What Went Right?

1. Small teams are much more productive
With only three people on the team all communication problems vanished, all three of us have been in the mainstream development industry for many years and even though it was expected it still came as a shock. With teams of 15-25 people a huge amount of effort is needed to track what everyone is doing, keep people motivated and maintain a singular vision of the game. With such a small team it was easy to keep the number of official meetings to a minimum. Instead of an entire afternoon wasted with no agreement, meetings would be over in minutes with everyone happy that their views had been aired.

All three of us had just finished a 2.5-year mainstream project for Xbox (Gunmetal) with an 11man team. Starscape was a revelation in comparison; it cost £50,000 and took 21man months. GunMetal on the other hand cost £2,500,000 and took 330 man months. Obviously GunMetal was a more labour intensive project featuring an FMV intro, 3D technology on a new and previously unknown platform. However it shows dramatically there is an exponential relationship between increasing the content/technological scope of a game and the resources needed to realise it. Is Starscape 16 times less of a worthwhile gaming experience for the player than Gunmetal? It's a fact that 16 times less resources were expended in producing it. Slightly vacuous question, but an interesting one never the less, in theory you could have had 16 Starscape type games for the cost of Gunmetal.

2. Small projects keep moral and creativity high
Normally a development project will last 2-3 years and each person will only be responsible for a very small part of it. In this situation its easy to see why a team's moral can drop from time to time and why sometimes a team will seem to lose the direction it started with. Initially we set ourselves the goal of writing a commercial quality game in six months, a daunting prospect. One of the unforeseen side effects was the joy of seeing the game develop and take shape so rapidly. As there were only three team members each had to take a large responsibility and play a significant part in how the game turned out.

It was actually one of my happiest development experiences to date, even with five previous mainstream releases in the last 7 years. With Starscape, new features would go in on a daily basis, the game was playable very early on and new ideas could be tried out within days instead of months.

3. Online community of beta testers
In mainstream development paid on-staff testers are used, quite often they are kept in a dark room out of the way and paid very little. They tend to follow strict testing plans and will repeat the process on the same game over and over for anywhere between 2weeks and 3months depending on the complexity of the game (i.e. how many bugs it contains). As a machine for finding bugs it is a system that works very well, but for creative feedback it tends not to be as effective. Testers are often housed elsewhere in the building, there can be layers of managers separating testers from developers, official channels of communication and a general feeling of apathy as they spend long hours ticking boxes on test plans (shame really).

Starscape used public testers brought in early with their own private forum on the website. We could reply directly to their comments and an equal number of design ideas and improvements populated the boards alongside the more predictable bugs. The small scale of the development meant we could turn people's suggestions into prototypes very quickly. There is a natural tendency for people to get excited about different aspects of a game and then just lose interest because it takes months to develop prototypes to explore their ideas. We could turn things around in days and this helped keep the testers motivated.

4. Small game with big ideas
I consider Starscape to be a 'complete' game in miniature, by this I mean it contains all the expected elements that go to make a well-rounded mainstream game. People have come to expect what I would call a shopping list of features, obviously some games by their nature have a different looking list, but here is my Starscape list:

• Primary compulsive game play mechanic
• An intro and outro animation
• A well developed back story and plot
• Well rounded characters to bring out the back story and plot
• Periodic cut scenes
• Periodic rewards for play i.e. player advancement, level bosses, progress screens
• Good music and sound effects
• A visual feast, games are visual entertainment after all
• Multiplayer

Ok so we missed out on multiplayer, but we certainly got every other element. Obviously we don't match mainstream titles in all elements, that's not the point, what we did match and in some cases exceed are people's expectations. All too often indie games are lacking in one or more areas and the public shouldn't put up with that.

5. Player orientated design
Long-term developments with on-staff testers tend to result in polarised game content. The only people playing the game are the development staff and they have been playing it so long it's almost impossible to put themselves in the shoes of the newbie. One of the most obvious ways this manifests itself is in the difficulty level, which tends to be enormously high, but it also causes more subtle problems. New players find it difficult to understand what to do next or what is expected of them. Focus groups help but the expense means they are few and far between, when reviewing these sessions developers tend to vent their frustrations loudly "what is that guy doing, you aren't meant to put that there, it's obvious".

Playing the same game day in and day out for two years can result in a kind of mental conditioning that is hard to notice. The game seems fairly well balanced, obvious and understandable, but is it? We found having a large online community of testers with new members coming in all the time really helped overturn our assumptions about what people 'get' and 'do not get'. It was always a surprise, and catching these problems early allowed us to rework elements. Obviously security was a major concern, we always made sure the demo versions were well marked as demos, timed out within two weeks and later we even locked them to the testers PC using an online registration process.

6. Open source middleware
Starscape uses OggVorbis for music encoding and playback, SDL for OpenGL support on Mac. Both libraries are free to use (under the GNU licence) and work on PC, Mac and Linux. In the past we have used commercial middleware with $50,000 price tags and so were worried about compatibility, support and overall quality of implementation. We were pleasantly surprised, both libraries proved to be excellent and are highly recommended.

NEXT > What Went Wrong?