Maybe you're right. There were lots of performance issues that were easily solved, like: we were writing to PostgGreSQL for every single game action, our poly counts were in the several hundreds to thousands but could have been well under 100 in many situations, and we were always globally consistent but could have been eventually consistent (through geographic sharding and a "speed of information" data propagation rule -- since all cells don't need any game state from distant cells due to players having to be in a physical geographic location to play). Though I'm sure there were even more huge issues not yet anticipated.
Ultimately, failure was always the mostly likely outcome here, even if we had millions to spend.
The most unfortunate part for me is that we didn't ever get to learn if people actually liked or disliked the game, since we didn't release it outside our own circles.
Ultimately, failure was always the mostly likely outcome here, even if we had millions to spend.
The most unfortunate part for me is that we didn't ever get to learn if people actually liked or disliked the game, since we didn't release it outside our own circles.