[R] Protocol for answering basic questions

Marc Schwartz MSchwartz at MedAnalytics.com
Thu Dec 2 16:36:25 CET 2004


On Wed, 2004-12-01 at 21:52 -0500, Duncan Murdoch wrote:
> On Thu, 02 Dec 2004 00:55:53 +0100, Carlos Javier Gil Bellosta
> <cjgb at wanadoo.es> wrote:
> 
> 
> >Over the years, while learning C, Java or Python, I have found very 
> >useful a few IRC channels on those languages where one could get (and 
> >provide!!) peer-to-peer support. Should a rather informal, open and 
> >publicited one exist for R, I believe it could channel many of these 
> >basic questions and, probably, many others some novices do not dare ask.
> 
> As far as I know, there is currently no IRC channel on R.  Why not set
> one up?  (I've got no idea what is involved in that.)  You'll need to
> attract participants to it.  Various ways are:
> 
>   - announce on this mailing list when you have created it.
>   - put a note about it in your signature
>   - use it, and be helpful to other users on it
>   - get a mention of it on an appropriate r-project.org page.
> 
> I've never used IRC and wouldn't be likely to be a participant (in
> general I dislike live online chat), but to each his own.  
> 
> Duncan Murdoch

It is relatively easy to set up an IRC channel. One needs to select an
IRC network and register and configure a channel on that network and
then make that information available to a target audience. The key is
that someone or multiple folks would need to be designated as channel
"ops", who in effect become the moderators for the channel and can act
accordingly when required.

There are various IRC clients for multiple operating systems that would
allow for relatively ubiquitous access.

In my mind, a significant part of the problem with IRC however, is that
knowledge is lost. There is no searchable log or archive of the
conversations and the exchanges that take place. Thus, there is no way
for others to take advantage of the knowledge sharing and this only
serves to perpetuate the repetition of queries and responses in that
environment.

Even worse, as the traffic level increases, many of the exchanges tend
to move from the public channel to DCC chats, because the main channel
traffic becomes impossible to follow. Thus, even those who may log in to
the channel as potential new users, can eventually find that there may
be a minimal amount of traffic in the main channel, as many of the
conversations are now taking place in private one-on-one or group chats.

While the above is clearly a generalization, I have experienced it
personally over the years in a variety of IRC channels. It can become
quite frustrating.

Part of the problem with attempting to separate out "basic" questions to
a new list is how does one define "basic"? Where is the line, or more
likely the grey buffer zone, between what may be clearly basic versus
something that ultimately becomes more involved.

How does one separate out what may seem to be a basic plot query at the
outset, when that query is actually entangled with the more involved
aspects of the plot/predict methods applicable to a linear regression
model or how data needs to be structured for a particular plot function?

These present difficult challenges and choices, many of which have been
discussed here more than once, at least over the past three plus years
that I have been in this community.

That does not mean that a single, highly motivated person or group of
folks cannot pursue an alternative venue for sharing knowledge apart
from these lists. That is quite common in most communities.

Indeed, this is the basis for the add-on packages in CRAN, Contributed
Documentation, the various GUI's that are available and the Bioconductor
group as examples. In each case, the basic functionality of R has been
substantially and meaningfully extended by individuals and/or groups who
defined a need and took it upon themselves to fulfill it. Those
additions have been made available back to the community at large, which
is consistent with the overall philosophy of open source initiatives and
the selfless desire to make contributions available to aid others.

As another example, how may different independent online support
resources are there for the various Linux distributions between online
forums, FAQ's, wiki's, e-mail lists, personal web sites and Usenet
groups? Some are "official" resources, while others are created
independently by users for the benefit of other users.

And...if one wonders whether 'basic' questions get asked in other
communities, one only need review a message that was posted to the main
Fedora list yesterday, which had a subject line of:

"Does Fedora include Linux?"

The key question in the body of the message was:

"So i was wondering does the Fedora Download include a OS (operating
system) with the download? Such as like Linux?"

The author of this message BTW is a 'webmaster' by his sig. You can
imagine what the range of responses to that question was.


There are several web sites already created by useRs that are available
which are highly complementary to these lists and to the main R web
site. There is nothing to prevent folks from continuing to advance those
types of activities and making those resources known to the community at
large. 

If those independent resources serve to advance the sharing of knowledge
with respect to R, terrific! That is what this is all about, users
helping other users. 

Ultimately, it is likely that each such resource will attract some
particular subset of self-selected participants, perhaps based upon
skill level, the nature of the participants and/or specific areas of
interest. In this way, users can locate and participate in one or more
support resources that best meets their particular requirements.

To make that a reality however, much like the recent discussion on
documentation, will take one or more persons who are motivated to make
it happen. 

Hopefully, some will step up and do just that. 

It is unreasonable and incredibly unfair to expect that R Core can or
will do it all. Given the selfless sacrifices that they have made to
bring R to its current state, it is incredible that they continue to
contribute to these lists as much as they do. One might want to keep
that in mind when on the receiving end of the occasional terse RTFM
reply.

Best regards to all,

Marc Schwartz




More information about the R-help mailing list