dakretz said:
>   This comes up all the time.

it has been suggested before, yes...


>   a. It's socially unacceptable by acclamation.

...and is usually deemed socially unacceptable, true...

but i'm not sure that that couldn't be turned around,
assuming people are genuinely looking to be tested,
in a sincere desire to improve.  it is a bit repugnant,
in my opinion...  but so is the secretive collection of
data on proofing accuracy that rfrank is doing now,
especially since he's not revealing that to _anyone_,
including the very people he is collecting data on...


>   b. All you prove is that the proofer did or
>   didn't catch an error you already knew about.

well, i don't necessarily agree with that.  i'd believe
your ability to catch the introduced errors would be
highly correlated with your general overall accuracy.
so if you value that metric, there's one good purpose.
(but, for the record, i believe that metric is valueless.)

further, the ability to detect the error _immediately_,
and show it to the proofer on-the-spot might well be
the very best feedback necessary to get their attention,
an argument which hasn't been fully considered before.

so there _could_ be some real value in this technique.

now, i certainly wouldn't do such error-injection on a
proofer without their express approval, because it is
entirely too _sneaky_ when you are doing it that way,
and jeopardizes the trust-relationship, which is vital.

but if a proofer _asked_ for it, i think it would be ok...

i never subscribed to the "i need to have some errors
to keep myself from betting bored" philosophy...  but
for a proofer who does, this could well be an answer.

bottom line, though, i just don't think it's necessary...

-bowerbird