Reproducibility
From SkepticWiki
Contents |
[edit] Introduction
For this reason, along with the result itself, we require him to describe the methodology of the experiment, or the conditions under which the observation was made, in such a way that we could, given the inclination and resources, try to reproduce his observations and see for ourselves that he's right (or, as it might be, wrong).
[edit] Discussion
Most experiments are not usually reproduced exactly, except by the original researcher doing the experiment. There are good reasons for this.
Consider the example of a man who claims that if you shine green light on water, this lowers its boiling point to 90°C. Suppose that we are skeptical of this claim, as well we might be.
The ultimate in reproduction would be to get the same scientist to do the same experiment in the same conditions using the same equipment in the same laboratory. This would certainly allow us to check whether he was simply lying. It would also allow us to check that he was not using his equipment incompetently --- for example, he might not be leaving the thermometer in the water for long enough for it to reach the temperature of the water. Or we might notice that his laboratory was halfway up a mountain, where the boiling point of water is lower for reasons that have nothing to do with green light.
However, if he did reproduce the experiment, we might still conjecture that the thermometer was broken, or the water contaminated with some chemical that lowers its boiling point.
Even a reproduction with the same brands of equipment is not terribly enlightening. If we found that the experiment only worked using a particular brand of electronic thermometer, we might guess that there is something about its workings that's affected by green light. If the same experiment failed using a mercury thermometer, we should certainly be deeply suspicious. In fact, when we find that an experiment requires an exact reproduction of conditions, we should be suspicious of the result. The best-supported results are those that can readily be reproduced by lots of different experimental setups all independently designed to find out the same thing.
[edit] Misconceptions: creationists and reproducibility
It is hard to misunderstand the concept of reproducibility, but creationists are equal to the task. Their blunder may be summarized as follows:
- Reproducibility is an important part of the scientific method. But scientists cannot reproduce the past four billion years of evolution in a laboratory. Therefore, believing that it happened is unscientific.
Now, this has nothing to do with reproducibility as it's defined in science: it is the experiments and observations that need to be reproducible, and in the case of evolution, they certainly are. One can go and look at, for example, the fossil record as many times as one pleases.
But scientists do not claim to have produced four billion years of evolution in their laboratories. If they did, the onus would certainly be on them to reproduce it, but there is no requirement on them to re-produce what they never claimed to have produced.
To fully appreciate the folly of the creationist demand, consider applying the same illogic to other scientific questions. When an astronomer says that Saturn has rings, we don't challenge him to "reproduce" this by constructing another ringed gas giant: his failure to do so has no bearing on the question. What we ask him is how we can reproduce the observations that led him to his conclusion, information that he can and will share.
Or, to take an example from the historical sciences, we do not ask archaeologists to establish an agrarian empire in Egypt, rule over it as god-kings, and get the population to build pyramids before we will admit that it is scientific to believe in the ancient Egyptian empire.
Indeed, just as failure to "reproduce" things in the creationist sense doesn't make true things false, success in such an endeavor wouldn't make false things true. If a man claims to have proven Mallory's tales about King Arthur by sitting a hundred men in armor around a big circular table and exhorting them to search for the Holy Grail, then we can see that this would prove nothing of the kind.
Of course, when creationists wish to pretend that creationism is scientific, they abandon their misconceptions about reproducibility, and do not require that "creation scientists" should reproduce the six days of Genesis.
For more information on this subject, see our main article on Creationists versus the Scientific Method.
