On Wed, Dec 14, 2011 at 10:45 PM, Bala Jeyaraman sodabottle@gmail.com wrote:
So stating the truth is being unconstructive?. You could have rolled this out in phases or as a "dark launch" as you are contemplating now, done more testing and saved us all this bitterness and trouble. Instead you decided to go live on a massive scale without adequate testing in live environment being accessed by millions of people everyday.
Although, the largest of the Indic language Wikipedias gets about 1/900th the traffic of English Wikipedia in traffic. [1] I think it's appropriate to be a bit more experimental at that scale. It remains to be seen whether we can get sufficient testing done via opt-in methods and dark launches, although I'm certainly strongly encouraging experimentation with both.
Your comparison with top web properties is flawed alongside another dimension: no other top web property has to solve the kinds of problems we're trying to solve here with a staff of three internationalization developers, one product manager and no QA team (!). Although we're larger now than a year ago and the year before, and we're still adding engineers and support resources, we have lots of big problems to solve in parallel. We don't have the luxury of being able to do just one thing extremely well, much as we might want to.
On the other hand, what makes Wikimedia unique is precisely the fact that we're all working together to continually improve and fix things. In the best case scenario, that's a happy partnership. And as you well know, folks like Santhosh, Siebrand, Gerard, Niklas and Amir have made tremendous contributions as volunteers, well before being on staff. So we should strive for being collegial and forgiving. I have a feeling that we'll all be doing this for a long time, so we might as well try to like each other. :-)
I applaud the i18n team for being bold and pushing things forward, while also reflecting on the process at this opportune time. I'm also grateful to you, to Srikanth and others for being both supportive and critical voices along the process, and getting very involved to fix things. Improving language support is a huge, daunting undertaking. WMF is only one player in this and it'll require continued, joint effort from all fronts.
All best, Erik
[1] http://stats.wikimedia.org/EN_India/ReportCardIndia.htm