[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: OT: unicorn, was: one of the many reasons why removing non-free is a dumb idea

> On Thu, Jan 08, 2004 at 10:23:30AM -0500, Raul Miller wrote:
> > Actually, absence of evidence really is evidence of absence.

On Fri, Jan 09, 2004 at 02:08:23AM +1000, Anthony Towns wrote:
> Uh, no it's not. Eg, "I don't have any bug reports for debootstrap 0.3;
> that's evidence that there aren't any bugs in it."

It's totally inadequate evidence, but nevertheless it's evidence.

> The lack of evidence is due to the fact that (almost) no one else has
> seen debootstrap 0.3, so there hasn't been any opportunity to file
> bug reports about it; not that there aren't any bugs.

And after many years of experience with software we expect that all
software of any complexity has bugs, regardless of any evidence to the
contrary.  This is complicated by the fuzziness of the concept of "bug".

But this is also a matter of degree -- if there's no evidence after
one person searches for one hour, that means less than if there's no
evidence after a thousand people search for five years.  [And if we know
something about the capabilities of those people that knowledge adds to
the evidence.]

If Sven has spent a lot of time attempting to find DFSG free adsl support
software for a specific card, and has contacted the manufacturer and
even the manufacturer of that card is not able find such software,
that's a different kind of evidence than no bug reports being filed on
a newly released piece of software.

This is not to say that such software will never exist.  It does not
prove that such software does not currently exists.  It is, however
evidence that it does not exist.

However, I think the point is that evidence isn't proof.

> > Or maybe there really is a little unicorn in your sock drawer.
> That's Occam's razor, which says you should draw the conclusion that
> requires the least on assumptions for which there's no evidence. ("There's
> no unicorn" requires no assumptions; "There is a unicorn" requires the
> assumption that it hides whenever you try looking for it, that it's always
> very quiet, and that it hid when you tried hooking up the flashlight and
> webcam...)

Occam's razor is another set of words for talking about the absence
of evidence.


Reply to: