Venture Africa Post Mortem, Part 1 (repost)

Venture Africa Post Mortem

I've just finished up Part 1 of my post mortem for Wildlife Tycoon: Venture Africa. This part focuses on Art, Programming, and Game Design. Part 2 will focus on Business, PR, and Team Management.

This was just intended as an internal document, but I've had a few requests that I share something like this, so here you go.

Wildlife Tycoon: Venture Africa was released on October 31, 2005. It was nominated for the 2006 IGF Gand Prize and the 2006 Slamdance festival. Part 1 of this post-mortem focuses on the technical development, art, and game design of the product. Part 2 will focus on business, PR, and team mangement. Items in bold are major take-aways from each section.

Technical
Venture Africa relied heavily on externally developed products. In particular, the Torque Game Engine, multiple 3rd party add-ons for the TGE, and Trymedia's ActiveMark technology for sales and security composed the majority of the technical content.

The Torque Game Engine is an excellent choice of 3D engine for a small developer. The game runs well even with large numbers of animals on fairly low-end machines. And Trymedia's ActiveMark system was very easy to implement, despite a few incompatibilities with Torque. I expect to continue to work with Torque and Trymedia's ActiveMark technology as they offer incredible power and ease of use at minimal cost.

From a programming perspective, the only major area of development was the artificial intelligence. The AI made the game fun to watch and interact with; it is what makes the product unique. Unique, interesting Artificial Intelligence is clone-proof, runs well on low-end machines, and is an unsaturated area of innovation in today's games.

The major technical problems encountered were due to externally developed code. The Torque/Flash integration for the user interface and HUD was buggy and incomplete. It did not save us time or money and the product was released with a number of bugs related to it, some of which persist today. Torque also required a little work to get the game to shippable state, notably its support for windowed gaming and the mac implementations of encrypted asset files. In the future, time should be allotted for polish (not just integration) of externally developed code, no matter how complete it seems to be.

The aforementioned bugs were not fully ironed out before the product went beta and we lost a potentially large amount of test data because of voluntary beta tester attrition. Because of lack of testing, the game was released buggy, and we enjoyed no release-week sales spike. Our problems with buggy releases could have been solved by fixing all of the known bugs before the game was released to beta testers.

Art/Sound/Writing
Making the game 3D instead of 2D may have been more trouble than it was worth. Plenty of downloadable Tycoon games were built in 2d and have been succesful. One wonders, however, if moving to 2D would cause us to suffer in comparison to our closest relative, Zoo Tycoon.

Art, Sound, and Writing were entirely contracted out to remote, part-time contractors. The user interface, fonts, and logos were very well received by customers and press. The animal models, while not as well received by the public, were appropriately low-poly, and allowed us to have large herds of creatures populating the levels.

Sound effects and music were similarly well-received, but were very sparse in-game. The in-game music added for the retail version of the game helped dramatically. We should strive to have a stronger aural atmosphere to our next game by including more music and stronger ambient sound effects.

Signature animated moments, such as the baboons climbing trees, delighted players. But despite having 10 unique environments, the levels tended to run together a fair amount, possibly due to lack of signature moments. Buildable environmental objects such as trees and bushes were the same in every environment, making each environment look even more similar. In the future, we should strive to differentiate animals from one another through more unique animations and sound, and we should reduce the number and size of environments significantly but populate them more densely with environmental models and effects.

This improvement, of course, will require more assets. Venture Africa had a very constrained asset list, which kept us on schedule and under our 20MB file size limit. The 20MB file size limit was not necessary in retrospect. This limit applies to casual games distributed from Yahoo and BigFish, but other distribution portals have games with up to 40MB download sizes. We should increase our file size limit from 20MB to 30MB in order to include more music, sound, unique animations and environmental models.

The story was told through animal-themed limericks on the loading screen for each level. This feature, while a pet favorite of mine, went under-appreciated by players. The limericks did not provide true incentive for the player to move forward: a traditional storyline would have sufficed.

Design
Everyone liked the concept of Venture Africa. Our download rate was high, we attracted the attention with a lot of press, and we were very in-demand with publishers. This is clearly an IP that merits another iteration.

The flip side is that the reaction to the actual gameplay was mixed, particularly with the trial experience. While the tutorial level and the second half of the game were fun, strategic, and replayable, the trial levels (not including the tutorial) were repetitive and simplistic. This is largely due to the animal unlock schema - players only started out with 2 animals and finished the trial experience with only 4 unlocked, though the game didn't get interesting until about the 6th animal. Since time-limiting is the most effective way to induce sales, the trial experience should include all the ingredients that make the game fun, and the remainder of the game should simply provide additional content.

Since the second half of the game is so much stronger than the first, we also needed to do a better job communicating to the potential customer what they are getting if they buy the full version. An attract mode would viscerally sell the game much better than a sales sheet ever could.

In comparison to the game's closest relative, Zoo Tycoon, we fell short on the information/education angle. The encyclopedia was not presented in an engaging manner nor was it accessible in-level. More information about the animals in general and in particular about the mood of each animal would have pleased the "management" gamers, though in leaving that information out we may have pleased the "casual" gamers and younger children. A random tip box and a better encyclopedia would help to provide a deeper glimpse into the strategy and a stronger educational angle, without sacrificing the simplicity and cleanliness of the game's presentation.

Some gameplay elements that were added as experiments were not that fun. "Be the animal" was a great sales feature but completely forgettable in game. Constraining the animal choices to 4 at a time and the shuffle button were also unsuccessful: the sandbox mode in which the player could choose any animal any time was far more fun. We should continue to experiment with gameplay elements such as these, but we should not hold onto ones that make sense from a game design perspective but just aren't fun.

Finally, the initial design goal was to make a Tycoon game that was targeted at the downloadable market. We were unsuccessful in the goal for a variety of reasons. Even though the camera was constrained, the 3D world was daunting for some. Also, our goal was to have each level take 4-7 minutes to play, but in actuality, the levels ended up taking 15-60 minutes to complete. For our next game, we either need to embrace the non-casual nature of the game and include features that fit that market (save/load being a key feature for longer games), or we need to scale back the scope of each level to attract the casual audience. This may involve reducing the size of the levels so that there is less camera manipulation necessary, reducing the length of time for each level, and removing features that don't easily fit that market, such as "Be the animal".