Re: Response to "Position Statement to the Dunc-Tanc "experiment""

[Michael Poole <mdpoole@troilus.org>]

Thibaut VARENE writes:

> Sorry for the wording but it's way more than I can take:
>
> On 10/27/06, Anthony Towns <aj@azure.humbug.org.au> wrote:
> [...]
>> > - How is the success of this "experiment" measured? (For the release as
>> >   well as for the entire project)
>>
>> An "experiment" is successful as long as it provides useful information.
>
> What the fuck is this definition of "successful"??!

  Experiment \Ex*per"i*ment\, n. [L. experimentum, fr. experiri to
     try: cf. OF. esperiment, experiment. See {Experience}.]
     1. A trial or special observation, made to confirm or
        disprove something uncertain; esp., one under controlled
        conditions determined by the experimenter; an act or
        operation undertaken in order to discover some unknown
        principle or effect, or to test, establish, or illustrate
        some hypothesis, theory, or known truth; practical test;
        proof.

The experiment may succeed and yet show that Dunc Tank is not a great
use of funds.  I have no idea or personal interest in whether it shows
that or whether it shows that Dunc Tank is a wonderful use of funds.
But an experiment is by definition a way to test a question, so if it
provides useful (relevant) information, it is successful *as an
experiment* even if it is not an ideal way to expend resources, which
includes developer time and attention as well as money.  (If the ideal
way to expend resources were known in advance, no experiment would be
necessary or appropriate.  However, no one has shown or plausibly can
show a provably ideal way to expend these resources.)

Michael Poole