[Rd] CRAN Server download statistics (Was: R Usage Statistics)

[Fellows, Ian]

Gabor,

Well, I think "discriminates" is a bit over the 'top'. It just measures something else. No statistic is going to be the One True Popularity Statistic. Downloads, Unique IPs and Top level downloads all measure different aspects of popularity.


Ian

________________________________________
From: Gabor Grothendieck [ggrothendieck at gmail.com]
Sent: Monday, November 23, 2009 3:15 PM
To: Fellows, Ian
Cc: hadley wickham; Stefan Theussl; R-devel
Subject: Re: [Rd] CRAN Server download statistics (Was: R Usage Statistics)

On Mon, Nov 23, 2009 at 3:51 PM, Fellows, Ian <ifellows at ucsd.edu> wrote:
> 6. Regarding package dependancies, I was thinking about also counting the number of top level downloads, as approximated
> by the number of downloads where a reverse dependancy was not downloaded in the next 5 min by the same IP.

Top level downloads discriminates against infrastructure packages,
i.e. packages that are not necessarily directly used by users or even
downloaded directly by users.  It can make it seem as if a package
that is widely used, but not directly by users, is barely used at all
so total downloads seems a better metric of popularity than top level
downloads.

I can't find it just now but remember reading a post on another list,
or maybe it was his blog, by the author of certain software (not
related to R) that was fundamental to Ruby on Rails but as
infrastructure would not be directly observed by the user yet Ruby on
Rails got all the credit and his key contribution got almost none
despite the fact that Ruby on Rails likely would have never achieved
its current high level of popularity without his contribution.