[R] Size of R user base

[Philippe Grosjean]

>On Tue, Apr 20, 2004 at 10:47:06AM +0200, Philippe Grosjean wrote:

>> The main role of this code would be, of course, to warn users that an
update
>> is available. One side effect would be that it should be possible to
monitor
>> access to the link in the centralized web site and to know:

Tamas K. Papp answered:

>I appreciate that it would be really nice to know the size of the user
>base: the fact that lots of people are using R would give everybody a
>warm, fuzzy feeling.

OK, as I said in my mail, the goal is different, but anyway...

>On the other hand, some people are getting carried away with this
>discussion.  I would consider the above suggestion a serious violation
>of my privacy: if I want to make this type of personal data (what
>software you use is personal data, in the legal sense) public, I can
>do it, but I don't want any software on my computer which "kindly"
>does that for me if I did not ask for it.  I don't care whether the
>data would be anonymized -- it is a matter of principle.

You have strong principles, excellent position probably for saving Freedom.
However, do you really apply them? If yes, it means you never use your web
browser except with https:// addresses (or even not at all), because you
could say the same thing about the web: everybody can track how you use it.
The fact is that you use internet (because you posted this mail). So what?
You accept the architecture of the web and the way it operates with no
confidentiality at all, except in secure sites, and you violently protest
against doing the same thing for software usage?

That said, and after thinking a while, yes, what's wrong in what I propose
is that the piece of code does not ask to the user if it can track its
use... a subtle, but important difference! However, think at the following
story. Do you know that voting in Belgium is not a right, but an
*obligation*. Why do you *have to* vote in Belgium (you would said, this is
my freedom: if I do not want to vote, I don't... then you can be jailed
according to Belgian's law)? The reason is that if vote is not obligatory,
only a fraction of people do vote, and there is a bias in the results (for
instance, because extremists tend to vote more than others). So, what's the
matter to ask people that use R to accept the lost of a microscopic fraction
of their Freedom in sending such a basic information as "I use it".

>So before people are coming up with ingenious (and possibly invasive)
>methods to estimate the user base, I would like somebody to explain
>why this is relevant.  R already has the "critical mass" it needs,
>given that it is a specialized piece of software.  It is always good
>to have a broader user base, but R is doing fine as it is.  In some
>markets, it is safer to pick a product with a wider user base (think
>VHS vs Betamax), and sometimes this argument applies to software (eg
>you should not make your bet on a commercial, closed-source software
>company with a dwindling user base, because if they go bankrupt, you
>can expect no updates etc).  But free software is entirely different.

No software is eternel. I think a good picture of the change in time of R
users would be an interesting tool to have an idea on how well it fits or
don't fit users' needs. Moreover, if you develop for both R and S-PLUS, and
think to develop only for one software only in the future, it would be nice
to know what fraction of the "S language and environment" users you miss
this way.

Best,

Philippe Grosjean