[R] creating an equivalent of r-help on r.stackexchange.com ? (was: Re: Should there be an R-beginners list?)

Liviu Andronic landronimirc at gmail.com
Tue Feb 4 01:11:42 CET 2014


Dear Don and Bert,
Allow me to address some of your concerns below.


On Mon, Feb 3, 2014 at 9:56 PM, Bert Gunter <gunter.berton at gene.com> wrote:
> I find SO's voting for posting business especially irritating. I wish
> merely to post or to read the posts of others without being subjected
> to some kind of online pseudo game and ratings competition. That alone
> keeps me away. But Don said it better.
>
On SO voting is irrelevant for either posting a question or an answer.
*Anyone* (with an account) can ask a question, and *anyone* can answer
a question. Their system of privileges is explained here:
http://askubuntu.com/help/privileges . But to summarize:
- if you're interested only in giving help, then the only really
relevant threshold is 10 and 50 votes (removing some new user
restrictions and allowing you to comment on posts, respectively)
- if you're interested only in seeking  help, then all thresholds are
irrelevant really

All other thresholds are relevant only if you're interested in
contributing to the organization of information, or in moderating this
whole forum-slash-wiki thingy. And as a note, given the quality of
your answers on r-help, Bert, I have no doubt that you will clock
upwards 50 upvotes in a couple of hours or so.


> I realize that I may be out of step with the masses here, and the
> masses should certainly decide. Hopefully I won't be around if/when
> they decide that R-help should go.
>
The proposal is not necessarily to close down r-help. From the myriad
lists it currently has, R Core could keep only r-help and r-devel, and
encourage new users to seek help on r.stackexchange.com. The scope of
r-help could be redefined.


> On Mon, Feb 3, 2014 at 12:42 PM, MacQueen, Don <macqueen1 at llnl.gov> wrote:
>> - They waste copious amounts of screen space on irrelevant things such as
>> "votes", the number of views, the elapsed time since something or other
>> happened, fancy web-page headers, and so on. Oh, and advertisements. The
>> Mathematica stackexchange example given in a link in one of the emails
>> below (http://mathematica.stackexchange.com/) illustrates these
>> shortcomings -- and it's not the worst such example.
>
Well, I've seen my fair share of advertisements on Gmail, Yahoo Mail
or what have you. I know some use dedicated clients, but not all do.
(And sofar I haven't noticed one single intrusive or distracting ad on
SE.)

As for the number of votes, this is actually the most useful bit of
this Q&A interface: it allows for the best questions (or most often
asked) to stand out from all the noise. And it allows for the best
answers (or those most authoritative) to stand out, too. Accepted
answers immediately indicate to others seeking similar help what has
worked for the OP. Very useful stuff.

Voting also naturally allows to differentiate between neophytes
(<100), and professional helpers (>1k; think of Brian, David or, as it
happens, Bert). If you remember long ago someone proposed on r-help a
reputation system for our professional helpers, only to be rebuffed
essentially because it is unfeasible in a ML interface. The SE Q&A web
interface---or similar---naturally handles this.



>> - In most if not all cases, one has to login before posting. I have too
>> many usernames and passwords as it is.
>
Fair point. However SE found a neat way around this: it keeps cookies
around and whenever you close the browser and reopen SE, it identifies
the cookie and auto-logs you in.


>> Right now, at this very moment, in my email client's window I can see and
>> browse the subject lines of 20 threads in r-help. And that's using only
>> about half of my screens vertical space. In contrast, in the Mathematica
>> stackexchange example, I can see at most 10, and that only by using the
>> entire vertical space of my screen. The "From" column in my email client
>> shows the names of several of the people contributing to the thread, which
>> the browser interface does not. In the email client, I can move through
>> messages, and between messages in a thread using my keyboard. In a
>> browser, I have to do lots of mousing and clicking, which is much less
>> efficient.
>>
Again, fair point, but with SE you quickly realize that this is
irrelevant. On ML, even more so on r-help, the only sane way to sort
and filter the messages is using time. If a question wasn't answered
in 24h (or, to be generous, a week), chances tend to zero that this
question will ever be addressed. On SE it is absolutely normal for a
question to be answered, with a high-quality input, 3 months or 2
years later.

It is also much easier to filter questions by topics: if you're
interested in GUI or plyr related questions, just display those tags,
and then answer relevant questions. On r-help you may only  guess from
the subject line what the question could possibly be about.

The Q&A interface also allows easily to redirect users to similar
questions that were already answered (goodbye "PLEASE do read the
posting guide"), thus identifying duplicate questions. It also makes
it much easier to search for topics of interest that were already
addressed in the past; much easier than scouring the mountains of
untriaged r-help content.

And do not underestimate the soft incentives induced by the voting
system. Users seek upvotes (you can set bounties, get moderator
privileges and so on), thus making them interested in giving
high-quality answers and asking high-quality questions. Very well
thought-out stuff.


>> As it is now, r-help messages come to me. I don't have to start up a
>> browser. So it's much easier to go take a quick look at what's new at any
>> time.
>>
Agreed. I understand the frustration from using a different medium.


>> True, I had to subscribe to the mailing list, which involves a username
>> and password. But once it's done, it's done. I don't have to login before
>> posting, which means I don't have to remember yet another username and
>> password.
>>
Mostly same happens with SE, the way they set it up.


>> What "...duplicated efforts of monitoring multiple mailing lists)"? I have
>> no duplicated effort...in fact, I have almost no effort at all, since the
>> messages come to me. There was some initial setup, i.e., to filter
>> different r-* messages to different mailboxes in my email client, but now
>> that that's done, it's as simple as clicking on the correct mailbox.
>>
Do you follow r-sig-gui or r-sig-teaching or r-sig-finance or
r-sig-robust? Does Brian follow them all? Probably not. People who are
seeking specialized help have a hugely reduced chance of getting
useful help.

On SE however, the efforts are not fragmented; all questions are asked
and answered in the same place. If a question pertains to 'plyr' and
'finance', either a finance type or a plyr enthusiast are as likely to
answer. For the r-sig-* MLs, one would need to subscribe to all MLs
and monitor them all; few do so.


>> In other words, in every way that's important to me, the mailing list
>> approach is superior. I do not support abandoning the mailing list system
>> for any alternative.
>
I'm not an SE  evangelist, and only truly discovered it about a month
ago or so (even though it seems that I had registered more than a year
ago), and initially I was quite very skeptical of this "fancy forum".
But when I actually realized how _efficient_ this Q&A interface is, I
quickly decided that r-help and associated r-sig-* were good to go the
way of the usenet. Long story short, the Q&A interface is impressive
in terms of economic efficiency, i.e. matching up supply and demand;
the ML is quite inefficient in comparison.

Kind regards,
Liviu


>
>> -Don
>>
>> --
>> Don MacQueen
>>
>> Lawrence Livermore National Laboratory
>> 7000 East Ave., L-627
>> Livermore, CA 94550
>> 925-423-1062
>>
>>
>>
>>
>>
>> On 2/2/14 1:49 PM, "Liviu Andronic" <landronimirc at gmail.com> wrote:
>>
>>>Dear Duncan,
>>>I discovered something interesting wrt to the licensing and mirroring
>>>of user-contributed material on StackExchange.  Please read below.
>>>
>>>
>>>On Sun, Nov 24, 2013 at 9:00 PM, Duncan Murdoch
>>><murdoch.duncan at gmail.com> wrote:
>>>>> I'm not aware of a discussion on this, but I would say no.
>>>>> Fragmentation is bad. Further fragmentation is worse.
>>>>>
>>>>> TL;DR
>>>>> =====
>>>>>
>>>>> Actually I'd say all mailing lists except r-devel should be moving to
>>>>> StackOverlow in the future (disclaimer: I'm not affiliated with it).
>>>>
>>>>
>>>> I would generally agree with you, except for a few points.
>>>>
>>>> 1.  I avoid StackOverflow, because they claim copyright on the
>>>>compilation.
>>>> As I read their terms of service, it would be illegal for anyone to
>>>>download
>>>> and duplicate all postings about R.  So a posting there is only
>>>>available as
>>>> long as they choose to make it available. Postings to the mailing list
>>>>are
>>>> archived in several places.
>>>>
>>>It seems that StackOverflow is officially proposing user-generated
>>>content for download/mirroring:
>>>http://blog.stackoverflow.com/2014/01/stack-exchange-cc-data-now-hosted-by
>>>-the-internet-archive/?cb=1
>>>
>>>"All community-contributed content on Stack Exchange is licensed under
>>>the Creative Commons BY-SA 3.0 license. " And it is currently being
>>>mirrored at least at the Internet Archive:
>>>https://archive.org/details/stackexchange
>>>
>>>So, in principle, it would be possible/desirable to:
>>>- spin the 'r' tag from StackOverflow and propose an r.stackexchange.com
>>>at
>>>http://area51.stackexchange.com/categories/8/technology . Such a SE
>>>site would be similar to http://mathematica.stackexchange.com/
>>>- involve R Core to give blessing for using the R logo, if necessary.
>>>This would be similar to what Ubuntu does with AskUbuntu:
>>>http://meta.askubuntu.com/questions/5444/is-ask-ubuntu-official-ubuntu
>>>- set a mirror on r-project.org for all the user content that is
>>>produced by r.stackexchange.com , and thus allow R Core to keep the
>>>info publicly available at all times. The mirroring on Internet
>>>Archive would still hold.
>>>
>>>
>>>> 2.  I think an interface like StackOverflow is better than the mailing
>>>>list
>>>> interface, and will eventually win out.  R-help needs to do nothing,
>>>>once
>>>> someone puts together something like StackOverflow that attracts most
>>>>of the
>>>> people who give good answers, R-help will just fade away.
>>>>
>>>The advantages for such a move are countless (especially wrt to
>>>efficiently organizing R-related knowledge and directing users to
>>>appropriate sources of info), so I won't go into that. I would only
>>>note that most 'r-sig-*' MLs would become obsolete in such a setup,
>>>and would be replaced by the much more efficient tagging system of the
>>>SE Q&A web interface (for example, all posts appropriate for r-sig-gui
>>>would simply be tagged with 'gui'; no need for duplicated efforts of
>>>monitoring multiple mailing lists).
>>>
>>>Opinions?
>>>
>>>Liviu
>>>
>>>______________________________________________
>>>R-help at r-project.org mailing list
>>>https://stat.ethz.ch/mailman/listinfo/r-help
>>>PLEASE do read the posting guide
>>>http://www.R-project.org/posting-guide.html
>>>and provide commented, minimal, self-contained, reproducible code.
>>
>> ______________________________________________
>> R-help at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-help
>> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.



-- 
Do you know how to read?
http://www.alienetworks.com/srtest.cfm
http://goodies.xfce.org/projects/applications/xfce4-dict#speed-reader
Do you know how to write?
http://garbl.home.comcast.net/~garbl/stylemanual/e.htm#e-mail




More information about the R-help mailing list