Above link points to a news article. The paper itself is linked here:
Cluster failure: Why fMRI inferences for spatial extent have inflated false-positive rates
I find it amazing that cross-testing and elaborate verification hasn’t been considered a minimum best practise when developing and working with these devices.
I find it upsetting that they still aren’t, even after the salmon thing (Wired, 2009).
Please note that both of these are talking about statistical methods, not software bugs. The journalist appears to be incorrectly assuming that any badness that comes out of a computer is a bug.