How testing became a discipline of its own — and what it shares with measurement everywhere else.
Solar cells, tumours, corn chlorophyll and a 14 373-citation paper on cosmological portals: the first query of the neuroscience assessment, what went wrong with the sampling frame, and what the index showed unprompted.
PET fell from 124 436 hits to 21 942 after we added a field filter. A clean result — except that we had also changed the search string. A small, real case of an analysis that cannot say what it seems to say, and how it got caught.
First technical note from a months-long assessment of neuroimaging: why „most cited“ is a measurement with an instrument, a sampling frame and a drift — and why attention is not evidence.
A working history of software testing — the people, the papers, the arguments still open, and the one problem nobody has solved.
Errors are findable. Conceptual problems are not. In AI-accelerated development the gap between these two grows — and only tight communication loops can close it.
Testing is not a phase. It is a measurement discipline — as invariant to engineering as dimensional tolerance analysis is to mechanical design.