Langmuir’s third test was simply that one of the characteristics of bad science is that the investigators make “Claims of great accuracy”. By itself, it seems reasonable to insist that accuracy is a good thing and we want accurate data. And the more accurate we can be, the lower our potential for experimental error.
In the current context, experimental error comes from measurement of either the inputs or the outputs. How can we have errors in the inputs? If you are doing a chemistry experiment, there can be errors in the measurement of the volumes, temperatures, pressures, time, and mass (among others). Likewise, there can be similar errors in measuring outputs. Both of these go into the experimental error – which is one reason we try to randomize experiments.
While we strive to eliminate experimental errors, we can only minimize them. We demand instruments that are both accurate (close to the target) and precise (with very little spread). In other words we want highly consistent and reproducible measurements.
However, when someone claims to have the only gizmo in the world that can measure something with near perfect accuracy, it is time to question. Or when someone claims his mathematical model captures all the nuances that other models seem to miss, some further examination of the models is required.
A good scientist welcomes those challenges once he has protection for his intellectual property. He should have a patent for his gizmo or a copyright for his model. If the scientist is not forthcoming with methods and data, then his observations are not confirmable and his theory is unproven.
Note that I have not talked about intellectual property vis-à-vis actual data. In my view, data is not copyrightable because it is in the public domain, available to anyone who can observe it. Think of the temperature outside your home or office right now. Is it reasonable to expect anyone to claim a copyright on the temperature value? If only one person has the ability to observe, he may certainly sell or license that ability to others. If the scientist expects a monopoly on his data, he should not expect the right to make his results public. His is the question of “public” versus “private” science.
With public science, a scientist will get a grant from some government entity, and as taxpayers, we have a right to see the fruits of his labor that we paid for. I believe he has a duty to make the publically funded data publically available. However, I am not a lawyer, and for all I know my opinion on this matter is completely wrong.
Private science, on the other hand, is paid for by a private entity, such as a corporation. Their duty is to release only what is necessary to support the legal protection of their intellectual property. If, by some miracle, I devised a teleporter, I need only disclose the minimum amount of data required to support a patent and to write that patent as vaguely and broadly as possible so that my patent would be granted. Or I could keep it a trade secret and not tell anyone how I was able to make it work.
But accuracy without confirmation from an independent source is simply an anecdote.