[Catalog-sig] Package comments

Laura Creighton lac at openend.se
Fri Nov 6 11:58:49 CET 2009

In a message of Fri, 06 Nov 2009 10:49:09 +0100, "M.-A. Lemburg" writes:

>There will certainly be some developers that don't like being
>rated for things they put on PyPI as a service to the community.
>Others will probably take it as challenge to get a high rating.

Yes, but I consider this to be a _very_ _bad_ _thing_.  Competition
is only useful if the contests can be made fair.  And this is precisely
what we cannot guarantee with a rating system.  Thus we can expect that
many of the packages with higher ratings are quite arguably worse than
those with poorer ones.  What is worse is that it sets up a way to
compare packages when they, themselves, need to be evaluated on their
own merits.  We could expand the rating system along the lines that
René Dudfield outlined as is used in pygame -- we could rank software
for ease of install, and completeness of documentation and the like --
establishing categories with strict rules so that we can make sure
the competition is fair, but there comes a point where what you want
to know is 'how good is this software' -- and the answer is very much one
of 'depends on your personal taste' and 'depends on what you want to do'.
There isn't any way to get around this.

But one of the reasons that these multiple ways to do things can
peacefully coexist is that we don't encourage people to go around
saying 'my package is better than your package' or 'I have more fans
than you do' or even 'the packages I use are better than the other
ones'.  People in comp.lang.python who want to write glowing
recommendations of why you should use X instead of Y have to
substantiate why they prefer X to Y, which makes for more informative
reading.  I think that a rating system will encourage the sort of
uncivil behaviour I have seen other places, places where there is too
much competition, unfair contests, not enough acceptance of multiple
ways to approach a given problem, and where people expect to be
spoon-fed simpler answers than their problem admits.

By having a rating system, we are saying that a rating system is
useful.  And I think it is at best useless, and more often very harmful.


More information about the Catalog-SIG mailing list