Developing Games for Android

The advent of smartphones has opened up an entire new range of platforms for game development, particularly for independent game developers. The iPhone's App Store model has been enormously successful, with a recent estimate of $2.4 billion in annual sales.

From a developer's perspective, app stores have the incentive of a distribution channel with powerful search and marketing capabilities. They also provide an intermediary that handles the logistics of selling and downloading -- generally for a cut of 30%.

You've probably heard the Cinderella stories of developers like Steve Demeter, whose triangular variant on match-3 games, Trism, reportedly earned him $250,000 in just two months in the iPhone App Store. Take also Ethan Nicholas' iShoot, which reportedly earned $600,000 in a single month. With the gold rush mentality and the iPhone's undeniable success with the App Store, why would a game developer choose to make a game for any other smartphone platform?

About six months ago, I started developing games for Google's Android OS, somewhat by chance. A friend who works for Google gave me a G1 developer's phone as a gift. At that point, I hadn't even heard of Android. In the process of learning about the OS, I came across the Android SDK, and downloaded it to give it a spin. I was instantly hooked.

While iPhone apps are written in Objective C, the Android SDK uses relatively more programmer-friendly Java. The iPhone store charges developers $99 a year to distribute their apps, while Android has a one-time $25 fee for developers. And the review process for iPhone apps grows increasingly lengthy -- sometimes weeks or more -- and it's somewhat arcane. Android apps go live as soon as the developer hits the publish button. Google handles the review process post-hoc, and is much more lax in terms of content.

With the greater ease of producing and distributing apps through the Android Market, why aren't game developers large and small clambering to produce quality games for the system? There are a number of reasons, but number one is, of course, money.

Mobile ad company AdMob released its July 2009 report (pdf) which notes the following:

So a major part of the equation is pure size. But another factor is the population makeup of Android users and their perception of the platform. The iPhone user population is likely much more diverse, with a greater cross-section of users representing more diverse demographics, while the Android user population is still likely predominantly tech-centric early adopters. Since Android is an open-source system, the perception may also be that the software that runs it should follow the same model. Thus many users may be reticent to spend money on apps.

Of course, the hope among Android game developers is that an influx of new phones on a variety of new carriers will increase and diversify the user base, increasing sales. However, current reports are gloomy. One case study is Larva Labs, a New York based mobile game developer. Larva recently shared its sales stats for August, and they weren't pretty. Despite having two games in the market that both spent time in the top 10 among paid apps in their categories, the daily average income for August was $62.39.

The developer laments issues with the market interface (poor organization and search) and a lax return policy (24-48 hour returns, no questions asked). But the issue of piracy is also raised, and is another likely source of poor performance of games in the Android Market. A cursory search of popular torrent sites reveals bundles of Android games for download.

The next release of Android is bringing much needed revisions to the market interface. As for piracy, as with any game market, it is difficult to determine the impact. The Android Market includes optional DRM, but it is unlikely to slow or deter would-be pirates.

Unlike the iPhone, an Android device does not need to be jailbroken in order to allow installation of pirated apps. A user can simply install pirated apps onto their SD card and use one of the freely-available app installers from the market. The return policy even allows a would-be pirate to download an app, make a copy, and then get a full refund. They don't even have to pay for the original copy they pirate!

One viable way to mitigate the effects of piracy is to monetize apps through advertising, rather than relying on direct sales. About half of the revenue from my games has come from ads as opposed to direct sales.

Despite the seemingly poor state of the Android market, the Larva Labs developers remain upbeat about the potential. The mobile analytics company Flurry notes the growth of new Android projects next to the relative decline of new iPhone ones. The AdMob market report mentioned earlier also notes that ad requests from Android have now edged out requests from Windows Mobile, both internationally and domestically. While iPhone is still the king of the hill, Android is pushing its way into the market.

There are currently only two handsets that run Android available in the U.S.: the T-Mobile G1 (aka HTC Dream) and the T-Mobile myTouch (aka HTC Magic). The G1 has a hardware keyboard, while the myTouch does not. Other than that, both devices have the same screen size, and so far cross-device development is still pretty straightforward. I've heard that both HTC and T-Mobile are pleased with the sales performance of both devices, though I'm not aware of hard figures. Based on analytics from my own apps, about 20% of my existing user base own the myTouch.

The good news: a spate of new devices are on the way. Motorola just unveiled its Android-powered Cliq, which will soon be available through T-Mobile. Another Motorola phone announcement is expected soon. Earlier this year, the estimate was that 18-20 Android-powered devices would be released by the end of the year. That remains to be seen, but there it is extremely likely that by the holiday season there will be at least half a dozen Android smartphones, probably on multiple carriers.

Things could get very interesting with this new wave of devices. The myTouch alone hasn't changed the game, but the incremental influx of new hardware, especially with the openness and customizability of Android, and choice of carrier, could start to see Android carving out a significant portion of the market.

On the one hand, that seems like good news to Android developers. An increase in the market can only be good, right? Well, an open issue is one of device compatibility. The current HTC devices have a trackball, while other devices may or may not, opting for a touchpad or alternative input device. Another issue is screen resolution. The current devices all have a 3.2 inch screen with a resolution of 480x320, but newer devices may have resolutions that are smaller or larger (e.g. there are already tablet PCs running Android slated for release).

In terms of RAM and processing power, there are not significant differences in the devices rolled out thus far. Although continued improvements and custom builds of Android can get the most out of existing specs, the newer generation of Android smartphones -- those poised to be released by the end of this year -- should have more RAM and more powerful chipsets. So building more CPU-intensive games for the platform will only get easier, but the decision to do so may become more difficult.

One advantage of programming for the iPhone is not having to worry (much) about hardware compatibility. This could potentially be a problem for Android game developers as new handsets emerge. The possibility of a backlash exists if users buy a new Android smartphone only to find that the most popular games in the market don't render properly on their screen, or that the control scheme for the game doesn't work. A developer can try to work their way around these issues by designing for multiple screen resolutions and including multiple input methods, but this of course means extra work.

Other issues include the limitations of the first-generation hardware that's currently running Android. Specifically, battery life and storage space are both big issues on the G1, and games can tend to hog both, especially the more features they use. I just finished a game for the second iteration of the Android Developers Challenge -- more on that in a moment -- and it uses the GPS functionality of the device. Unfortunately, the GPS finder is a huge drain on the battery.

As for storage space, the G1 only has 70MB of storage allocated for apps. That means designing lean games in terms of music and art assets, since users may not want to delete a bunch of existing apps to make way for your 10MB masterpiece. The newer generation devices have several times the capacity, so in a sense these problems are being resolved as new hardware rolls off the shelves. But the limitations are in place for most of the current generation of users and developers.

Likely because Google is behind the curve in the smartphone game, it has instituted a strategy designed to lure developers to the platform with the promise of cash -- not in the form of market profit, but in prize money. Last year Google held an open contest for the best apps, the Android Developer Challenge, a free-for-all in which candidate apps were selected by a panel of judges.

This year, it kicked off the Android Developer Challenge II, with $2M in prizes. Rather than lumping all entries into one bin, ADC II is divided into 10 categories, including two for games (Casual/Puzzle and Arcade/Action). The deadline for submissions closed on August 31st, and winners are to be decided by two rounds of judging: one by users and one by a panel of judges.

However, If you hadn't heard of the challenge, you're probably not alone. Google didn't do much to spread the word, probably assuming that the lure of large cash prizes would carry the news effectively. With the overall winner set to take home $250,000, the strategy seems sound. But the jury is out on whether or not the ADC II has been effective in drawing game developers to the platform. The first contest had about 1,700 entrants, but no numbers on the second challenge have been released as of this writing.

I spent the summer working on an entrant into the ADC II, a puzzle/RPG hybrid along the lines of Puzzle Quest, with the twist that the player must physically visit locations near them, such as coffee shops and grocery stores, in order to unlock content.

Developing the game presented some unique challenges, learning how to incorporate the Google Map API, for example. And some aspects of development still need a lot of work. For example, the functionality for playing sounds and music is extremely limited and has been historically buggy. However, overall the experience was very rewarding, and grassroots developer communities provided answers to nearly every issue I encountered. In particular, the sites anddev.org and the Android Developers Google Group have been helpful.

Still, there are existing issues that are problematic for the platform. There are quite a few games for the iPhone that utilize an underlying physics engine, but I'm currently not aware of any games for Android that are physics-based. Ports of some existing 2D engines to Android have apparently not performed well enough to use them for games. Again, as with the other issues related to developing on first-generation hardware, newer devices will probably render them obsolete.

For now, if a developer decides to implement a game exclusively for a particular smartphone platform, and the choice is between the iPhone and Android, the tradeoff is between trying to get noticed in an incredibly crowded and competitive market where the potential payoff is huge for those at the top, or entering a market with low barriers, little competition, currently low returns, but the possibility of potential growth. Hopefully there will continue to be developers that take both roads, and the smartphone market will continue to grow as a diverse outlet for game development.