Discussing the XBox One's cloud claims with Jon Blow. Spoiler: it doesn't matter.
TL;DR: The XBox One cloud is a PR stunt. There's very little incentive to use MSFT's cloud offering in games that also need to work on other platforms. Likely, very few games will do so, and you can anticipate bad user experience for those.
Recently, I poked Jon Blow on Twitter via the Indie TL:DR account, that I curate. I had seen him complain a few times about how the 300k servers available for the XBox One claim seemed ludicrous and I was getting tired of it. Here's a transcript:
BREAKING | @Jonathan_Blow learns that Google has over 2 million servers and that 300k is nothing. Stops whining already. Everyone breathes. - indie_tldr
Sure, but those servers are responsible for most of Google's revenue. It's, like, the entire company. Xbox Live is a small fraction of Microsoft's revenue.
My response: It's not about XBox live, it's about running half of the games on the cloud. Which is admittedly idiotic. That said, given Microsoft's reach, 300k is not a hard claim to back up. Whether they get used is another story.
Jon's worried about costs - and he's right:
Sure, but where does the revenue to maintain those servers come from? Either Live subs or game prices... with corresponding ramifications.
Agreed. There are several interpretations here - Personal thought is that they're basically saying: we have up to 300k server available if any game blows up - they sure don't want people making the link between "game computations in the cloud" and the Sim City fiasco.
As for the cost of keeping the servers running:
- on a large scale it's probably less than expensive you/we think, and
- they might leave the cost of servers dedicated to game X up to the publishers
You have two models: subscription games (mmo, etc.) - that works. And then you have expensive titles with short lifespans. That works too because you can just allocate many instances on the fly as success comes, then free them later. Of course, MSFT might play with things like commitment etc., make game studios play the cloud billing game.
But more importantly, who today can afford to develop on a platform so radically different than all the other ones? As far as I understand, PS3's 8 cores and weird arch was already "hard". But cloud AI/physics/etc... up a notch.
Jon hits the nail on the head:
It's not just that it is hard... it has extremely limited applicability. Most of the time it would make games worse.
That too. Game devs clearly won't fall for it. Who is it supposed to fool? The press?
Andy Lancaster responds with a probable truth:
I’ll tell you who, the public. Guy I work with is totally pumped because of “the cloud” he can’t say why though.
And that, ladies and gents, is why this announcement was hot air. It's impractical even for the bigger studios, and it's just meant to make it sound juicy. Which it isn't.