[gentoo-project] Re: Plan, then communicate (no-list)

[Steve Long]

Alec Warner wrote:
> The community currently has no good means to rank problems in the view
> of users other than the forums; which currently have their own issues.
>
> User Coverage: Not everyone has a forums account.  Not everyone uses
> their forums account.  We have no idea how many users we have
> (ancidotal numbers suggest ~200000; see
> http://dev.gentoo.org/~antarus/bouncer-stats.txt).  It is difficult to
> know what percentage of users responded and thus becomes difficult to
> judge how important something is (we have only the respondants data to
> use).
>
> Arguably you could say that anyone who didn't vote doesn't care; but
> you have to factor in people who didn't learn of the vote during the
> voting period.
>
I could and I would; forum users see their accounts similarly to how gentoo
devs view their @g.org badges. Spammers are soon dealt with, so that anyone
who has posted more than 50 posts (not counting OTW) and been a member for
more than 3 months is not a bot. Let's say there's 100,000 active users. In
a poll say 5% of votes are fraudulent. It can be factored into the
calculation of significance. And when 90% say they don't like the way they
get treated by Gentoo devs, there is a real issue. Call it communicaton,
call it what you want, it's a real and valid concern.

And IMO it holds Gentoo back.

> User Education: This is that whole Cathedral thing.  Below I'll talk
> about Daniel's goal of maximizing developer impact and this plays a
> big part.  Many developers don't talk to users because its draining
> and they want to work on projects that they have a high impact on.  I
> could sit in #gentoo and field questions all day (I've done it before)
> but I have things I could spend my time on that are more worthwhile to
> the project (and we are lucky enough to have a crack team of awesome
> contributors that staff that channel).
>
There's also developer education. A junior dev (aka code-monkey) who comes
in at the start of their career is not expected to show the same level of
maturity as a 40 year-old.

> Talking to users is exhausting when the user really has a
> misconception about a given problem, program, or feature.
Yeah it's called requirements analysis (whichever model you use.) That's why
it's such a source of problems.

> It takes
> time to educate people why something works the day it does and
> documentation only helps so much.  Give bad service and the user is
> off to the forums to complain about how he was mistreated by that
> Antarus guy on #gentoo-portage and how much Gentoo sucks.
>
So allow more advanced users to help the less knowledgeable. All your doing
is formalising what happens on irc.

> That being said; talking to users who know what they are doing (doubly
> so when they know more than me) is a delight and I'm generally happy
> to take the time to respond.  If there was some way to aggregate user
> complaints into concrete problem sets I'm all ears.
>
Votes. Require 75% majority from users if you want.

> User Validation: Most systems that users can use to respond on a large
> scale don't have a means to validate whether they use your software or
> not.  This is more of a trend game; needing to look at the aftermath
> of any given aggregate data and look for areas where people may have
> given feedback that we should throw out (like automated voting).  I
> don't think this problem is necessarily solvable or that big a deal
> but it is something to consider/
>
Yeah see above about statistical significance. We're not looking for a 5%
end of the normal here (which is what could perhaps be used to identify a
minority.)


-- 
gentoo-project@l.g.o mailing list