We have a skype chat where a whole bunch of the ex-Zynga senior technologists (we called them CTO’s) get to chat about stuff we learned.
One thing that came up was our attempts to come up with web-mobile-backend cross platform technology. As a key driver of this effort, I learned that this was a misguided attempt.
Because I like the immediacy of the dialog I am including the skype chat:
 kostadis roussos: if I was going to enumerate my biggest professional mistake it would be to completely misunderstand the mobile gaming market need for specialized gaming engines targeted at specific games.
 kostadis roussos: The reality was that I didn’t understand the importance of having a hyper-optimized game for a specific platform and that it was more important to have the BEST game implementation than leverage.
 kostadis roussos: That leverage was basically unimportant – as a gaming company – … What was important was delivering the best possible user experience.
 XXX: it’s a fundamentally hard problem to solve and i think what I took away is being polyglot is preferable
 XXX: it seems like you’re really just talking about the failures of flash/html5 on mobile, not so much a “specialized gaming engine” since cocos2d / unreal / unity are all cross platform
 kostadis roussos: @XXX – intriguingly in 2013 when we did a whole bunch of analysis 8/10 games had their own custom engines.
 kostadis roussos: Maybe the data has shifted – and I’ll accept that notion but you really need to have a highly customizable engine where you are not giving up any performance or user-experience .
 kostadis roussos: My point is that I completely misunderstood the centrality of that point. I didn’t get that we need to think of a game as CMS with a very optimized and specialized “rendering” system and that we needed to have the best one for any game genre or we would lose.
[ kostadis roussos: And so I pursued – html5 as a solution, flash as a solution and then PlayScript language we built – all of these were attempts to solve the wrong problem. We should have been building the best UX system for I/E games in native languages with some C++ for the business logic bits that could be shared across multiple platforms.
 kostadis roussos: Oh well. Lesson learned.
My only personal consolation was that I gave the prezi guys some good advice based on my learnings. They feel good about the outcome.