Wednesday, December 2, 2009

iPhone TouchPets post-mortem talk at PAGDIG

Last night's PAGDIG meet-up had Andrew Stern (Catz, Babyz, Facade) of Stumptown Game Machine give a talk about the development of TouchPets Dogs for iPhone. Good talk and I took some notes. Here they are.

TouchPets Dogs is an iphone-based, modernized take on the 'pet simulator' genre that Stern gave birth to with Dogz and Catz back in 1995. Not surprising that when NGMoco wanted to do a game along these lines, they came to find The Man :-). The game uses a business model that is kind of a hybrid of pay-to-play and pay-for-upgrades, and indeed they are evolving the business model on the fly based on performance. It also uses gesture input, accelerometers and all the usual iPhone platform candy.

Ok, so, notes [with added commentary in braces]

  • Probably among the largest and most ambitious iPhone projects done to date. 12 months, 5-6 people full time, 8-10 people part time, plus 4-6 people part time at the publisher [using a rate for a mid-range studio of 8-10k/man month, this ballparks the development budget at somewhere between 850k to 1.3M(!) not counting NGMoco's costs, marketing, etc. That certainly is higher than even the high-end stuff we're seeing on iPhone today, that has a feel of "few hundred k". Even if the studio was cheaper than my estimate, it almost certainly wasn't under, what, 700k?]
  • This was about 2X the scope and time of what was originally proposed, as publisher kept growing scope and ambition of the project. Server complexity, social elements, in-game transactions...
  • Great things to say about NGMoco as a publisher. Supportive of them doing what they wanted in the game, kept them funded as scope grew, great relationship with Apple, etc
  • The game is a pet simulator, but has heavy focus on stats (to involve more 'gamer types'), careers, stories/missions, a social network, facebook connection, inter-pet relationships between players, in-game transactions.
  • 850k people have downloaded and connected. Peak server load has been about 25k people. Game is only periodically connecting, so that means "some number more than 25k" playing simultaneously [100k?]
  • Dogs go to sleep if not fed. Need to buy bowls of food to keep them away. Amounts to pay-to-play. Some user backlash to this, looking at maybe shifting toward free to play (and keep playing) but premium items/missions/etc are for pay.
  • Push notification if your dog gets lonely [does this translate to "come feed me money!" :-)]
  • All attributes to cost, rate of decay, etc, etc are all on server, so they can evolve over time despite clients in the wild.
  • Uses NGMoco's Plus+ network, which was good to get a community aware of the game and quicker to connect.
  • Online infrastructure complicated and tricky. Communications between their server, Apples for appstore/transactions, NGMoco's Plus+network. As scope grew, server grew wicked complicated (e.g. needed to do sharding, manage issues with players with 500 friends inviting them all for playdates, etc)
  • Graphics: All in OGL 1.1, no realtime lighting, 3000-5000 polys/frame max, 2 textures only. "I think iPhone is more powerful than the Wii"
  • Used no engine, but lots of sample code from PowerVR SDK
  • Can't mix all Apple's really good UI with OGL, so if you want UI in your game, have to build it yourelf [seems like a middleware opportunity here. Do a exact copy of all Apple's UI functionality in GL]
  • Did some easy physics (ball collision, etc). Cartoon physics: Throw frisbee off left side of screen, it wraps around and flies in from right. Move viewpoint over to where the wall is though, and THEN it collides with wall rather than wrapping.

What went wrong:

  • Product spec always changing
  • Complexity of system grew beyond means of core team
  • iTunes rules and constraints - moving target plus they were pushing the envelope here*

* [lots of questions and talk about this afterward. One of the challenges being echoed from XBLA, then iphone, and now Facebook. High dependence on single gatekeeper, with no commit from gatekeeper on how policies/APIs will change, whether notice will be given, etc. People are betting their companies on stuff that can be pulled out from under them with hous notice]

What went right:

  • Good team
  • Just enough time, budget, freedom given by NGMOco to actually build a great game
  • iPhone as a platform is wonderful. great simulator, powerful, somewhat challenging to fit everything on a small screen

It was a good talk and we went over to Stumptown's studio afterward for a release party, complete with snack foods served on dogfood bowls. Woof!

1 comment:

Alice said...

Very useful, this stuff. Keep it coming :)