David Gerard wrote:
2009/11/15 William Pietri
<william(a)scissor.com>om>:
[...] I'm just saying
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,
we can automatically optimize which we show and how. [...]
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
posted about this on the wiki, explaining in detail what's going on,
but I can't find the link right now.
Just to be clear, I'm not suggesting this approach for this year, but
for future years.
As others have pointed out, you could do this incrementally. If a slogan
doesn't test well in its native language, there's no reason to go to the
trouble of translating it. And even if it does, it may not be
particularly translatable, so I think translating is something could be
optional, based either on central expert judgment or the individual
judgment of volunteer translators.
As to browser testing, I think we should be able to abstract formats
from content enough to make sure that the containers get tested
thoroughly, while providing enough flexibility to those suggesting
content. Quite a lot of people have the skills to come up with a clever
line, a clever image, or a basic mix of the two, and those formats would
be easy to constrain and browser test.
That would leave out some of the more esoteric format possibilities, but
I think that's ok; this would still be a step forward in terms of
engaging the community and the wider public, and I suspect ad format
innovation is anyhow not something where we want to be on the forefront.
William