On Sat, Nov 14, 2009 at 9:41 PM, David Gerard <dgerard(a)gmail.com> wrote:
that we
don't have to speculate; we can run all the ones that don't seem
blatantly counterproductive, and find out how well they do. Even better,
It is not in fact that easy - because every slogan has to be
translated into a pile of languages (by volunteers), and every banner
has to be tested thoroughly in all translations (some this year broke
in IE6/7). So there really isn't that much room to move. Erik Moeller
----------
~~~>>We don't have to speculate if this bacteria can live in low pressure;
we can try it.
~~~>It's not easy at all! It will cost trillions to lift the test apparatus
into orbit and get it onto a path to mars. Then we have to consider the
cost to get the scientists there!
~~~Um. Why not try it in a lab. On Earth.
---------
Er. David. Why not test the messages to the extent that doing so is easy,
and leave the trouble of the full deployment for messages demonstrated to
actually perform?
It's also pretty likely that different messages will perform differently
in different places. If we were really getting smart about this we should
be predicting the most likely clicked message based on which *article*
the reader is viewing (based on its categories). ... tracking it separately for
different *languages* is an obvious first step.
Make it possible to request a particular entry with the right url while
it is out of rotation... and put up a browser compatibility checklist.
Let the community get test coverage if they want their messages seen,
they could do worse than wikimedia's prior "well it worked in safari"
testing.
And, frankly, if the message is only shown to 1:1000 viewers it would still
get good click-testing coverage but if it caused some minor hiccups in corner
cases the world would not end.