Thank Branko, nice blog and very interesting to see the differences in the datasets and the underlying assumptions of each. I always considered the Maddison numbers to be slightly too high for non-Western countries and it's interesting to see the role that urbanity plays in that.
I recently made an infographic on this topic actually, glad to see that world trends in global economies are interesting to the World Bank! The Maddison project has such a great advantage over the other dataset as it places a value on history and trends which are key to understanding the time we are in now. By only providing GDP figures from the 20th century, you get a really partial view of the world and it normalises the financial dominance of the Western world. The Maddison project shows that the last 160 years have been completely bizarre and beg the question of why China/India weren't the richest countries in the world over the past century and a half.
Take a look at the charts if you're interested: http://infogr.am/Share-of-world-GDP-throughout-history/