HN Explorer

11 comments

[ 5.0 ms ] story [ 42.8 ms ] thread
tosh
> The distance to decision makers and the number of developers working on a project is clearly and unambiguously the issue that is the best predictor of future problems with a code base.
valerij
> Another shocking discovery for me personally, is that the only one that I've actually used myself - code coverage - has the lowest recall. In other words, almost all the issues it predicted turned out to not be real issues.

sounds like survivorship bias to me

drewcoo
I'd be willing to bet that it was sloppy coverage and those false positives were along component/team boundaries.
drewcoo
Take Conway's Law into account and it's maybe not so surprising.

https://en.wikipedia.org/wiki/Conway's_law

yodon
The original paper on which this blog post is based is actually an explicit experimental test of Conway's law (and it's also a paper which happens to have found significant validity for Conway's law in the data they analyzed).
franciscop
Could a mod please update the title to make it less click-bait?

"According to Microsoft, the number one predictor of software bugs is" sounds like all those side/bottom extra articles.

dang
I took a crack at it, but the article has been flagged at this point so it's a bit moot.