How do scientists carry out science, and how should they? These are questions that have been raised by reviewers in a recent edition of Science magazine, one of the leading science journals. One of the telling quotes is “resources (that) influence the course of science are still more rooted in traditions and intuitions than in evidence.” What does that mean? In my opinion, it is along the lines, for those who have, much will be given. “Much” here refers to much of what is available. Government funding can be tight. And in fairness, those who provide funds want to see something for their efforts, and they are more likely to see something from someone who has produced results consistently in the past. The problem is, the bureaucrats responsible for providing the finds have no idea of the quality of what is produced, so they tend to count scientific papers. This favours the production of fairly ordinary stuff, or even rubbish. Newbies are given a chance, but there is a price: they cannot afford to produce nothing. So what tends to happen is that funds are driven towards something that is difficult to fail, except maybe for some very large projects, like the large hadron collider. The most important thing required is that something is measured, and that something is more or less understandable and acceptable by a scientific journal, for that is a successful result. In some cases, the question, “Why was that measured?” would best be answered, “Because it was easy.” Even the large hadron collider fell into that zone. Scientists wanted to find the Higgs boson, and supersymmetry particles. They found the first, and I suppose when the question of building the collider, the reference (totally not apt) to the “God Particle” did not hurt.
However, while getting research funding for things to be measured is difficult, getting money for analyzing what we know, or for developing theories (other than doing applied mathematics on existing theories), is virtually impossible. I believe this is a problem, and particularly for analyzing what we know. We are in this quite strange position that while in principle we have acquired a huge amount of data, we are not always sure of what we know. To add to our problems, anything found more than twenty years ago is as likely as not to be forgotten.
Theory is thus stagnating. With the exception of cosmic inflation, there have been no new major theories that have taken hold since about 1970. Yet far more scientists have been working during this period than in all of previous history. Of course this may merely be due to the fact that new theories have been proposed, but nobody has accepted them. A quote from Max Planck, who effectively started quantum mechanics may show light on this: “A new scientific truth does not triumph by convincing its opponents and making them see the light, but rather because its opponents eventually die.” Not very encouraging. Another reason may be that it failed to draw attention to itself. No scientist these days can read more than an extremely tiny fraction of what is written, as there are tens of millions of scientific papers in chemistry alone. Computer searching helps, but only for well-defined problems, such as a property of some material. How can you define carefully what you do not know exists?
Further information from this Science article provided some interest. An investigation led to what then non-scientists might consider a highly odd result, namely for scientific papers to be a hit, it was found that usually at least 90 per cent of what is written is well established. Novelty might be prized, but unless well mixed with the familiar, nobody will read it, or even worse, it will not be published. That, perforce, means that in general there will be no extremely novel approach, but rather anything new will be a tweak on what is established. To add to this, a study of “star” scientists who had premature deaths led to an interesting observation: the output of their collaborators fell away, which indicates that only the “star” was contributing much intellectual effort, and probably actively squashing dissenting views, whereas new entrants to the field who were starting to shine tended not to have done much in that field before the “star” died.
A different reviewer noticed that many scientists put in very little effort to cite past discoveries, and when citing literature, the most important is about five years old. There will be exceptions, usually through citing papers by the very famous, but I rather suspect in most cases these are cited more to show the authors in a good light than for any subject illumination. Another reviewer noted that scientists appeared to be narrowly channeled in their research by the need to get recognition, which requires work familiar to the readers, and reviewers, particularly those that review funding applications. The important thing is to keep up an output of “good work”, and that tends to mean only too many go after something that they more or less already now the answer. Yes, new facts are reported, but what do they mean? This, of course, fits in well with Thomas Kuhn’s picture of science, where the new activities are generally puzzles that are to be solved, but not puzzles that will be exceedingly difficult to solve. What all this appears to mean is that science is becoming very good at confirming that which would have been easily guessed, but not so good at coming up with the radically new. Actually, there is worse, but that is for the next post.