Fashion Research Institute's Shengri La

I’ve received several comments with wild-eyed claims and various anecdotes about OpenSim, including a recent one about a simulator with a build of 100,000 prims. Folks, this entry is for you.

While I’m waiting for Spirit to be groomed and tweaked and made ready for my next assault, I’m going to take the opportunity to talk about why we’re doing what we’re doing.  The Fashion Research Institute didn’t actually set out to be alpha testers of open source code. 

As CEO of the Fashion Research Institute, I’ve done my due diligence about virtual worlds. I personally have explored all of the virtual worlds out there in the last year of developing the Fashion Research Institute, and our virtual world-based product design and development technology solution.   But after a hot-eyed tour of the many virtual worlds out there: Blue Mars – stunningly beautiful.  World of Warcraft - lots of users.  Stardolls? Shopping for the tween set…and the many other worlds out there…it became crystal clear that none of the existing virtual worlds was going to be what we needed for our solution.  

These virtual worlds all had issues, not least of which is that most of them are games.  Entertainment for the marketing demographic of choice, which means we can’t use it for our solution - the Fashion Research Institute isn’t serving the media and entertainment industry.  We’re building an enterprise-ready virtual world-based technology solution. 

There’s nothing playful about it, unless you regard business like Edith Wharton: “He had the Saxon love of games, and the best game of all was business.”  We’re in business in the apparel industry, and part of our business demands that we have an appropriate platform.  As I’ve reiterated at my many talks, the real value proposition for virtual worlds isn’t in marketing or serving the consumer base.  It’s in helping enterprises succeed at their business by using virtual worlds to enable their work flow – at which point, the consumers will follow.

The Fashion Research Institute was facing a dilemma.  Second Life tm has graphic quality that is ‘good enough’, and a richly immersive experience.  But Linden Labs’ tm Terms of Service agreement alarms me as an entrepreneur.  It’s fine for individuals, but an enterprise that is serious about their business information and intellectual property would never allow their proprietary information to sit on a Linden Lab server. 

And then, OpenSim was presented to me as an option.  It was an option that was ringed and garnished with a lot of cautious warnings like ‘well, you know, this is very alpha code’, and so on.  And at the point where I first went in, in October of 2007, it really was quite rocky.  But it was also very clear that it was our future, and I’d better embrace it.

And to that end, I had my people set up the first of our OpenSims, and we started playing with them.  I now have the abandoned ruins of four or five OpenSims laying about on my boxes, and of course, Shengri La Spirit alive and well on an IBM-hosted Blade.

Fast forward to where we are now: testing the code.  And, I’d like to think, doing a service to the OpenSim community, and in the spirit of open source, making our data available for everyone to see and use, in the form of this blog, and feedback from Kurt, Sean, Dale, and Zha into the community.  Open source means just that: being open about what you are doing, and showing your work.  Being transparent about it, so everyone can benefit. 

For example, I’ve had a lot of technologists tell me that the prim limit in OpenSim is arbitrary.  I am first and foremost a visual learner – I like to look and see for myself….and that means actually seeing the  performance limitations for substantive builds.  Now, it is true, I could have just asked my IBM team to create a script that would have rezzed prims in a loop till the system ground to a halt.  It wouldn’t really have impressed anyone, particularly those who write loops. And we wouldn’t have learned anything in the process – a machine cannot alpha test because it isn’t human and it does not have the sensitivity to learn from the experience.  All it would have done is dumped in as many prims as it took to grind the machine to a halt. 

But having a server full of prims, with no active observer, or worse yet, an observer who is unable to log and report what she observes, really doesn’t serve any useful purpose.  You can’t actually learn where the FUNCTIONAL prim limit is – you know, the one where the overall user experience degrades to the point it becomes unacceptable to the human user – a clearly human condition that a program can never identify. 

So we’re building out to find and push the functional prim limit, on a specific box, and we’re benchmarking the performance of that machine, with the given installation, and with a lot of user parameters being fed back.  I make no secret about the fact that we’re performance tuning as we go along; that we are not yet pushing textures, inventory, scripts, or a range of other parameters (that’s coming, soon enough).  We’re systematically focusing on prim limits first, which in our case is a human-created substantive build that uses primitive-based objects, including basic system, tortured system, sculptured or flexible primitives.

And we’re going to keep running out onto the ice until we fall through, at which point we will know where the functional prim limit is, for this set of parameters, and we’ll push it further.  When we find that functional prim limit based on our parameters, tuned for the IBM Blade hosting it, we will have a benchmark, which we will share so that the OpenSim community also has that benchmark. 

And this is why Spirit is so important.  Benchmarking performance, and sharing our data.  If you, my reader, have done something awesome with your OpenSim and you haven’t shared your data….well, anyone can SAY they did something.  But in the Spirit of scientific exploration, if you haven’t shared your data, you’ll forgive me if statements about ‘what you did in your OpenSim’ aren’t received as anything more than your marketing material to be circular filed. This is an open source community effort, and in that Spirit, I’d ask you, “Where’s your data?”

I’m not clearing space on my calendar to beat on Spirit because I love games or alpha testing.  I’m doing it to move the platform forward, because alpha testers who can actually test and provide worthwhile feedback are tough to find.  And I’m talking about our work because I feel strongly that the results of my alpha testing are important to the community as a whole, and that there are some very dedicated and capable people out there who will grab the results of what the Fashion Research Institute is doing in our collaboration with IBM, and run with them. 

Personally, I cannot wait to see the results.  Thank you again, to all of the dedicated open source & OpenSim supporters, coders, programmers and technologists who share their work openly and publicly.  You rock.