|
|||
Black box testing is different. A black box tester deliberately isn't informed of details of how the unit works, though he may be able to deduce a great deal of it if he's sharp. He shoots everywhere. Of the two, white box testing will turn up more bugs, but black box testing will turn up more surprisng and generally more serious bugs. That's because the white box tester is to some extent constrained by his knowledge. By shooting at the welds, he misses weak spots in the armor. If you only had to choose between them and only have one kind, you'd definitely choose black box. Of course, that's not really a choice we generally face, and we usually use both kinds. But in some cases black box testing is all that's possible. There are now and have always been a lot of areas in technology where art has developed strictly empirically, with theory lagging badly behind or sometimes not being available at all. The electric generator and electric motor were developed before Maxwell explained why they worked, for instance. A modern example of that is drugs intended for treatment of mental disorders. While it's unquestionably the case that many of them work and even work well, in most cases no-one has the slightest idea how. For instance, most antidepressants work by boosting the levels of the three monoamine neurotransmitters (serotonin, norepinephrine and dopamine). But this can't be the whole answer, because some of these drugs take as long as six weeks to have clinical effect, whereas their effect on neurotransmitter levels happens immediately. And given that no-one has any idea just what depression actually is, it's hard to say how these drugs affect it. "Depression" is a description of a clinical condition, not a description of a failure mode. It's like describing diabetes as a failure to control sugar levels, not as a failure of the pancreas to correctly secrete insulin in proper quantities. As a result, Drugs to treat depression have been developed through serendipity or by varying other drugs which were known to work. Another example of a strictly empirical science is epidemiology. Without knowing what the disease pathogen was, it was possible to determine in the early part of the 19th century that cholera was passed through contaminated drinking water. Without having any idea at all what caused AIDS, it was possible to determine that it was passing through body fluids such as blood and sexual secretions. Only later were the comma baccilus and HIV identified respectively as the disease pathogens, which then explained the empirical result. But the empirical result stands solidly even without a theoretical explanation, and if there is a conflict between the two then usually the theory is wrong. Empirical studies proved that HIV didn't pass via casual contact long before HIV was found; and it remains true today that HIV isn't passed through casual contact. Study of HIV hasn't changed that fact, because our knowledge of that fact wasn't derived from study of the virus. It isn't always necessary to explain why something is the case in order to prove that it is the case. A valid observation is a fact irrespective of whether there's an explanation of the fact. It isn't necessary to explain the genetics of a genetically modified food in order to prove that it's safe to eat, and in fact safety testing doesn't rely on genetic explanations. Which is a good thing, because the science of genetics is much too immature to permit a reasonable explanation. Rather, safety testing uses the time-honored approach of feeding the stuff to animals and observing whether they prosper or get ill. If you feed enough of it to enough different animals with no sign of ill effects, there's a sufficiently high probability that it's safe. That, in fact, is the same way that drugs are initially tested before they are used in humans. We can't look at the chemical formulation of a drug to determine if it will be harmful because we don't have nearly enough knowledge of animal biochemistry to determine all the ways that it might affect things. The only way to find out is to try it and see, which is why animal testing is essential. Once that's done, you try it in people. Greenpeace, in its neverending shrill attempts to throw every possible roadblock into the path of development of genetically modified foods, had demanded that Monsanto explain a certain gene sequence found in a particular approved soybean variant, and then says "If Monsanto did not even get this most basic information right, what should we then think about the validity of all their safety tests and experiments, which are based upon these data?" Which is to say that they're claiming that Monsanto is basing its claims of safety on "white box" genetic analysis. That's not correct. Monsanto is basing its claims of safety on "black box" studies where these soybeans were fed to animals who were not harmed. That's empirical data which doesn't require an explanation. And, in fact, that's how we've alway 
| |||