The Freeze: Closing the Loop to Launch


Okay, you're in the last stages of the open Beta, the game is stable, and it's time to prepare for the official launch day. Now what?

Now is "the freeze," meaning you stop messing around with changes in both the client and server code. They don't necessarily have to be frozen at the same time; it depends on your client distribution model.

The Client: Retail Box

Since you're going to need a minimum of 30 days from the time you send the client to the duplicator to the time it shows up in a box on retail shelves , you have to finish up any last-minute polish on the client and freeze it at least six weeks before launch day.

Why six weeks instead of four? Shouldn't you keep working on the client right up to the moment you have to send it to the duplicator? If you're going to be selling the client in stores, you'll want at least two weeks of "frozen" status to test the retail version for any missed client-side bugs and just to ensure that it is truly stable. You'll also want to do fresh installations on as many different machines as you can get your hands on. QA will already have done as much machine configuration and compatibility testing as they can, but it can't hurt for everyone on the team to do fresh installs of the frozen version, work with it for two weeks, fix any lingering bugs or anomalies, and make sure it actually works correctly and intuitively.

Once the frozen client has been tested for two weeks and is shipped off for duplication, don't touch it . In the past, teams have continued to work on and add to the client after it was shipped to the duplicator, on the cheerful assumption that hey, we can just download the changes to the player with a patch, right? The problems come with the size of the patch downloads; they tend to be large. World War II Online and Anarchy Online ( AO ) both had 75MB downloads waiting for new players after the install process. For a dial-up user running at 56k, which still constitutes the majority of US and European connections, this can easily be a 7- to 10- hour download, depending on the quality of the connection.

The hard- core gamer tends to be more philosophical about large patches; just as price is of small object to them, so too is a 10- or 20MB download, for the most part. However, any patch for new, registering players larger than about 4MB can be a serious block to entry. Learn from the example of EA's Majestic , where more than 90% of 100,000 potential players abandoned a 10MB download of the free episode before completion of the download.

If at all possible, it is better to hold off on client patch downloads for two to four weeks after launch. By that time, you'll be past the initial rush for the doors and things will have steadied out a bit. Of course, if serious client-side bugs pop up during that period, you'll have no choice but to fix and patch them; just keep them as small as you possibly can.

The Client: Self-Publish Online

Bypassing the retail channel to publish the client online is a hot topic these days. We don't currently recommend this, mainly because most people still don't have broadband connections and forcing a long download is more market-limiting right now than requiring someone to walk into a retail store. That will certainly change over time, maybe as soon as 2005, but it is the reality of the situation right now.

If you're still determined to publish the client via online downloads, you'll need at least a two-week freeze period to establish stability, and it is recommended that you hold off on patches as much as possible for the initial two- to four-month post-launch period.

The Server

When the client is frozen, it is a good idea to also freeze the server code, but that might not be realistic; there always seems to be some cleaning up and tweaking to do. If you're going to make changes after the freeze, it is generally a much better idea to make server-side changes; they are easier, don't generally require a download to the player's client, and mistakes made in the patch are easier to fix.

In general, the name says it all; if you're going to freeze the software, freeze it. Every change after that greatly increases the chance of a problem at launch.

Wiping the Database Clean and Rewarding Testers

After 6 “12 months of testing, the PTs who stuck around for a few months will have become attached to the characters they created. This is only natural; if some people didn't form emotional attachments to their characters , we wouldn't have a business. However, this can present a tricky situation, as most games wipe the player database clean at the end of Beta, and this represents a loss in the minds of the testers. As the Beta phase draws toward conclusion, you'll start hearing talk from the testers that, you know, maybe you should be different from all those other guys and not wipe the player database clean for launch.

The players don't seriously expect you to leave the test characters intact; that would have the effect of unbalancing the game heavily in favor of them from day one of live operations and they know it. Long- term testers will already have the advantage of knowing the game better than new players, and that can be a pretty hefty headstart. What you can do is plan in advance how you will reward the testers for helping you out. This can take several forms that don't impact the gameplay directly, which should be your goal:

  • Reserved character names ” Allow testers who were active a certain minimum period of time to reserve a character name for use after launch. Since persona names are key "fame" identifiers in PWs, this is considered a great reward.

  • Special tester sigils or items ” Players love to brag how they were involved with a game "since the first Beta, before you punks even knew the game existed!" Unfor tu nately, most games have no way to easily acknowledge that status, so plenty of poseurs pop up. You can make this a meaningful status; if the game features unique clothing or jewelry pieces, character titles and ranks, or some other way to uniquely mark an online persona, create one specifically for testers and make sure they each get one. This is almost never done, yet it would be remarkably popular.

  • Extended free play ” The industry-standard reward is to offer some amount of free "tryout" time for newly registered players, usually the first month of play. You can thank your testers by extending this time, although you have to weigh this option carefully ; it can eat into your margins if your tester numbers were high.

The point is to reward your testers in some visible, recognizable manner.



Developing Online Games. An Insiders Guide
Developing Online Games: An Insiders Guide (Nrg-Programming)
ISBN: 1592730000
EAN: 2147483647
Year: 2003
Pages: 230

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net