Friday, April 20, 2007

Ponderings on the Future of Second Life

People often think my faith in those at Linden Lab is a bit too much. Seeing the current state of the grid and all of the problems it has, I can understand why. I guess part of the reason I have faith is that I can see a direction that I would be taking Second Life in the long term, and I believe that the path I see is somewhat similar to the path Linden Lab is walking on. The way I see Linden Lab's business model heading long term is a cross between MySQL AB and InterNIC. I don't think their desire is to be an ISP, which is currently their role in many ways, and there are quite a few other ways them to make money, since they hold the keys to the domain, as it were.

For those who don't know, MySQL is the world's most popular free, open source database server. There are literally millions of installations of MySQL out there, and some of your favorite Internet services use it: Linden Lab/Second Life, Craig's List, ClassMates, Google, Hot or Not, Yahoo!, Tomb Raider, the City of New York, and of course, SLBoutique and SLTrivia! MySQL AB, the parent company, gives away this amazing product for free. Out of the millions of installations, key customers (some listed above) pay for service. They charge on a sliding scale, based on level of service; some have up to "X" incidents per year where they can call MySQL AB in, and others have "all you can drink" type service. I believe about 15,000 total companies pay MySQL AB, the company, for service. They make a pretty penny, but compare that 15,000 versus a total of 8 million installations. They also control the source code; anyone who modifies the source code has to submit their changes back to MySQL AB, who can then decide whether or not to incorporate the change. They also employ developers actively working on the project. They're a profitable company. I can see Linden Lab heading in this direction with the code for both the viewer and the sim server.

Some things on the Internet need to be centralized, but distributed. An example of the central and distributed model is InterNIC and DNS. InterNIC is the authority where you register domains, like,, or DNS is the system that allows browsers, email servers, email clients, and basically anything that uses the Internet for communication to point these names at the correct servers world-wide. I foresee Linden Lab doing the same with the grid; you pay them a small fee of $X per year to have the right to plug into this "square" of the map. They maintain what squares go where. If you're Anshe Chung, and you have an entire continent, you'll probably want to host them in the same data center, so border crossings are smooth (relatively), and pay Linden Lab for a full joined block of squares. The land store is the infant-stage of this registry.

The long and the short of it is, Linden Lab could eventually host only a few servers: their master user server cluster, the "grid map" server cluster, the money L$ server cluster, and probably a few beyond my foresight. You could either host a bunch of private sims, with your own core servers, or have your sims subscribe to Linden Lab's master servers. This would plug you into the "main grid" map, use the Linden Lab L$ economy, and allow anyone with an account access your sims. Think of this option as how the world works today, except that instead of sims all being centrally located, they're hosted at various data centers world wide.

There could also be subscription servers for asset storage. It isn't a big leap of faith to imagine that the ( SL viewer + Linden master organizational servers for one virtual world + simulators hosted world wide + Amazon's S3 acting as asset server = "The Metaverse" ).

I've talked to various Lindens over the years about the possibilities, and none of them have said this is the direction they're headed. This is just the way I see things starting to play out, and the direction I would be heading if I ran Linden Lab. They're very bright, dedicated, capable people, and I think of lot of the problems that are occurring now are because something like I've stated above is the long term vision, and they're focused on long term goals instead of short term band aids. At least, that's what I hope!


  1. And I seriously hope this scenario doesn't happen. I don't want a private company controlling the metaverse.

  2. That is definitely a valid concern, Sean. However, I see a large difference between a company with a closed source product (IE: Microsoft) and an open source product (IE: MySQL AB).

  3. Just a license-nerd type comment. The situation with MySQL's license isn't quite as you state in your post ("They also control the source code; anyone who modifies the source code has to submit their changes back to MySQL AB, who can then decide whether or not to incorporate the change.").

    MySQL has a dual licensing model, both GPL'd, and with a commercial license. In a situation where a user chooses to use MySQL under the GPL, they are NOT required to submit changes to anyone...though many people do. You are required to release your changes under the GPL if you /distribute/ the updated code; but you're not even required to do that if you only use it internally.

    OTOH, many commerical companies buy the commerical license, which provides commercial support. This is the model that allows MySQL AB to make money, and is a perfectly acceptable dual-licensing model under the GPL. It's also why MySQL AB does require copyright to be signed over to them for changes that go back into the primary project; they couldn't dual-license otherwise.

    Not trying to be overly pedantic, I promise...but you mentioned some of the common misconceptions about GPL'd and dual-licensed code in your post, and I thought it worthwhile to comment. Thanks for understanding!

    WRT the rest of the post, I agree in the abstract...LL does probably hope to go a distributed model. They'll have to compete on quality at that point, as there will be no reason that alternate asset repositories, grid map servers, or even currencies couldn't develop. Should be interesting!

  4. Thanks for the clarification! I will be the first to admit the legal aspects of the GPL are not my strong suit, just a cursory understanding of the basic flavors.

  5. Okay then Flipper... here's my long answer... :-)

    Something like the model you are suggesting is likely. Joe Miller has said they will be allowing people to host their own grids at some point. It's just a question of what will be centralised - the grid map info as you suggest? The accounts and logins? The asset servers? The economy? (you betcha on the last one!).

    The fact that the software is open source really has little to do with it, though. They could simply license or sell closed source server software and still achieve the same model.

    You say they don't want to be an ISP... sure... why would they? people can connect to the Internet already. What they might want to be like is a portal to a closed metaverse, like AOL and Compuserve were portals to closed networks.

    In the short term letting people host their own SL server software (and even whole grids) will be far better than what we have now, but make no mistake though that if we want to connect to the Second Life grid we will have to abide by LL's rules, and LL cannot be compared to InterNIC (which is now called ICANN, btw) because the latter is a not-for-profit organisation entrusted with the management of the domain name system with guidelines from the US government and the former is a private company which will ultimately be answerable to its investors and shareholders and will have its own interests at heart (no matter how wonderful Phillip Rosedale's vision is).

    What you suggest is nothing like the Internet or Web which are built on open protocols and standards (such as TCP/IP, HTTP and HTML) that were developed over many years by the scientific and technical community for the benefit of all. These protocols are administered by neutral, not-for-profit bodies such as W3C (and ICANN).

    If you are suggesting that this will be the Metaverse, then what you are suggesting is akin to a private company like Microsoft controlling the Web - having control over the naming, the structure, the protocols and the methods for money exchange.

    Is that really the model you want?

    LL will be vying for dominance over the next few years with products like Multiverse (who are trying to do a similar thing right from the start) just like MSN tried to compete with AOL, but in the end I believe open and common standards and protocols and inter-operability will win out, and private metaverses will be swept away as sure as private networks like AOL, CompuServe and MSN were swept away by the Internet.

  6. Old habits die hard with InterNIC, heh, thanks for reminding me. I don't think we're necessarily disagreeing. With open sourcing the back end, anyone will be able to run their own grid. I'm not trying to say you must use Linden Lab's user server, economic server, and grid, however, I think there's will be the biggest contiguous world and community (at least for the foreseeable future). For many people like me, what makes Second Life the best current option isn't the technology, its the people, the freedoms, and the communities. One reason Croquet isn't really appealing to me, for example, is that it is set up to be thousands of smaller separate communities. I'm fairly sure people will have the option to host a completely separate grid on their own, but LL's power will be in the power of having one large contiguous world. The metaverse probably won't end up being one program, one interface, or one option; more likely a set of tools from which people can pick and choose according to their preferences.



  7. Here's an apropos quote I found at

    "But down the road, Linden plans to let companies run its software on their own servers, effectively creating their own Second Life-like worlds, says Glenn Fisher, director of marketing programs at Second Life. Users would have the option of providing a gateway to the Second Life community or keeping their worlds closed."

    I agree with you that much of the value of SL comes from the social aspect and the community. Private worlds that aren't connected to the main grid won't be as much fun, but then there may be an opportunity for another company to come in and offer interoperability and an economy that works between these independent worlds/grids, or maybe someone will come up with some protocols that will allow these independent worlds/grids to just link together without a central body - that would be cool! Then it would be more like the web. There'd be no economy, but maybe people will just use something like Paypal.

    From the demos I've seen of Croquet they do allow linking between worlds. They are called portals from memory and you can choose who to link to and who can link to you. They based the idea on web hyperlinks, so you can have a connected web of 3D spaces.

    It's a similar model to ActiveWorlds, if you've spent any time there .