develooper Front page | perl.qa | Postings from July 2008

Re: counting usage, not downloads

Thread Previous | Thread Next
From:
Eric Wilhelm
Date:
July 6, 2008 14:53
Subject:
Re: counting usage, not downloads
Message ID:
200807061452.49851.ewilhelm@cpan.org
# from Andy Lester
# on Sunday 06 July 2008 10:33:

>FWIW, my line of thinking has nothing to do with how many people are  
>downloading or installing a module as an indicator of quality.

Note that I said "usage" -- not downloads, not installations.

I do not suggest counting downloads (and we cannot do it anyway.)

I do not suggest counting installations (it could be done, and might be 
interesting (see popcon), but lacks meaning.)

Usage is neither of these.  (Sorry about the echo.  It seems like I 
said "don't count downloads" and then 3 people said "no, don't count 
downloads".  Some trend of disagreeable affirmation?)

>I'm far more interested in human commentary.

Usage is a vote of confidence - it represents a human decision.  It 
gives you a number which represents the users who are likely to notice 
if the module breaks.  Further, it is a sort of voting which can both 
1) be done by the machine and 2) has meaning.  In that sense, it 
represents the shortest applicable form of human commentary.

If a usage count comes from e.g. your internal project's META.yml, that 
is clearly a direct human decision, and a data-point which is currently 
unavailable.

I would love to read a handful of carefully-written reviews, but it is 
far easier to gather meaningful numbers.

>Counts of who uses a given product puts the Britney Spears as the  
>highest quality musical artist.

"Highest quality music" is clearly even more amorphous than "best 
module" - and quality is never conveyed by popularity even in products 
where quality is way more objective.  However, there is clearly a 
*demand* for the "Britney product" and that indeed says something about 
the product, but has no meaning when compared to tea sales in china.

Different products have smaller markets.  If you're looking for two guys 
exploring multiple time signatures with 11 songs in 16 minutes and no 
lyrics, you are probably not waiting for the radio to play it for you.

  http://cdbaby.com/cd/seangrindcore

So, you have to judge using your own parameters, but what if you had 
available the information of how many users a module had?  If it had 6 
reviews and 2 usage counts, that means something different than 2 
reviews and 2000 usage counts -- but both have meaning.

Anyway, I'm only writing all of this in the hope that someone might get 
around to implementing it before me.  That someone might not be you, 
Andy (given that you'll be busy writing careful reviews of 13830 
modules.) :-D

--Eric
-- 
I arise in the morning torn between a desire to improve the world and a
desire to enjoy the world. This makes it hard to plan the day.
--E.B. White
---------------------------------------------------
    http://scratchcomputing.com
---------------------------------------------------

Thread Previous | Thread Next


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About