here are some of the findings from my analyses
of the various "experiments" done over at d.p.

the best proofers miss between 5% and 25% of the
errors over the course of proofing an entire book.
the worst proofers miss an even higher percentage,
but it is not all that much higher, probably 10-40%.

there is no evidence for the position that proofers
"get bored" and therefore miss a higher percentage
if the text they proof is clean (i.e., has few errors)...

p3 proofers are no better than p2 or p1 proofers.

some errors withstood over 5 rounds of proofing;
there was nothing obviously "difficult" about them.

the best predictor of whether a page is now "clean"
is how many people proof it without finding an error.

if the last person to proof a page found an error,
then you cannot reliably predict it to be error-free,
no matter how confident the proofer believes that...

if anyone wants to dispute or discuss these findings,
i'd be open, and will ask about your supportive data.

-bowerbird