[i_think_so]

Is it really a constant rate? Or is it a Law of Large Numbers kind of thing, where past a certain scale the randomness gets smoothed out and looks constant? Or something else?

(Obviously some developers are better or worse than others, so I presume your observation is assuming developer skill as a constant.)

[jpollock]

Well, I think there are two things at play here.

1) As org size grows, it's the team's average quality that matters (so yes, large numbers).

2) Even with a single team, the velocity will increase to match the acceptable level of quality.

Management will push the accelerator until they get too many bugs, then it will be "we need fewer outages".

So, in an team+environment, you end up with a constant (in time) detection rate, which basically means a constant in time injection rate.

If the teams' velocity increases without increasing quality, the bug injection rate (and detection rate) will increase.

AKA if the AI is slightly worse, but 10x faster, stop carrying the pager. :)