Request for comments about the Ubuntu Signpost

[Andrew Sayers]

Hi Dougie,

Before I talk about specific points, I'd like to bore you with a bit of 
theory:

I see the signpost as a specific implementation of a 
FAQ/Flowchart/chatroom model that could be a powerful approach to a 
number of support and documentation issues.  The general idea is that 
you start by answering questions in a chatroom, write up each answer as 
it's provided, then gradually develop a hierarchy of questions that 
people ask.  So long as each question in the hierarchy has no more than 
about 7 answers, it should be quite intuitive to navigate.

Since the hierarchy of questions follows from the questions asked in the 
chat room, the key issue is: what sorts of questions belong in the 
chatroom?  A chatroom about renting movies will lead to a different 
hierarchy of questions than a chatroom about car repair.

In the specific case of the signpost, the original idea was for the 
chatroom to handle questions of the form "where should I go to 
ask/learn/talk about X?".  I think that's different from "how do I solve 
problem X?", which is more like the domain of #ubuntu.  The signpost was 
born[1] out of a thread[2] in ubuntu-devel-discuss which also discussed 
doing something similar for support questions[3].  If anyone shows an 
interest in the long, hard process of writing up support answers, I'll 
be sure to point them in the direction of this list.

The signpost cheated a bit on the chatroom side of the equation. 
Everyone that's been around Ubuntu more than a week has seen questions 
that belong somewhere else, and it'll be a while before people hear 
about #ubuntu-signpost, so we wrote up an initial version based on our 
experiences.  My expectation is that the signpost will mutate quite a 
lot while adapting to real users.

One of the issues brought up in a post to ubuntu-devel-discuss[4] was 
about people becoming your "new best friend".  That is, if you answer 
somebody's question, some people will start looking at you as the first 
place to go when they have any problem at all.  Different groups deal 
with that issue in different ways - for example, #ubuntu has made it the 
purpose of the channel to answer any possible question.  I personally 
plan to deal with the issue by avoiding giving any actual answers, but 
instead pointing people in the direction of help.

Dougie Richardson wrote:
> 
> I think there's such crossover here that a joint effort would avoid
> any duplication of effort and produce a more complete end result - as
> you said above about having every answer - I think that certainly be
> our goal.
> 
> Combining our efforts would or working together would be my preference.
> 

My personal interest in the signpost is to make sure that everyone asks 
the right questions in the right places.  If I'll be manning 
#ubuntu-signpost much of the time, I'd rather not be expected to answer 
questions about using the Swahili localisation of Avidemux.

Rather than answering every question, I would prefer to look at 
pioneering the FAQ/flowchart/chatroom model within the limited domain of 
pointing people towards answers, then trying to get other teams 
interested in the model.  For example, #ubuntu is full to bursting, and 
answering common questions (or perhaps just helping people to formulate 
better answers) strikes me as useful to them.  If they develop a 
flowchart, it would be trivial to link to it from the signpost.

> Do you mean the UF Beginners Team?  Because there are several members
> in the doc team.  I'd perhaps also get in touch with the learning team
> as they are very forum oriented and working in a similar area.
> 

I do, and will do :)

> Phil and I had discussed work on data mining resources like the forums
> for common questions.  I don't think you can be purely reactive people
> may expect answers to questions they think are too obvious to ask.

That's a good point.  Perhaps "evidence-lead" would be a better phrase 
than "reactive", as my concern is about writing up questions that bear 
no resemblance to the style or content of real user questions.

Mining the forums strikes me as a useful job, although I think it would 
have to be done in collaboration with forum helpers.  They'll be the 
ones to maintain it, so they need to buy in to the solution.

My limited experience with forum helpers suggests that they prefer to be 
given an idea to critique, rather than half an idea to develop.  That 
brings me back to pioneering the model in a small field, then 
advertising it to wider groups once we've worked the kinks out of it.

How would you feel about me writing up the above on a wiki page, which 
people could then edit as our understanding of the model improves?

	- Andrew

[1]https://lists.ubuntu.com/archives/ubuntu-devel-discuss/2009-July/008977.html
[2]https://lists.ubuntu.com/archives/ubuntu-devel-discuss/2009-July/008971.html
[3]https://lists.ubuntu.com/archives/ubuntu-devel-discuss/2009-July/009031.html
[4]https://lists.ubuntu.com/archives/ubuntu-devel-discuss/2009-July/009037.html