[R] Problems in Recommending R
h.wickham at gmail.com
Tue Feb 3 17:00:01 CET 2009
On Tue, Feb 3, 2009 at 9:03 AM, Peter Dalgaard <p.dalgaard at biostat.ku.dk> wrote:
> hadley wickham wrote:
>>>> 1. avoid clicking on the "R version 2.8.1" link - that takes you to a
>>>> directory listing of strangely named files
> Yes, this is of course much harder than avoiding to read the two bullet
> points labeled "Getting Started"
You think people read text before clicking on links? ;) Next you'll
be trying to persuade me that people read the R startup text. I
certainly had never noticed that link - the wrong words are higlighted
by the link - the important part of that sentence is "download R",
"CRAN mirror" is just an implementation detail.
>>>> 2. recognise that you need to click on an CRAN (what is a cran?)
> Clicking on the words "CRAN mirror" never comes to people's mind. I mean,
> it's not like the first line of the link explains the acronym, is it?
Which first line? You have to click on the link to find out what it means.
>>>> 3. successfully select a mirror that is up-to-date (with no
>>>> information about which mirrors are up-to-date)
> Of course the R project website is required to keep track of that.
Well someone needs to.
>>>> 4. click Windows (ok, this one is easy)
>>>> 5. guess that base is the "distribution" that you want
> Yes, there are two links so people will take the other one.
Why would anyone ever click on contrib? Why not move the content of
base to that directory and then provide a link to contrib? That would
save one step in the process.
>>>> 6. phew, you're there (but don't follow the advice to download from a
>>>> mirror near you or you'll be back at step 3)
>>> Its even more confusing than that because actually you're not there
>>> yet! You have to click on the unobtrusive patched link and then
>>> download that or you get the version with the bugs.
>> Wow, I'd never noticed that before!
> However, you are only supposed to need it if you are one of the exceedingly
> rare users who happens to be bitten by a bug that went undiscovered through
> several weeks of user testing.
> (Warning: The above may contain traces of sarcasm...)
More information about the R-help