Wednesday 5 March 2008

Mega-cash prizes for revolutionary science

Editorial

Stimulating revolutionary science with mega-cash prizes

Bruce G. Charlton (a) and Peter Andras (b)

Medical Hypotheses - Volume 70, Issue 4, 2008, Pages 709-713

(a) Editor-in-Chief – Medical Hypotheses, Newcastle University, NE1 7RU, UK and (b) Editorial Advisory Board – Medical Hypotheses, Newcastle University, NE1 7RU, UK

Available online 5 March 2008.

Summary

We argue that the most ambitious science is intrinsically riskier science, more likely to fail. It is almost always a safer career strategy for the best scientists to seek to extend knowledge more modestly and to build incrementally on existing ideas and methods. Therefore, higher rewards for success are a necessary incentive to encourage top scientists to work on the most important scientific problems, ones where the solution has potentially revolutionary implications. We suggest that mega-cash prizes (measured in tens of millions of dollars) are a suitable reward for those individuals (or institutions) whose work has triggered radically new directions in science.

***

Rewards for successful revolutionary science

Revolutionary science may be distinguished from ‘normal’ science in that revolutionary science aims at generating qualitative advances on established science while normal science aims at incremental progress [1]. While there is a grey borderline, there is a clear distinction between research that aims at transformation of a scientific discipline and that which aims at extension of that discipline [2].

Naturally, since revolutionary science is more ambitious, it is more likely to fail. For every successful instance of revolutionary science there are likely to be manifold examples of failure [3]. By contrast, ‘normal’ incremental science is much more likely to succeed, especially when it is performed by able and well-trained scientists working in well-resourced institutions.

Given that revolutionary science is a high risk endeavour which usually fails; it is likely to thrive only when the incentives rewarding the rare instances of success are greater than for normal science. Therefore we would argue that it is insufficient for successful revolutionary scientists merely to get the usual rewards of prestigious professorships, respect from within the scientific profession, and a modestly high level of reasonably secure income. Something more is needed: lots of money.

The money incentive in science

To compensate for the intrinsically greater risk of failure, successful revolutionary science requires greater rewards than normal science; rewards such as higher prestige, better jobs and/or more money. We suggest that more money is the most promising incentive to encourage revolutionary science, because it is the factor which is most-controllable.

Historically, the main reward for scientific success was high prestige. It was possible for a successful genius to transform a whole science – as Darwin transformed biology and Newton transformed physics. After such success these scientists became public figures so there was an incentive of ‘immortality’ (e.g. Darwin and Newton have appeared on UK banknotes). But now that science is divided into numerous sub-specialties there is less prospect of general public esteem and the rewards of high status are confined to recognition within what may be a relatively small and specific scientific discipline.

Nor is revolutionary science likely to be rewarded with better jobs, except at the very top stratum of eminence such as Nobel laureates. The prospect of rewarding successful revolutionary scientists with better jobs than successful normal scientists is unlikely to happen. Economists usually suggest that the basis of the salary and conditions of a job are the (probable) marginal effects that an individual has on future productivity of an organization [4]. This implies that salaries are not given for past achievement, except when this enhances future prospects for the organization. For example, the top research universities in the USA seem to compete to have Nobel laureates on their faculty as a symbol of their scientific eminence. This signal is apparently effective even when the laureates are semi-retired and unproductive.

But this is due to the extreme rarity of Nobel laureates [5], and at lower levels of eminence there is little advantage for an institution to employ people who were, in the past, successful at revolutionary science – and to employ them at the enhanced rates of pay and privileged conditions of service which are necessary to compensate them for having chosen the high-risk path of revolutionary science.

A scientist of high ability who chooses revolutionary science as a career will probably and on average attain lower outputs, lower numbers of citations, lower-impact publications and less professional prestige than if they put the same efforts and abilities into normal science. (To put in another way, a scientist would probably gain a higher rate of science production by aiming directly at attaining high science production, than by aiming to solve a major problem.)

An able and committed scientist choosing a normal science path would therefore usually be able to out-perform himself as a revolutionary scientist, in terms of immediate and measurable performance measures. A longer term strategy will tend to generate results and produce rewards later. Even if everything goes perfectly and a successful revolutionary scientist ends-up with a Nobel prize, the chances are that this will come near or after the end of their professional career.

But the prospect of a major prize such as a Nobel is surely too unlikely, too remote and too uncertain to be a widespread or realistic motivation in the career paths of young scientists in their twenties or thirties. And it is at this age when scientists must decide whether or not to pursue revolutionary science [6].

Incentives for the best young scientists to pursue revolutionary science

The critical decision to embark on a high-risk strategy of revolutionary science would typically come not much later than a scientist’s late-twenties – just after they have finished their PhD or equivalent research training [6]. Of course, some young scientists will be so idealistic (or so unrealistic) that they ignore the potential rewards or costs of their career choices. Nonetheless, incentives usually make a difference to human behaviour, and if there were more powerful incentives, then presumably more scientists would opt to practice revolutionary science. So, what are the incentives which might affect a young and gifted scientist’s career selection?

Let us imagine an outstanding young scientist who has the potential to be first rate. He or she is faced by a choice between a high risk strategy of trying to make a revolutionary contribution to their branch of science, or else to ‘down-shift’ to a less ambitious but lower-risk path of doing a large volume of high-quality normal science. The best scientists are, in other words, in a position to choose between a small chance of success at the goal of making a first rate qualitative breakthrough, and a much better chance of success as a second rank scientist – perhaps making numerous incremental and quantitative advances in the most prestigious areas of existing science.

As argued above, at this age the small chance of a Nobel prize at (say) 65 years old would strike most objective people as hopelessly unlikely, and vastly remote compared with the much more immediate prospect of generating high-impact, highly-cited, well-funded normal science leading reliably to a job at a prestigious institution within the decade. Our feeling is that the only effective incentive to encourage this young scientist into the high-risk strategy of revolutionary science is the prospect of becoming fabulously rich at a stroke – in other words the prospect of winning a mega-prize for revolutionary science.

The prospect of vast riches in the near-ish future is an important motivation to encourage high achievement in creative areas of popular culture in fields such as music, acting and sports. The competition in these areas is intense, and the percentage chance of success is very low – but ‘the mass market’ rewards success very highly and the reward comes while the person is still young – so the incentive is that much more powerful. Music, acting and sports are among the few fields in which young people can become fabulously rich while still young, and there is therefore no shortage (indeed, a huge surplus) of young people competing in these areas of endeavour while eschewing safer and more secure careers [7].

Our point is not that it is necessarily socially beneficial to have so many young people chasing so few niches in music, sports and acting – rather to emphasize that a large and not-too-distant financial reward seems to be a powerful incentive in creative activities. If it works in the arts, we believe the motivating effect of riches would also work in the sciences. We see no reason why revolutionary science would be an exception.

Why mega-cash prizes for revolutionary science?

Traditional science prizes gain their prestige from their extreme rarity – for example, Nobel prizes are awarded to a maximum of three people per year in each discipline (and this rarity would still apply even if the number of Nobel prizes were expanded, as we have advocated [2] and [5]). So, Nobel prizes are essentially prestige rewards. The financial reward attached to Nobel prizes (although it is fairly generous – over one million US dollars shared among the recipients) is of such little motivational significance that media reports often neglect to mention the sum of money.

But in the early days of Nobel prizes, the large amount of money attached to each award was a major reason for the prestige of the prize. If mega-cash prizes became used as an incentive to promote revolutionary science, then the intrinsic status of the prize would again be dependent mainly on the amount of money (rather than the sheer status of winning the prize) because the intention of using mega-cash prizes is that there be many such prizes without the large numbers diluting the motivational effect of the prize.

Traditional science prizes (such as Nobels) therefore constitute a zero sum game in which the more prizes are awarded the less prestige attaches to each prize – because there is only a fixed amount of status to be shared. But this does not apply to monetary rewards – a prize of one hundred million dollars remains well-worth winning whether it is unique or whether 10 or 100 other people also get a 100 million dollar prize. This is important because there need to be enough prizes for revolutionary science that competitors should feel they have a realistic chance of winning one of these prizes so long as their research succeeds in its aims. This would enable there to be a significant incentive to do revolutionary science across as many scientific disciplines as there are mega-cash prizes.

Our feeling is that mega-prizes to stimulate revolutionary science would need to be of the order of magnitude of tens of millions of US dollars in order to replicate the kind of incentives seen in popular creative activities such as music and sports.

Potential problems with mega-cash prizes

There are (at least) two major problems in using mega-prizes to stimulate revolutionary science. However both problems are potentially soluble.

The first problem is that the prizes must have a basis in objective fact, to guard against the prizes being awarded for subjective reasons (for example for political reasons – as has arguably been the case for the MacArthur ‘genius’ fellowships (www.macfound.org) – where significant numbers of recipients seem to have been chosen for their symbolic value rather than their measurable achievement).

To guard against this possibility, our proposal is that mega-prizes should be awarded only to those who have objective evidence of having performed successful revolutionary science. This could be accomplished by specifying objective scientometric criteria which must be satisfied before a person or group can be considered eligible for a prize.

A second difficulty is that mega-prizes must be awarded early enough in a person’s lifespan to act as an incentive to aspiring scientists who are planning their career strategy in their mid-twenties. We think this implies that these prizes should usually be awarded before the age of fifty.

The paradox is that revolutionary science prizes are intended to promote ambitious long term strategic research. Yet to be an effective incentive the prize must be awarded in the medium term (and not the long term). In other words, a compromise is necessary.

In practice, a mega-cash prize before fifty would encourage the brightest 25–35 year old scientists to adopt a time horizon of about ten years for research coming to fruition – on the assumption that a decade-level of long-termism is worthwhile in a context of normal science which typically rewards achievement in units of more like 3–5 years.

A prize awarded before fifty would allow post-doctoral scientists to work on a problem for about a decade, then would allow approximately another decade for the importance of this work to become apparent and for a revolutionary breakthrough measurably to influence the research practices of other scientists.

Scientometrics as a basis for revolutionary science prizes

To be an effective incentive, a prize needs to be relatively transparent and objective, so that the aspiring scientist can be reasonably confident that if they do in fact achieve success in revolutionary science, then this will probably be noticed and they will in likelihood be considered for a prize.

The necessary level of transparency and objectivity can best be achieved by having a first stage selection process based on objective scientometric criteria. This, in turn, requires the development of scientometric instruments for measuring revolutionary science [2] and [8]. Such scientometric instruments are not currently available, and the development and testing of scientometric methods for measuring revolutionary science would – we hope – be a very valuable by product or spin-off benefit of using mega-prizes to reward revolutionary science.

A proper consideration of the topic of scientometric methods for detecting and measuring revolutionary science requires separate treatment, but for the present is may suffice to say that the basis of objective recognition of successful revolutionary science is that the success of revolutionary science can be defined by the fact that revolutionary science affects the direction of normal science. In other words, what makes revolutionary science a success is that is changes normal science then becomes normal science.

In terms of scientometrics, this means that revolutionary science can (in principle) be identified retrospectively from the fact that a small set of revolutionary science communications generate a much larger set of normal science communications which are referenced-back to that originating revolutionary science. In a nutshell, successful revolutionary science can be defined as the work that leads to the evolution of a new scientific specialty.

A new scientific specialty is typically either a sub-specialty of a previous science or a hybrid science. A sub-specialty example is when new branch of medical science develops around a class of disease – e.g. lung disease or liver disease; further new sub-specialty sciences may then arise from sub-classes of these disease classes – e.g. asthma sub-specializing from lung diseases, or hepatitis from liver diseases; and then the process may continue. An example of a hybrid science would be biochemistry – when a new specialty was created by the simultaneous application of theories, techniques and other forms of knowledge from both sub-sets of biology and chemistry knowledge.

This means that some revolutionary science (although not all of it) should be identifiable from a sophisticated analysis of citations within the scientific literature. For example, there may be an observation that over a specified period of time (between, say, 5–10 years) a new cluster of inter-referenced science communications has emerged. This could be detected by the emergence of a new set of specialized conferences, journals and papers – detected using specific terms and methods. More sophisticated methods could include the analysis of emerging citation clusters representing new foci of scientific activity [e.g. [9] and [10]].

The main difficulty is that not all scientific communications are accessible to analysis. For example, as well as formal scientific papers and other public communications there are many un-recorded informal interactions between scientists occurring verbally and by ephemeral or inaccessible media such as notes, letters and private e-mails – not to mention the thoughts inside each scientists head! So scientometrics will only sample from a sub-set of science communications, and it is inevitable that some examples of revolutionary science will remained undetected or under-estimated.

Nonetheless, given the advantages of objective quantification, we would advocate that the field of possible recipients of revolutionary science mega prizes should initially be defined by scientometric methods. After this potential field has been defined, then expert peer review methods could be used to determine whether the identified potential examples of successful revolutionary science were primarily the work of a small number of individual scientists (potential prize recipients); or alternatively the work of teams within a small number of research institutions (in which case the prize could be awarded to the institutions).

Prizes instead of program grants

In conclusion, we suggest that revolutionary science could be encouraged by increasing the monetary incentives for successful revolutionary science – especially the incentives as they operate on the best young scientists as they choose their career paths in their mid twenties to early thirties.

This could be accomplished by a change in behaviour of the large grant awarding bodies – a shift from funding research programs with grants and towards rewarding successful revolutionary science with prizes. For example, a research foundation working in a specific scientific field might at present spend 100 million dollars per year – and might spread this money among ten 10 million dollar program grants. In all likelihood, this money will at present be spent on normal science, and will produce modest incremental progress.

We are suggesting that such a research foundation might instead spend 100 million dollars in a single prize, awarded to a relatively young scientist or a few scientists in recognition of a significant success in revolutionary science.

In the short term, this kind of prize would serve merely as a retrospective recognition of research which had been done anyway – but after a few years the mega-cash prize would begin to work as a prospective incentive; shaping the behaviour of young scientists towards more ambitious scientific problems which (if successfully solved) would be eligible for such prizes.

There is a previous literature on the use of prizes to stimulate scientific research [11], [12], [13] and [14] – however, these types of prizes have either implicitly or explicitly been orientated towards problem solving as quickly as possible and therefore using the simplest possible methods – since this ‘research and development’ approach is most likely to win the prize.

What is novel about our argument here, is that we are suggesting that prizes may also be set-up such that they encourage revolutionary science. Furthermore, we advocate the use of scientometrics as a screening mechanism before peer review as a method of preventing corruption and ensuring that the research being rewarded has had an objectively verifiable consequence of revolutionizing (i.e. changing the direction of, or opening-up new fields for) the practice of science.

In other words, mega-cash prizes might encourage some of the very best young scientists to make more long-term and high risk career choices. The real winner of this would be society as a whole; since normal science can successfully be done by second rate scientists – but if the first rate scientists do not make the decision to tackle the toughest scientific problems, then solutions to these tough problems may be delayed, or they may never be solved.

References

[1] T.S. Kuhn, The structure of scientific revolutions, Chicago University Press, Chicago (1970).

[2] B.G. Charlton and P. Andras, Evaluating universities using simple scientometric research output metrics, Sci Publ Policy 34 (2007), pp. 555–563.


[3] D.L. Hull, Science as a process, Chicago University Press, Chicago (1988).

[4] Caplan B. George Mason University – Econ 103: Labour markets and labour regulation. [accessed 02.01.08].

[5] B.G. Charlton, Why there should be more science Nobel prizes and laureates, Med Hypotheses 68 (2007), pp. 471–473.

[6] F.L. Holmes, Investigative pathways, Yale University Press, New Haven (USA) (2004).

[7] R.H. Frank and P. Cook, The winner take all society, The Free Press, New York (1995).

[8] Charlton B, Andras P. The ‘down-shifting’ of UK science? – The decline of ‘revolutionary science’ and the rise of ‘normal science’ in the UK compared with the USA. Med Hypotheses, in press. doi:10.1016/j.mehy.2007.12.004.

[9] A.F.J. Van Raan, Reference-based publication networks with episodic memories, Scientometrics 6393 (2005), pp. 549–566.

[10] Van Raan AFJ. Scaling rules in the science system: influence of field-specific citation characteristics on the impact of research groups. ; 2007.

[11] D.F. Horrobin, Glittering prizes for research support, Nature 324 (1986), p. 221. Full Text via CrossRef

[12] Davis LN. Should we consider alternative incentives for basic research? Patents vs. prizes. Presented at DRUID summer conference 2002. ; 2002 [accessed 02.01.08].

[13] B.G. Charlton, Mega-prizes in medicine: big cash awards may stimulate useful and rapid therapeutic innovation, Medi Hypotheses 68 (2006), pp. 1–3.

[14] J.E. Stiglitz, Scrooge and intellectual property rights: a medical prize fund could improve the financing of drug innovations, BMJ 333 (2006), pp. 1279–1280.