Too many cooks spoil the soup

Since the 1990s, much of the debate on research ethics has revolved around fraud and misconduct in relation to falsification, fabrication and plagiarism. In recent years, there has also been a considerable focus on big data, as described in the new report by the National Research Ethics Committees in Norway on big data in research (translation pending).

Fraud and manipulation of data have a long history. In the 1830s, the challenge was to improve the quality of the data collection, which was crucial for the development of science as a profession. It was therefore important to separate the scientific approach from other undesirable practices, such as hoaxing, forging, trimming and cooking. Data collection had to be disciplined.

Bad data

The scientific revolution in the 17th century was closely linked to the experimental method, for instance formulated by Francis Bacon in his philosophical work Novum Organum from 1620. Using new instruments and scientific analysis, it was possible to produce new facts in the laboratory, for example about vacuum and pressure or light and optics.

However, people were not necessarily willing to trust new knowledge about the natural world produced with machines in a laboratory. Another important innovation was therefore the establishing of ‘facts’ as a reliable category in the 17th century.

‘Data’, however, did not come into being until the 19th century. This time, it was not related to controlled experiments in the laboratory, but rather to thousands of measurements in the study of natural phenomena such as ocean currents, the Earth’s magnetic field and meteorology.

The object of study here was fluid and dynamic, such as weather and wind. Many of the observers were not professional scientists, but rather affiliated with the navy, trading companies or expeditions. They also lacked standardised instruments and measurement techniques. Their data were therefore not reliable as a source of knowledge. Consequently, these research fields had a low status within the established universities, where the exact sciences dominated.

The decline of science

Poor data quality was one of the catalysts for an important debate on the decline of science in the UK in the 1830s. The main figure was Charles Babbage, a Cambridge-educated mathematician who was also the editor of The Nautical Almanac. In order to handle large collections of data, he began designing a mechanical calculator – an important forerunner of the modern computer.

However, Babbage struggled to secure financial support for his mechanical calculator. Practical work had low status at universities and in the Royal Society. In the book Reflections on the Decline of Science in England, and Some of its Causes from 1830, Babbage criticised this elitist attitude.

The Royal Society in London had declined, he claimed, and the snobbish members were more concerned with sociability than with science. And many were officers and lords rather than researchers. They had failed to uphold Bacon’s legacy and squandered the scientific revolution of the 17th century, according to Babbage.

In 1831, Babbage was a driving force behind the establishment of the British Association for the Advancement of Science (BAAS). The association quickly became an arena for practically oriented researchers and played an important role in the professionalisation of science in the British Isles. It was in connection with the establishment of this association that a fellow Cambridge scholar, William Whewell, invented the term ‘scientist’, which was first used in 1834.

Babbage’s main argument was that professional natural scientists should also be interested in practical studies benifitting society, such as meteorology, electricity and magnetism. They should also take responsibility for more disciplined data collection. It was not until the middle of the 19th century that these experimental natural sciences were introduced at universities.

What’s cooking?

Babbage’s book was also about research fraud. It was impossible to develop the sciences as long as the observations were so mediocre. He criticised officers and polar heroes who collected measurements from distant shores, and he showed how they manipulated data to concur with their theories. To describe this, he formulated four analogies for various forms of fraud: hoaxing, forging, trimming and cooking.

The most serious of these, hoaxing or falsifying the object of study, was fortunately not so widespread. Here Babbage referred to an honourable knight from Malta who, in 1788, claimed he had discovered a completely new species of animal on the beaches of Sicily. However, the animal was in fact made up of different parts from other marine animals and did not actually exist. All data concerning the animal were therefore misleading.

The other form of fraud was forging data, i.e. fabricating results that were not based on direct observation. Forging was a more permanent form of fraud, because the output data were also included in tables and scientific publications. The damage was therefore greater, and it took a long time to reveal such fraud. Babbage concluded that forging was also not very widespread.

The third form was far more extensive, namely trimming data. This entailed cutting data here and there that did not fit, like when you trim a hedge or cut your hair. This haircut approach to data was apparently not so dangerous, according to Babbage, but it was nevertheless ruinous for the researcher’s integrity and credibility. Precision was therefore vital in both observation and analysis.

The worst type of fraud, however, was cooking. This was not just about omitting some data; here all observations were cooked together in a soup, making it impossible to know what was true and what was just spillage. As long as there was a sufficient colletion of data, the chef could always find results that suited the taste. And if it did not quite fit, it was simply a matter of finding another recipe that included all the selected ingredients. The researcher’s task, however, was to follow the recipe and distinguish the wheat from the chaff.

The boundaries of integrity

In the 1830s, the new role of professional scientist had to be separated from the lords and the officers of the Royal Society. Also, they had to maintain a distance from practical crafts such as blacksmiths, gardeners and cooks.

Data were not reliable in themselves, but had to be cultivated using scientific methods, which also had to form the basis for data collection. The credibility of the sciences required both individual and institutional integrity. And the challenges of fraud and manipulation did not disappear with the advent of the computer.  

References

Babbage, Charles (1830). Reflections on the Decline of Science in England, and Some of its Causes, London, pp. 174–183.

Higgitt, Rebekah (2012). ‘Fraud and the decline of science’. The Guardian 13. September.

Hyman, Anthony (2002). ‘Charles Babbage: Science and reform’, in Perter Harman and Simon Mitton (red.). Cambridge Scientific Minds, Cambridge University Press, pp. 79–93.