On 8/12/2011 17:15, Simon Phipps wrote:
On 12 Aug 2011, at 22:01, Rob Weir wrote:
On Fri, Aug 12, 2011 at 4:30 PM, Simon Phipps<[email protected]> wrote:
I suggest the right question is "which project members need which data and why". The
answer today may well be "none", since we don't actually have any resources to visit yet.
This is also likely to change over time, and we'll need to add analytics as and when people request
(and justify) according to their specific needs and remove them when they're no longer justified.
I suggest we resist the idea of capturing bulk analytics "just because", and
instead devise a lightweight process for justifying and requesting collection of data.
I'd guess there is already a process to copy somewhere in Apache - any mentors with
suggestions where to look?
I see that tracking code is used with the websites of most of the
groups you are affiliated with: LibreOffice (Piwik), ForgeRock
(Google Analytics, including in the community pages) and OSI (Google
Analytics). And as was mentioned before, OpenOffice.org uses Google
Analytics currently.
It is indeed endemic. We have a unique opportunity to address the issue
thoughtfully. And by the way I am delighted you're paying such close attention
to my career.
Have you given them similar advice?
Where possible, yes. Abuse of personal data is something which concerns me
greatly.
Or is there something special
about OpenOffice at Apache
Yes. At the moment as far as I am aware AOOo has no significant resources of
interest to non-project-members and no groups of members with active
applications for the data from analytics. Both situations will certainly
change, but on best YAGNI principles I suggest doing what's needed when it's
needed, on the basis of actual documented requirements.
that suggests that we should not be
optimizing our website based on visitor stats like others, including
LibreOffice, are?
When we have an end-user website capable of optimisation along with project
members stepping forward to harvest the data, process it and act upon it, that
will be a fine thing to do. What's needed, when it's needed.
S.
Two points, not definitive but worth considering:
1) Technically, is it easier to build in analytics now, or would it be
just as easy to add them later?
2) Might we want to do something dramatic (<sarcasm> say, actually make
a release... </sarcasm>), and measure the effect on the site? Suddenly,
months of un-analyzed data become a valuable baseline.
$0.02
--
/tj/
