First the link to this week’s complete list as HTML and as PDF.
***
An important study by Andriole et al. on the efficacy of testing for prostate cancer has just been withdrawn. It has turned out that 80 % of the control group, assumed not to have undergone precautionary testing, has been privately tested too. So yes, the original report was partially wrong. But is it invalidated? We have two curves, cases and mortality. There are many ways to diagnose prostate cancer apart from precautionary testing and the number so diagnosed in the control group was about two thirds that in the intervention group (1300 vs. 2000 in year 4). This will include all cases however diagnosed, including those by private testing. The originally published result was, that 50 % more (early) diagnoses do not lead to a measurable number of prevented or delayed deaths. This result stands and is not affected by the new disclosure.
***
Although he makes an important point, I believe Herley slightly overstates the facts when he claims ‘That is, no possible observation contradicts a claim of the form “ if you don’t do X you are not secure.” ’
. You can make an (admittedly weaker) claim by looking at those, who in decades have never yet suffered any malware on their machines. You will probably find, that all of them have consistently ignored received wisdom while as far as I can tell just about everyone religiously following the mantra always to use the newest bloat and to update at the earliest possible opportunity has been hit at least once. It’s just that the adherents of the newer-is-always-better religion are constitutionally unable to see that.