Originally published here.
Primmies, winner of the 2005 Second Life Game Developer Contest, is the delightful strategy/puzzle/action game that resembles a multiplayer, three-dimensional Lemmings, challenging you to navigate your adorable tribes of Primmies across treacherous terrain, while preventing them from marching suicidally off the edge of the randomly generated maps which appear from a whirling vortex.
In naming it the overall winner, veteran game developer and contest judge Doug Church praised it as "a focused and self-contained game... a game sandbox to play in," full of nice details that provide "a fun closure to it."
There's really just one problem with Primmies. Primmies is dead.
Co-creator Jeffrey Gomez discovered this shortly after the recent release of SL version 1.7, when he returned to the land he'd won from the contest, and fired his game up. The thing of it was (without overgeeking on the particulars) Linden Lab had just changed with 1.7 the way objects in the world handled collision detection. "Everything works but the Primmies themselves now," Gomez told me. "Who either collide with one another now (bad) or don't collide at all (worse)."
Or as Gomez surmises it, "Let's just say the cows have come home to roost."
For in his eyes, this was a sign of larger issues for scripters like him, as they attempt ever more ambitious projects that require months of development, through several updates. "[Y]ou had asked why big production games don't work in Second Life," he tells be. "That's why. The commands are just too subject to change at the higher levels... The higher-function stuffs, collisions especially, are changed pretty often." (In this particular case, an element in the collision system that would somtimes cause regions to crash was fixed, and in the fixing, Primmies was broken.) "It's tough working with code no one really understands enough to follow out to every logical conclusion," Gomez reasons. "Makes stuff impossible to predict."
Unsurprisingly, LL development coordinator Chris Linden sees it differently, pointing to Preview, which as the name suggests, is a pre-release version of Second Life where Resident developers and builders can test how the upcoming upgrade might impact projects they have in the works. "What is frustrating from our perspective," he e-mails me, "is that we advertise the Preview grid months before we launched 1.7. Jeffrey could easily have come to Preview and tested Primmies and discovered the problem and contacted us. If done early enough, there is a good chance we could have done something."
"The devs have been very supportive of fixing it..." Jeffrey acknowledges. "It's really not the fault of the coders. Or Preview grid. Or anything. It's a logical fallacy at the systems level. That being, when a new update goes live, it goes live everywhere. Contrast this with most systems builders and designers, who know it might break stuff, and basically give people the option to upgrade."
He explains his perspective in systemic terms, in comparison to the Internet itself.
"Current assumptions by the [Second Life] system:
1) Linden Lab controls all data.
2) LL employees are human beings, hence, they make mistakes.
3) All patches are final and happen everywhere at once, due to point 1.
Compare this with the way the Internet works:
1) No one controls all data.
2) People screw up.
3) Any changes to the system happen ONLY WHEN ADOPTED."
If there's anything to be gained from the inadvertant death of Primmies, perhaps it's Jeffrey Gomez's exhaustive white paper on where he believes the future of metaverse development should go. [5/17/06: Seems like Jeffrey's white paper has been moved-- will check with him on new location. - HL]