Standing in the street trying to hear yourself think

Evan eapache at
Wed Jul 8 22:03:41 UTC 2009

On Wed, Jul 8, 2009 at 1:28 PM, C de-Avillez <hggdh2 at> wrote:

> On Wed, 2009-07-08 at 12:47 -0400, Evan wrote:
> >
> >         The only issue I can find with this approach is that many new
> >         users are
> >         coming from windows. Have you tried using windows "help"? It
> >         does use
> >         an approach similar to this, and I would be afraid that many
> >         of those
> >         users will dismiss this as soon as it starts. Everytime I have
> >         attempted to use the help in windows, the Q & A ends with
> >         frustration
> >         on my part when it says basically "can't figure out what is
> >         wrong".
> >
> >         If those new users can be convinced this will not be the
> >         results every
> >         time in Ubuntu, this could be an excellent help system.
> >
> > You have a point. Hopefully the addition of a "Get Live Help" button
> > will mitigate the problem though. Even if the automated help doesn't
> > give any useful suggestions, it should be able to reliably determine
> > where it ought to put you in IRC. How bad would it be for it to say
> > "Sorry, the automated system couldn't solve your problem. Please wait
> > while I connect you to a human who should be able fix it." ?
> >
> I can see something like this working -- as long as the requester gets
> paired with one single person, in a PVT IRC session (or something
> similar). If we just drop the requester into, say, the #ubuntu channel,
> then we will not have accomplished anything.

I was thinking somewhere in between. Dumping them into a generic channel is
definitely wrong, but I don't think doing individual pairing is necessary
either. As long as they're going to a channel a few levels down
(#ubuntu-graphics-ati for example) then there should only be a few people
looking for help and a few people helping at any time. Additionally, is it
possible to check the 'population' of an IRC channel? If so then the program
could do a check, and if no-one was there it could say: "Sorry, the
automated system couldn't solve your problem. Additionally, there is no-one
available for live help on your topic at this time. I have posted a message
to the Ubuntu support forums containing the necessary information to help
solve your problem, and you will be notified via email when a response is

An overview of the program flow would look something like:

1. Asks a few questions, similar to [1] to determine the area and type of
2. Scrape the related wiki pages (as well as prepackaged help docs) with
suggestions and related problems, and display them to the user in a nice and
friendly way (eg "Your problem may be [this]: try [this] to find out. If
[this] is your problem, do [this] to solve it.")
3. If none of the wiki helps work, the user clicks "My problem isn't solved.
Click here to get help from a real human being."
4. Run apport-collect into a nicely labeled package on the desktop, and
notify the user that "Some useful information for diagnosing and solving
your problem has been collected into file X on your desktop. Please keep
this file handy while working on the problem."
5. Checks the proper IRC channel. If there is someone there, open an IRC
client into that channel. If there isn't, go to 6.
6. Automatically post problem to proper forum, and attach apport-collect
package. Alert the user: "There is no-one available right now for live help
on your topic. Your problem has been posted to [give link]. You should get a
reply within two days. You will be notified by email when a response is

Again, as commented on this thread some times already, we at least need
> some basic requirements as far as *documnetation* is concerned:
> (a) comb through the forums, bugs, answer.launchpad, and others for
> issues worth being cataloged;
> (b) organise them in a more user-friendly search structure (subject
> matter, keywords, version, level of expertise required, etc, etc).
> (c) re-word and clean up (including adding a reference to the original
> source).

I've done a little looking, and we probably don't want to be using the wiki
for this at all. If I understand the descriptions correctly, we should
really be working on reorganizing and cleaning up [2]. Perhaps once [1] is a
little more complete, it could become the homepage for [2].

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the Ubuntu-devel-discuss mailing list