Bug Voting

[Eric Dunbar]

On 25/04/06, Sitsofe Wheeler <sitsofe at yahoo.com> wrote:
> On Tue, 2006-04-25 at 00:25 -0500, Jonathon Anderson wrote:
> > Would it be possible to implement a system of mod points where launchpad
> > users can "bump" bugs to higher priority with votes? I know this already
> > happens to some degree with dupes and comments, but if there was some
> > way to track how many users complained about a certain bug, it would be
> > easier to figure out which bugs affect the most people and thus which
> > should be fixed first (all else being equal).
>
> Bug voting sounds nice and fair but I simply don't believe it works
> unless it's intent is as a placebo to placate users (which is a fairly
> good point for implementing it in the first place I suppose...).

This sounds like a problem which includes the "elitist attitude" card
-- "the developer knows best" (and, no, I'm not accusing you of
holding it).

There is a pretty important piece of information to be had from a bug
frequency tracking mechanism (whether it is through voting or
otherwise) and that is how many people are affected by it.

And, the number of people affected also might, at times, be an
excellent indicator of where to put scarce volunteer resources.

Of course, as with anything "voting" would merely be a guide -- a lot
of people might see a problem with a menu behaviour but can still get
work done or a workaround is readily available. Far fewer might have
an issue with, let's say an SMB client used by Nautilus, but that
problem is a show-stopper. Without the fixed client they simply can't
get their task completed.

That said, using the preceding as an argument against listening to end
users in a systemic manner (more coordinated than through bugzilla) is
nothing more than an excuse for inaction and ineffectiveness.

Some control has to be relinquished by developers and it will be for
the better -- open source isn't just an exercise for and by developers
anymore. End users are now an integral part of the equation (even if
the end user is the developer or her clients) and there are any number
of reasons developers need end users as much as end users need
developers: end users pay some developers' salaries (if they're paid
like devs for Canonical or Red Hat or IBM or Novell or...); end users
might provide some devs with a feeling of satisfaction/need; end users
might provide (some) devs with fame and (even fewer) fortune.

One somewhat analogous situation is in the world of pharm and
biotechnology. Biotech and pharm companies research the sexiest and
most lucrative diseases and spend millions[1] on R&D trying to create
pharmaceuticals or trying to fabricate new concepts of what
constitutes disease (that warrant a patentable treatment ;-). The
overwhelming majority of research efforts target a _tiny_ fraction of
the world-wide disease burden. Instead, it's up to governments and
non-profit funding agencies to encourage and support research into the
diseases that afflict millions -- the majority (or even billions...
1/3 of the world's population carries TB).

[1] they might spend millions on R&D but they spend billions on marketing

Open source is a major deviation from the pharmaceutical R&D method
since it is not driven by the profit motive -- OSS doesn't have to
worry about devising strategies to make money or focusing on lucrative
research for people who can pay. However, a balance between focusing
on users' needs and the desires of developers needs to be struck. It's
a two-way street. End users need devs to write the software, and devs
need the end users to provide whatever it is they happen to provide to
the devs (whether it be salary, satisfaction, glory, principle, etc.).
Note: this is obviously not an exclusive model -- not every developer
needs an end user to write for.

> People think that if they can only get the bug voted on repeatedly
> something will HAVE to be done about which may not necessarily be the
> case. Just wait until a bug that's been hanging round for years has +100
> votes...

... and then try to explain that fixing a broken graphics widget might
be a lot less difficult than a broken network services client but that
the latter is more important in the grand scheme of things ;-)