11th October 2005
Cititel Mid Valley
Kuala Lumpur, Malaysia

Drat! That was my first thought when I saw the breakfast served at the hallway. I should have known that there would have been at least some sort of breakfast, and immediately regretted having my breakie at a local 'mamak' restaurant in SS2. Registration was easy and fast. Signs were clearly shown and there were people aplenty that were more than willing to guide you to the main conference area (which was composed of two adequately sized rooms, which could fit around 40-50 people).

The game development industry is still in its infant stage in Malaysia (and most ASEAN [Association of Southeastern Asian Nations – Ed] countries for that matter), and it was with some surprise that I noted the sizable crowd (around 30? excluding the AGDS staff) already chatting and exchanging cards at 9:30 AM.

At 10 AM sharp (more or less... an amazing accomplishment in Malaysia, I am sure) we had our honored speaker from Valve.

Talk 1: Valve's Digital Distribution
Mike Dunkle - Director of Cafe Operations, Valve

The focus of the presentation was more on Valve's Steam Powered distribution system, which essentially provides content directly to their customers over the internet. The main thrust of this form of distribution is mainly to capture the indie development scene and motivated to revive the enthusiasm that comes out of that scene.

Steam is actually a main part of Valve’s business platform. Its embedded nature within the platform provides excellent support to its other core business components. With Steam in place, Valve can now support games faster and unobtrusively, as any patches are done when the customer logs into Steam. This is in contrast to the traditional file patch distribution that is commonly used to address bugs and fixes. The other benefit is the scalable framework of the distribution architecture available.

Currently Steam is running off 200 servers. Valve purchases bandwidth with multiple vendors to ensure that the Steam infrastructure is in order and in place. They pay $15/$20 per 1 Mbit of bandwidth. The cost is high to ensure that the user will not feel disadvantaged in their usage of Steam. Mike notes the need for servers to be set up in Malaysia to handle the poor connectivity to Steam from his previous night's experience. These high bandwidth costs are partially recouped through adverts from other companies. (I wanted to ask about the potential of building a similar distribution system using P2P, like Bittorrent, but forgot. Not that the "full game" is hosted on Bittorrent, but maybe a partial or 3/4 shared P2P distribution would be more logical.)

The obvious benefit of using Steam was the successful dampening of the piracy effect (I disagree with his assertion that piracy was successfully prevented via examples such as the fact that the new cracks on Half-Life 2 does not even require a person to log into Steam). Surprisingly, piracy was their least concern.

Online cheating is their main concern, as this was a driver to reduction in customer base as customers would leave the game altogether if they had bad experiences within the game environment. Mike reminded the audience at this point that we need to keep in mind that the end users are not the average young adult male, but is the 13-year old male to the 65-year old person. Steam thus needs to provide sufficient flexibility to ensure that the experiences of these end users will not be ignored.

The topic then shifted to how Steam distribution is better in giving the end user instant gratification of the game when compared to the installation through traditional CD/DVD media formats. According to him, Half-Life 2 was built (under Source which has Steam embedded within the code) with the ability for the gamer to start playing using 70MB of data. Instead of waiting for the whole game to be installed from the DVD, gamers can now start playing with the first 70MB while the rest of the data is installed in the background gaming environment.

Mike also explained Valve's commitment in building an online community that drives the content through SDKs. Valve has a 5-step plan that would help modders to distribute their own mods on the internet. Valve provides time, guidance and feedback on the mods that are released. These online modders would create a stronger support for the engine itself.

The Q&A session brought to fore the following:

• Valve is not ditching the retail system, as 75% of their business still comes through them.
• Revenue sharing with developers under the Steam system is 60/40 (developer/Valve).
• Developers will not need to use Source in order to use Steam.
• Mike encouraged developers to separate their retail distribution contract from the digital distribution contract/rights. Hold on to them separately and not together. Control in your distribution is essential in maintaining good profit margins.
• Valve also experienced problem with returns using Steam. But these are only less than 1% of the sales received. The problems are commonly in the form of charge backs from banks upon rejecting certain transactions upon the customer's request. This will need to be addressed when the number grows more significantly.
• The cost structure of Valve's game is roughly 3% for customer support, 10% for bandwidth, 1% for payment charges and the rest is the profit margin. Reasonably good margins.
• Steam distribution's success is shifting Valve into trying out episodic gaming distribution next.
• Marketing function should be kept within your own company, essential as it helps create identity. This identity will be specific to your company and therefore, must be guarded.

Talk 2 - Breakout Session: Multiplatform Game Engine Development
Brett Bibby - CEO of GameBrains Sdn. Bhd.

This is a fairly technical guide for programmers who are in or are considering venturing into cross-platform development. The salient points of Brett's session are as follows:

Naming Conventions are extremely important. Avoid the use of spaces. [We presume he means in paths, which are treated differently on different platforms, not whitespace in code, which improves legibility – Ed.] Good conventions will help in the development of content heavy games. He prefers camelCase over underscores.

Folder Structure: Use naming conventions (e.g. singular not plural). Organize code into sub-folders. Very practical advice, which is not usually practiced by programmers. Need to be enforced as it helps in debugging.

For file naming, it is good to use your own extensions. Will enable you to perform multi-phase parsing. Create your own.

API Standards: Brett uses (Scope - Module - Feature - Property - Action) convention instead of the opposite which is commonly advocated.

There are two models in managing your filing structure:

Driver Model - which is better for experienced, or mature teams;
Module Model - which is better for newer developers as they will need to re-arrange and re-write the codes once again for other platforms.

There two problems in implementations:

• There are many features that are specific to only one platform
• There is a need to create something that fits all

Brett covered a few more:

• PIMPL - Good for hiding information.
• Modified PIMPL.
• Use Unicode strings, avoid char.
• Do not use multiple inheritance.
• Do not use factory pattern asset loader.
• Brett heard recently that Sony will drop games that have more than 30-second load times.

Brett concluded with explaining that good structure and tidy workspace will increase the productivity of any developer team and must never be underestimated. About 20 people attended this session.

Talk 3 - Breakout Session: Programming Stickiness in Games
Allan Simonsen - Managing Director of startup game development company Paradise Isles Ptd Ltd

Sorry, I did not take any notes from this, as the after-lunch effect was kicking in. There were about 20+ people in the room.

The main topic dealt with lessons learnt from casual games and MMOGs in engaging users and encouraging long-term playability. Allan made it clear that the word "addictiveness" was purposely avoided in dealing with this subject as it gives off the wrong connotations.

Pokémon was used in understanding the user's obsessive compulsive behavior in "catching 'em all." He dealt also with the concept of the need for players to have "bragging rights" which is translated into the game in terms of global ranking and replays.

Allan also emphasized getting good customer/player feedback and where and what to avoid in getting them. The dangers of online forums were also highlighted, as a minority of the gamers are involved in these forums and they usually exaggerate their intention and feelings online. Focus groups seem to be the better way in which to extract information about your product.

Free labor, the concept of allowing user-created content to create hardcore fans, was briefly touched also. One of the highlights of this talk was Allan's explanation of how developers/designers should know SQL and other analytical skills. He focused specifically on data mining for user online behavior and how to leverage these into improving your product content.

Customers love it when you have addressed their grouses.

Page 2