I mentioned how relativistic effects in the electron orbitals of gold atoms are responsible for its reddish tint. This is where I start to wonder if it is all some delicious hoax by the wicked Henry Gee or one of the people who read my book for the Royal Institution book club, and therefore knows that indeed it plunges headlong into alchemy and metallic transmutation in its final chapters. What are you trying to do, turn me paranoid?

Saturday, August 02, 2008

Might religion be good for your health?
[Here is the uncut version of my latest Muse for Nature news online.]

Religion is not a disease, a new study claims, but a protection against it.

Science and religion, anyone? Oh come now, don’t tell me you’re bored with the subject already. Before you answer that, let me explain that a paper in the Proceedings of the Royal Society B [1] has a new perspective on offer.

Well, perhaps not new. In fact it is far older than the authors, Corey Fincher and Randy Thornhill of the University of New Mexico, acknowledge. Their treatment of religion as a social phenomenon harks back to classic works by two of sociology’s founding fathers, Emile Durkheim and Max Weber, who, around the start of the twentieth century, offered explanations of how religions around the world have shaped and been shaped by the societies in which they are embedded.

That this approach has fallen out of fashion tells us more about our times than about its validity. The increasing focus on individualism in the Western world since Durkheim wrote that “God is society, writ large” is reflected in the current enthusiasm for what has been dubbed neurotheology: attempts to locate religious experience in brain activity and genetic predispositions for certain mental states. Such studies might ultimately tell us why some folks go to church and other don’t, but they can say rather little about how a predisposition towards religiosity crystallizes into a relatively small number of institutionalized religions - why, say, the 'religiously inclined' don't simply each have a personal religion.

Similarly, the militant atheists who gnash their teeth at the sheer irrationality and arbitrariness of religious belief will be doomed forever to do so unless they accept Durkheim’s point that, rather than being some pernicious mental virus propagating through cultures, religion has social capital and thus possible adaptive value [2]. Durkheim argued that it once was, and still is in many cultures, the cement of society that maintains order. This cohesive function is as evident today in much of American society as it is in Tehran or Warsaw.

But of course there is a flipside to that. Within Durkheim’s definition of a religion as ‘a unified set of beliefs and practices which unite in one single moral community all those who adhere to them’ is a potential antagonism towards those outside that community – a potential that has become, largely unanticipated, the real spectre haunting the modern world.

It is in a sense the source of this tension that forms the central question of Fincher and Thornhill’s paper. Whereas Weber looked at the different social structures that different religions tended to promote, and Durkheim focused on ‘secular utility’ such as the benefits of social cohesion, Fincher and Thornhill propose a specific reason why religions create a propensity to exclude outsiders. In their view, the development of a religion is a strategy for avoiding disease.

The more a society disperses and mixes with other groups, the more it risks contracting new diseases. ‘There is ample evidence’, the authors say, ‘that the psychology of xenophobia and ethnocentrism is importantly related to avoidance and management of infectious disease.’

Fincher and Thornhill have previously shown that global patterns of social collectivism [3] and of language diversity [4] correlate with the diversity of infectious disease in a manner consistent with avoidance strategies: strangers can be bad for your health. Now they have found that religious diversity is also greater in parts of the world where the risk of catching something nasty from those outside your group (who are likely to have different immunity patterns) is higher.

It’s an intriguing observation. But as with all correlation studies, cause and effect are hard to untangle. Fincher and Thornhill offer the notion that new religions are actively generated as societal markers that inhibit inter-group interactions. One could equally argue, however, that a tendency to avoid contacts with other social groups prevents the spread of some cultural traits at the expense of others, and so merely preserves an intrinsic diversity.

This, indeed, is the basis of some theoretical models for how cultural exchange and transmission occurs [5]. Where opportunities for interaction are fewer, there is more likelihood that several ‘island cultures’ will coexist rather than being consumed by a dominant one.

And the theory of Fincher and Thornhill tells us nothing about religion per se, beyond its simple function as a way of discriminating those ‘like you’ from those who aren’t. It might as well be any other societal trait, such as style of pottery or family names. In fact, compared with such indicators, religion is a fantastically baroque and socially costly means of separating friend from foe. As recent ethnic conflicts in African nations have shown, humans are remarkably and fatefully adept at identifying the smallest signs of difference.

What we have here, then, is very far from a theory of how and why religions arise and spread. The main value of the work may instead reside in the suggestion that there are ‘hidden’ biological influences on the dynamics of cultural diversification. It is also, however, a timely reminder that religion is not so much a personal belief (deluded or virtuous, according to taste) as, in Durkheim’s words, a ‘social fact’.


1. Fincher, C. L. & Thornhill, R. Proc. R. Soc. B doi:10.1098/rspb.2008.0688.
2. Wilson, D. S. Darwin’s Cathedral: Evolution, Religion, and the Nature of Society (University of Chicago Press, 2002).
3. Fincher, C. L. et al., Proc. R. Soc. B 275, 1279-1285 (2008).
4. Fincher, C. L. & Thornhill, R. Oikos doi:10.1111/j.0030-1299.2008.16684.x.
5. Axelrod, R. J. Conflict Resolution 41, 203-226 (1997).