Wednesday, 25 May 2011

Metrics: A Changing Metaverse

It was interesting to read  Hypergrid Business' monthly survey of the virtual worlds of the metaverse. SecondLife lost 41 regions while Opensim based grids collectively gained 1,102 for the top 40 grids. But HB is actually tracking 144 grids of which 80 report region counts. What is particularly interesting about these surveys is the picture it paints of the changing state of the metaverse which is in a real state of flux.

OSgrid and Kitely lead with significant gains in regions while  Avination, which rose so spectacularly in the early months of the year, lost regions and active users. InWorldz continued slow but steady growth and Meta7 closed due to legal issues. At the same time more grids came online. So it appears users are moving about a lot and new people are experiencing the free metaverse for the first time. The big loser of course is SecondLife which, while losing 41 regions might not seem much, has lost 500 regions in the past year.

There are so many options now compared to a year ago and the cost of setting up has dropped significantly too so it's no wonder Linden Labs, which still charges $1000 to set up a region and monthly tier of $295, is losing out. SecondLife still enjoys high traffic compared to Opensim based grids but no one really knows what the free metaverse traffic figures are. SL can report their traffic as it's all under one roof but the free metaverse is a partially disconnected cluster of small grids with no overseeing server to collect data.

Gaga checking a club out in Avination. Yeah, the dance pole works!

Hypergrid helps to connect grids but a lot of the grid owners keep it disabled to prevent content theft and, of course, partly to keep their residents on their own grid. Commercial grids of course are in competition with each other and since they rely on renting sims to users they obviously have a vested interest to operate on the same business model as Linden Labs which means a walled garden approach. And yet, even that is not stopping users from moving about no more than Linden Labs can prevent it's residents discovering what is out there in the free metaverse.

With the release of Opensim 7.1 at the start of the month things could be about to change still more. With 7.1 there is greater security to prevent content theft while still allowing users to travel to other grids with their same name and appearance. Once someone has found a place to call home (virtual home) and enjoy the community it offers they are less likely to travel so much but with 7.1 the barriers to travelling are reduced. What you spent on fixing up your avatar, for instance, wont be limited to the grid you call home so you wont arrive at some grid you visit looking like crap and having to rush around trying to find freebies to get fixed up for the visit. Moreover, you wont necessarily have to spend any more money to look as good as you did at home. Avatars are vain - believe it!

On the other side of the coin you probably wont be able to take anything you buy with you when you leave a 7.1 grid either unless the owner has the Outward bounds permission set to allow it. Most commercial grids, if they upgrade to 7.1 probably wont allow it anyway but that's not to stop you visiting to attend an event such as a music gig or dance with friends and just hang out. You might want to attend a business meeting or an educational class. Perhaps you are a role player in some ancient theme like Romans or the fantasy realms of Elves, Steampunk, etc, etc. And you set out to interact in war or peace with a neighbouring grid that follows the same theme. It's really little different than teleporting to a neighbouring region in SecondLife. You just find yourself on another grid and can still look the same everywhere.

Role Players at Role Play Worlds grid

The owner of Avination, Melanie Thielker has gone on record saying she will enable hypergrid once the security is better and she said she would press for this earlier this year. As she is also one of the code contributors to Opensim then we have no reason to doubt her. But Avination has not yet adopted 7.1 so it remains to be seen if they will open up to the rest of the free metaverse. Certainly, if Avination dose I think they will benefit with increased traffic and, since Avination has a reputation already for gambling there is another reason to keep the door open. InWolrdz on the other hand is unlikely to become hypergrid enabled even if they could because they have chosen to fork off from the main branch and develop their own code on top of Opensim which means they may be too far removed to be compatible now. This may yet prove a mistake for them as the metaverse becomes more connected.

There are of course still limitations to Opensim which may not please people coming from SecondLife where they are use to most things working after a fashion. Most notable is physics which are still better in SL than OS simply because Linden Labs can afford the commercial licenses. Opensim is still limited to ODE which most developers regard as very basic. certainly, you see it's limitations when sailing a boat on Opensim-based grids which is clumsy at best. SecondLife has Havoc physics which is far superior but this could also be about to change with the rise of Aurora sim. In Fact a lot of things are going to change with Aurora sim!

Decent physics still give SecondLife the edge when it comes to sailing and sea battles

Even Hypergrid Business survey might have to change as a result of the advances in server code that is being rapidly developed by the Aurora team. Most notably, region sizes. You see, Aurora developers have managed to change the structure of regions which they call var-regions. These var-regions can be up to 256 times the size of a standard SL region. Unlike mega-regions in Opensim, which are clusters or child regions attached to a parent region in order to avoid the problem of border crossings, var-regions are just a single region that has been expanded up to 256 times the normal size. So, where Hypergrid Business collects data on total regions on grids this would be misleading in the future when looking at grids running Aurora sim code. On a Opensim grid if you see 256 regions you can certainly count them regardless if some are connected as mega-regions. On an Aurora-based grid you might see just one region and count it as one while, in fact, it covers the area of 256 regions. The metaverse is certainly changing.

I have a big article coming up later this week which is an in-depth review of Aurora sim that I have been working on for weeks so check back again soon. But I will leave you with video from Skidz Partz - one of the Aurora team - to get an idea of just what is on the metaverse horizon.


  1. Gaga --

    You are absolutely right. I'm going to have to rethink how I track grid sizes.

    Some options:

    Total number of acres (standard -- 16 per region)
    Total number of square meters (standard -- 65,535 per region)

    Total number of square kilometers (can't do the math in my head!)

    Or convert all grid sizes into "standard" OpenSim regions.

    Any suggestions?

    -- Maria

  2. Hi Gaga,

    I think Aurora Sim are doing some VERY interesting things. It's just unfortunate that they can't do so as part of the standard OpenSim project. Opensim would be much further along if all the development that was done by the various groups that forked it ended up back in the main project...

    BTW, you have a small typo in the link to Kitely :)

  3. Yes Maria. I could see this problem coming up and, short of actually going to Nova grid and checking region sizes (extra work!) you may have to ask grid owners running Aurora based grids to supply relevant information about the regions they are supporting. Not all viewers support the var-region protocol (latest version of Imprudence and Astra viewer does) so you need a compatible viewer to do an eyeball check on the map. I think it is probable that Aurora grids wont have as many regions anyway. For example; Avination as a little over 1000 regions which could actually be reduced to just 4 regions in Aurora with the same total area.

    It's too early too say if a commercial grid like Nova will support var-regions although, if I know Enrico at all, I suspect they will. It could be they will price land on area rather than by region count which could lead to some interesting real-world parallels. Perhaps the grid owners will start to talk in terms of land area sold too and your metrics may need a new column to handle this. But right now it's any ones guess.

  4. Hi Ilan

    Yes, it is unfortunate that the Aurora devs split off from the core but it's not hard to fathom the reasons. Not looking at the viewer code for six months before you can submit a patch to core made it impossible for the Aurora devs to make the improvements they wanted to. And, apart from that, they wanted to rebuild Opensim from the ground up so it has a more modular structure and far easier to understand.

    When OpenLife forked early on in the development of Opensim they complained their contributions were being ignored but everyone knows, I think, that their real reason was commercial interest. InWorldz forked too and their devs made similar complaints about core Opensim. They actually described the code as a mess and, at the time, I locked horns with them over it because I felt they were biting the hand that fed them.

    More recently I have changed my views some because, after four years, so much in Opensim core is still not working which has held people back from investing - myself included. I hate to say it but I think the Opensim developers, for all their great work and dedication, have a rather blinkered view on where they want to go with Opensim. I think the devs have various personal interests and are not really listening to the community that has sprung up. Saying that may get me into trouble but I have always supported Opensim and followed it almost from the beginning and have heard the same complaints over and over again. Too much still dosen't work while "shinnies", as they have been called, seem to get priority while physics and some of the basic services and scripting functions have been left un-dealt with.

    Aurora devs have dealt with all this missing stuff in a little over 7 months and gone further with some interesting innovations like var-regions. What's more (and this is important) they do listen to the small, but growing, community they have attracted.

    And they remain open source while other forks have gone proprietary.

    I will be saying a lot more about Aurora sim in a new article soon.


    P.S. lol, sorry for the typo and ty for pointing it out. I fixed it.

  5. Hi Gaga,

    First, thank you for the fix :-)

    I completely understand why the various groups that work to improve OpenSim outside the main branch do it that way I just think that we, as a developer community, need to find some way to work together to advance the goal of a free VW architecture for the world. It is a waste of development resources to split our forces. I wold hate to see some proprietary architecture become the standard when we had such a big head start in building the "Apache of virtual worlds" (please note I'm not referring to Aurora Sim, which is currently open, but to some undefined solution).

    Maybe it's time to revisit the code submitting policies to OpenSim (which, I must make clear, we respect and follow). Many other non-GPL licensed open-sourced projects don't have such strict requirements about not viewing GPL code before submitting to the upstream. I understand why this precaution was initially put in place, I'm just not sure it is still required given the amount of SL GPL-free code that already exists in OpenSim. If we find someone accidentally (or deliberately) included GPL code we should remove that patch and rewrite it. That is what other non-GPL projects do. IANAL but I believe that having to sometimes rewrite code would be less costly to OpenSim than the alternative we have now.

    We don't look at viewer code and have submitted just one OpenSim patch to date. This rumbling is just my own personal $0.02 on the subject.

  6. I really agree with the idea that these forks would be better served as being enhancements to the core app. Aurora Sim is looking pretty exciting, but I don't want to be locked into a niche system that doesn't have the wide range of support OpenSim does, and isn't necessarily guaranteed to be compatible with Second Life's future changes. I really think OpenSim could benefit from a more module-friendly architecture, so developers would be more encouraged to simply "plug in" their new changes, without disrupting the core app, instead of completely forking into a new and ultimately incompatible project.

    You mention that you think the physics engine support is going to change with Aurora Sim. How do you think that's going to happen?

  7. Hi Justine

    Thanks for commenting and I'm all for a more modular approach which Aurora sim actually is and has been re-coded from the ground up to make it easier to understand and work with. Some of the Aurora devs do still contribute to Opensim as it happens so the break has not been absolute. I think the real problem boils down to Opensim's licence which, for fear of a stopping order, is condemned to follow slavishly in SecondLife's foot steps.

    Linden Labs has generated a lot of anger from their community by pursuing unpopular aims which they doggedly forge ahead with and this reflects on Opensim where the devs feel obliged to conform. LL has a deep belief, it would appear, that in order to grow they must become more like Facebook regardless if this goes against the mood of the residents. Whatever else LL is doing to improve the virtual experience is lost in value for all but the sworn and dedicated diehards if the masses are in contempt of the company.

    Aurora devs felt the overwhelming influence of SecondLife too and came to the conclusion, it would appear, that a stopping order was unlikely and in order to break out of the restrictive mould and get creative they had to break away. The saving grace is that they promise compatibility with Opensim where ever possible even if they have to build bridging modules to achieve it, and they remain open source. Moreover, there is a strong gaming influence amongst the developers and they are listening to their community.

    They is a new breed of people developing for the 3d metaverse that want to give people a truly virtual experience and not a social networking tool, and they believe the community is telling them that. Linden Labs and Opensim do not appear to be listening. Fortunately, Kitely and Aurora teams are.

    Personally, I have always felt passionate about the emerging 3d metaverse, that it should be free and open, and connected. I have always supported Opensim for those reasons but it has been too long in the making. Aurora sim and Kitely have the right ideas in my view and for that reason I feel the wind of change like a breath of fresh air.


  8. Oh, the physics. I nearly forgot that. Yes, The team has done a lot of work on improving ODE but recently I learnet the Revolution Smythe, lead dev of Aurora, as been working on PhysX too. Another of the devs has actually had it working on a test sim from what I gather and, in answer to a question about PhysX I put to Rev the two got together on it. It appears they had both been working on it in isolation so this all looks promising now they have their heads together.

  9. Hi Gaga,

    Thank you for the vote of confidence, we do try to listen.

    I'll be looking forward to your article about Aurora Sim. It sounds like they are working on some much needed improvements and doing so in a way that fosters the creation of a successful open-source project. If they stick with open source, continue building a capability gap over the standard OpenSim branch and maintain compatibility so people can HG teleport from OpenSim-based grids then maybe one day Kitely will offer Aurora Sim based worlds as well.

  10. Gaga -- There's always a big difference between things done in small groups and things done by large committees.

    A small group, or an independent company, or even a single determined guy, makes all their own decisions, and can forge ahead really quickly.

    Large open source project, however, move ahead by consensus. Think of how slowly HTML standards evolved! Everyone had to get on the same page before they could do anything.

    Meanwhile, independent companies -- Flash, Quicktime, Unity and many others were able to roll out functionality that the general WWW lacked.

    At the beginning, HTML was widely decried for being ugly, insecure, and with none of the functionality that people wanted.

    In the short term, if you really need a particular feature, you're better off with a commercial product like Flash (or, in the case of 3D, with VenueGen or Web.alive or one of the OpenSim spinoffs). But it's not a good idea to bet against the big projects in the long term. Necessary features - eventually -- make their way in. And the organizational structure means that you've got wide buy-in for whatever they decide to do.

    For example, having your own viewer would be great for OpenSim technically. But in a practical sense, it would cut OpenSim off from the millions of Second Life users out there -- and also from all the Second Life tutorials, videos, training classes, and all those other materials.

    So unless the new viewer was either identical to the SL ones -- or so brilliantly simple that no training was required -- it would be a hard sell. The realXtend folks, for example, haven't been able to get much pickup for the Naali/Tundra project despite some significant technical advantages.

    But in the long term, of course, we're all dead. So we might as well enjoy short-term benefits of Flash or the OpenSim-spinoffs. And that short term might not be so short -- Flash has been around for something like a decade, and the Web is only now starting to add some of its functionality with HTML 5.

    -- Maria