No coin nor oath required. For personal study only.
The foundations of our science are empirical, i.e., based on direct observation, to the extent where that is possible. For science to move forward and build on other discoveries, it is necessary that any group of experimenters be able to reproduce the results of any other group to within recognizable experimental error. This requirement is so critical that it is virtually an unspoken rule that any scientific paper published must provide sufficient information to permit another researcher to reproduce the published results. Were this practice violated too severely, there would soon be little that is scientific about our science.
It seems unlikely that many would quarrel with this view of empirical science. There do appear to be quite a few, however, who are willing to support a different rule----even a completely different set of rules--with regard to the publication of scientific papers based essentially on computation.
In chemistry we are beginning to accumulate a set of tools and equipment which may well permit us to obtain vast amounts of chemical information by means of computation. This is the same information which only a few years ago could be obtained exclusively by experiment. Thus we now have two nearly parallel approaches to the practice of chemistry: the empirical and the computational. It is quite natural that both groups of researchers be interested in publishing, thereby sharing the results of their efforts with their colleagues.
All would be well were these two approaches to chemistry identical in all aspects. They are not; and therein lies a major problem which could ultimately prove a threat to our science. This problem seems especially acute in chemistry, for reasons not wholly obvious at first glance.
In experimental chemistry, for instance, someone runs a specific reaction using a specific piece of equipment to measure something. (This is obviously simplistic, but it will serve the purpose of the discussion.) The results are then published and subjected to the scrutiny of other chemists. The results are thereby verified or refuted on solid, reproducible grounds. In computational chemistry we are facing an element which could render impossible the simple scenario I have described, one which is strong enough to threaten the process of science which has served us so well. This element is profit--and its power cannot be ignored.
Computational chemistry essentially began in the early 1960s with the availability of large-scale solid-state computers. Computer access was still quite limited, but interested researchers began to experiment with many different concepts, and within a decade people had some fairly dependable ideas on what approaches would work and what sort of directions to take. By the early 1980s, the colossal advances in computer hardware capabilities and the virtual collapse in computing costs put this computing power within reach of any interested researcher--power significant enough to practice computational chemistry. This ready availability of hardware naturally created a vast market for computational chemistry software where one had never really existed before. Chemists have historically been entrepreneurs and, recognizing the opportunity for profit, many of them brought commercial software products to market to meet this need.
It was quickly discovered, however, that computer software is very different, as a property,
๐ SIMILAR VOLUMES