[koolkao]

"3. The a priori likelihood that your site will be popular enough to demand the scalability GAE can apparently deliver is very low, indeed nil. Are you willing to pioneer on an inconsistent infrastructure when you could use a proven relational one?"

The perceived probability of "getting popular enough" is for most start-ups non-zero. Else many start-ups would not even exist at all. That is, the expected value of the entire start-up endeavor, E(start-up)--say in the unit of dollars--is driven by these rare, but extreme outcomes.

Given this is the case, my impression is that: P( failure to scale | RDBMS ) > P( failure to scale | BigTable/simpleDB ) (that is, it is much more difficult to scale up RDBMS than BigTable/SimpleDB)

and P( success | failure to scale ) is near 0,

then it makes sense to prepare for that possible rare outcome of having to scale, in order to preserve that expected value.