[By the way, Stevan changed my Subject line - but I suppose it's a relevant followup]
Stevan Harnad wrote:
I would say there is no particular lesson to be learnt from such cases, precisely because they are rare, and no one cares.
How do we know how rare they are? The problem I see with this, and with the Bogdanov's, is that they are rare cases of (very likely) incorrect research articles which are out there sitting with essentially no comment; just a "part of the literature", BUT THEN somebody notices them and makes a noise. How many similar cases are there which are NEVER noticed? What measure is there of "likely incorrect results that have received no public commentary"? With peer review there is some assurance that at least one or two experts have read the article and found it unobjectionable - an imperfect filter, but a filter none the less.
Qualified experts discount quackery; others know that they should not trust results until they are peer-reviewed, except if they are from known experts, and even that, only tentatively.
Do they? Scientists may have an intuitive grasp of this - but Swedish reporters perhaps not? Is there a line that should be drawn, when we make information available to all the public? Do any of the e-print archives out there actually require that something be published in a refereed journal first? If so how is that verified?
The bottom line is that you cannot build on a fraudulent or quackish or otherwise erroneous finding: It soon collapses under its own weight.
Not necessarily - in this particular case mathematicians have already been building huge edifices on the assumption that the Riemann hypothesis is true; any "collapse" would actually be a proof of its falsehood, and also welcomed (though with a smaller prize). Can there not be a more obscure case of some minor result that is published, never properly checked, and somehow absorbed into the mythology of a field? So that hundreds of scientist-years of effort are spent that rely in some small part on it, and then all called into question at a later date? There seem to be a few cases of this in bio-medicine in recent years - we have to worry about this anyway. But it seems there's much more of a danger of this sort of thing when the old checks on author self-publication are removed... I realize it's not a new issue, but perhaps this new example can clarify some of our thinking on this. Arthur