[Fwd: Re: Debian Package Popularity Contest - Turnout]

[Ellen Reitmayr]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


Hi,

Thomas noted that Debian has a package called 'Debian Package Popularity
Contest' which logs which packages are installed on a system and sends
an email. The user actively has to install the package, but then he is
no longer reminded that this data is collected and sent to the project
admins.

They have quite a high turnout (see http://popcon.debian.org/). I asked
them about their experiences to see if users are generally willing to
reveal such data.

Voila:


- -------- Original Message --------
Subject: Re: Debian Package Popularity Contest - Turnout
Date: Fri, 23 Sep 2005 16:19:53 +0200
From: Bill Allombert <allomber at math.u-bordeaux.fr>
To: Ellen Reitmayr <ellen.reitmayr at relevantive.de>
CC: Avery Pennarun <apenwarr at debian.org>, Bill Allombert
<ballombe at debian.org>,   Petter Reinholdtsen <pere at debian.org>
References: <4333C7CA.8050009 at relevantive.de>

On Fri, Sep 23, 2005 at 11:15:54AM +0200, Ellen Reitmayr wrote:
> I'm a usability girl from KDE and OpenUsability.org. We've thought of
> integrating feedback dialogs and tools to collect usage  data (what menu
> items, config options etc are mostly selected) with several KDE
> applications. However, we are not quite sure if the broad majority of
> users will accept those mechanisms and if they will take part.
> 
> I've had a look at your statistics and I see you have a very high
> turnout. Could you please tell me a bit more about your experiences with
> the popularity contest?

Hello,

First my (too quickly written ) advice:

You absolutly need to start with a privacy model and _never_
break/change it. The users must know exactly what data will be reported,
how they will be handled and what will be done with the stats.

Do not try to to report too many data:
1) users are more likely to disagree with that,
2) it will be harder to handle them
3) it will be harder to give meaningful statistics.
4) they can be used by malicious entity to break security/privacy.

so keep it simple stupid

> If you have information on this, please tell me:
> 
> - - Are the participants rather power users (who also use other feedback
> channels such as mailing lists) or do you have the impression that by
> means of the popularity contest, you also reach users you ususally do
> not reach? (the latter is our main goal in KDE)

All the data we have about our users are the report themselves! You are
free to interpret them.

Experience show that users do not know about popularity-contest
and so don't install it. However if they read about it in a mailing list
or a news site, they might decide to install it.  For example, the two
peak in the graph are the time when popcon was slashdotted.

> - - The emails are sent automatically. Did anyone comment on that, e.g.
> said he'd rather like to take control by manually sending emails?

They can remove the cron job, if they wish, running it manually, etc.

> - - Is the user reminded that popularity contest is still running?

No. But technically popcon is not running. It is a script that is
triggered by a weekly cronjon. It is not a long running process.
I think it is best if user forget completly about it, this remove
the temptation to bias the result.

> - - How are the drop-out rates?

I don't think anyone remove popcon, but sometime the computer is put
offline, reinstalled, etc so the box no more report.

> - - Have you received any negative feedback by users who complained about
> the 'Install debian popularity contest?' dialog during the installation?
> (Do you have the impression that offering the feature itself is
> controversial?)

Debian Sarge does not have this dialog, so I cannot comment.
I never remember complains about the feature itself.

Cheers,
- --
Bill. <ballombe at debian.org>

Imagine a large red swirl here.


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFDNBQN2mIgQiX+cvcRAilcAKC8cpLWLdJ8LVKy8e65CHyKpfMugwCgo9pZ
2yYdWbHIgV0IOhF7NaHcVns=
=oQH6
-----END PGP SIGNATURE-----