Linden Lab Working To Beef Up Second Life Stability, Usability

Linden Lab plans sweeping improvements to the Second Life platform this quarter, designed to improve stability and usability, and to make it easier to connect Second Life with the rest of the Internet.

Linden Lab is upgrading the Havok "physics engine," software that is a fundamental part of the foundation of Second Life. The physics engine controls the way Second Life objects interact with each other and react to gravity.

The company also is deploying Mono, an open source implementation of the Microsoft .Net Framework, to improve performance of scripts. And it's continuing ongoing development of a Web browser that will function inside Second Life, and a lightweight Second Life client that will support text chat and voice.

The bulk of these changes are planned for this quarter, although some of them will stretch out for the full year, said Joe Miller, VP of platform and technology development at Linden Lab. We met with Miller (whose Second Life name is Joe Linden) at Linden Lab's San Francisco offices recently.

Most of the changes are directed at making Second Life more stable, Miller said. "You shouldn't notice any difference except the service will be more stable and the performance will be more constant," he said.

Linden Lab is upgrading the Second Life service to Havok version 4.6, the latest version of the Havok physics engine. Havok is fundamental to the software foundation of Second Life, which now runs the 8-year-old Havok 1 engine.

The upgrade will be extremely significant in two areas, said Miller. Second Life will be more stable -- Second Life servers, known in the Second Life community as "sims," will crash accidentally much less frequently, and will be much more resistant to griefers' attempts to intentionally bring down the servers. Linden Lab is confident that it will be able to close every known means of crashing a server by the time the Havok 4.6 deployment is complete.

The software is being deployed on production Second Life servers -- known in the Second Life community as the "main grid" -- with the permission of server owners, Miller said.

The other visible effect of going to Havok 4.6 is that the performance of Second Life servers will be more predictable, Miller said. Servers become slower and less responsive -- a condition known in the Second Life community as "lagginess" -- when they fill up. And they fill up quickly -- most servers have a capacity of 40 avatars, although that can be customized by owners of some servers, which can support more than 100 users in rare cases. With the Havok 4 deployment, Linden Lab hopes that server responsiveness and performance won't slow down as the occupancy nears capacity.

In another architecture change, Linden Lab is deploying Mono as a foundation for running the Linden Scripting Language. LSL is the language used to control behavior of objects in-world. Mono is being deployed this quarter, starting on the beta grid. Mono will allow scripts to run up to 700 times faster than they now do, theoretically, although in practice performance has been 100 to 200 times faster than current rates, Miller said. The goal of the Mono deployment, as with the Havok 4.6 rollout, is to make Second Life more predictable and stable. Eventually, Mono will allow developers to write code in languages other than LSL, such as C#, Miller said.

Linden Lab also is making progress on a long-term project: Allowing users to display a Web browser in-world, Miller said. Using the so-called "browser-on-a-prim" technology (a "prim" is the fundamental building block of Second Life objects), developers would be able to display the contents of Web pages on presentation screens, the walls of buildings, or other objects in Second Life.

The inability to display HTML in-world is a fundamental problem with business and education in Second Life, making it difficult to display presentations in-world. Developers have to resort to cumbersome workarounds and a single page of a presentation can take up to a minute to come into focus in-world.

And Linden Lab is working on several measures to improve customer retention. Currently, about 90% of people who try Second Life give up on it quickly. Anecdotally, that's pretty similar to the retention rate for most Internet activities, but Linden Lab would like to do better. They have regular, weekly focus groups of average consumers brought in off the streets, to test Second Life usability, Miller said.

The company is also developing software to make it easier for users to invite nonuser friends in to Second Life for a special event. Second Life is the host of frequent events, including live music concerts, lectures, press conferences by businesses, classes, and games. Using tools planned from Linden Lab, a resident would be able to send an e-mail to his friends containing a URL that, when clicked, would launch software that (with the recipient's permission), tests the recipient's computer to see if it's powerful enough to run Second Life, then installs the client software and assigns the recipient a temporary name for entering Second Life. "That's social networking 101, the ability to bring someone else in on your aegis," Miller said.

Linden Lab is working on a new lightweight Second Life client that would be used only for text chat, instant messaging, and voice communications. Miller said he expects the client, which will run on Windows and the Mac, to be available in beta in February.

Most of these changes are tactical, patching Second Life to make it more usable and fixing obvious problems with the platform. But what about a year out? What will Second Life look like in 2009?

Miller said Second Life in 2009 will change from one grid to multiple grids. Linden Lab said in April it plans to open-source the Second Life server. The company open-sourced the client a year ago. Next year, users will be able to run their own Second Life servers, optionally behind a firewall or temporarily, for an individual event. Residents will be able to bring the same identity with them from one private Second Life grid to another.