OpenSimulator: The Choice for 2010

Open Simulator Logo

With Google Lively out of the picture, and an uncertain future for “new” virtual worlds to be launched in a world gone insane with the perceived financial crisis, it’s natural to ask over and over again, if there is an alternative to the Second Life® virtual world platform, what it will be.

Since early 2008 I have been very skeptic about “alternatives” to Second Life. In my mind, and specially after talking to several gurus and enlightened residents like Pavig Lok, any “competitor” to Second Life will require those mandatory features to succeed:

  • Collaboration and user-generated content. This mostly means being in a shared environment being able to work together. It goes way beyond simply being in the same place and chat together (either in voice or text), but being able to do things together, be it simply playing the same game, driving the same vehicle, using the same object, or, well, building the same building. And there can’t be a restriction on how you work together: it has to be both synchronously in real time (everybody seeing the changes being reflected immediately in the environment) or asynchronously (you might work on something, log off, and your friend comes in and finds things exactly as you left them and can continue working on it). Finally, user-generated content is not simply having an off-world Web site where you can buy clothes and furniture (like on IMVU or Moove): it means dealing with intellectual property rights; it means permissions; it means group ownership; it means an economy based on buying and selling digital content.
  • Persistence. This is a much more subtle feature: the notion that you can rez a prim, and it will be still around after five years. You might take this for granted, since in SL this is the way things happen. Many platforms might have user-generated content but rely on peer-to-peer networks to show that content to other users: OpenCroquet is the best example of that model. Under those models, either you save your work “somewhere” or it is lost once computers get disconnected from the P2P network (sure, you can upload them again once you connect).
  • Contiguity. Most virtual-worlds out there have a model where worlds are  sharded (the same copy of the content is replicated among several servers; a set of users connects to one local copy of the content, and although they can talk to users on other shards, they’re not in the “same” virtual world at the same time. World of Warcraft is a typical example) or roomed. Under this model, each server just holds a part of the world — from a single room, to a single region, to a “game level”. When moving to another room/region/level, you are actually changing your environment and point of view, by connecting to a different virtual world. Granted, tricks like “portals” and the ability to chat across rooms/regions/levels will provide you with the illusion of contiguity. Second Life uses a single world with the “tiled” paradigm (the world is divided into many regions, all side-by-side, but it’s a single world — at least on the mainland you can walk from one point to the other without going through “portals” or “teleports”).

And of course they require a valid business model to be still around after a few decades. Some virtual worlds have all the above requirements, but they are simply venture capital burners until Google/AOL/Yahoo/Microsoft buys them. 3D content hosting works for Linden Lab, and one might imagine that there are other possible models (e.g. advertising or sponsorship), but we have seen few of that happening. Subscription payments to get access to user-created virtual worlds seems, however, to be a dead end — probably the reasoning behind it is that if users create the content anyway, why should they pay a third party to see it?

(Movie available on the next pages, do read on! 😉 — Gwyn)

Second Life’s closest competitor is Second Life itself

Virtual Worlds, it seems, come and go. Over the years, I’ve added to my skepticism a certain amount of cynicism as well. The Silicon Valley hip culture has brought Geekdom the ultimate playground: just implement your crazy idea, and sooner or later someone will pay you for that. A lot of money. Don’t worry about long-term goals: worry instead on applying your skills to develop an awesome product, and you’ll be fine. Sooner or later, it’ll work out. Forget business plans — start with something, develop it to the full extent of your abilities, and you’ll succeed. Trust us. We’re business angels and venture capitalists, and we know how it works: for every ten good ideas, no matter how crazy, one will succeed (and pay off for 8 worthless business ventures and one which is so-so).

Well, the cycle seems to be always the same, even after the dot-com era. Companies are still launching impossible-to-sell virtual worlds (not to mention Web 2.0 websites). The whole messiness of the way these “new products” are launched is appalling. They have no business models. They have no idea what they’re selling. Granted, they often have awesome technology, excellent designers, programming experts, and good evangelists. They have talented teams with experience.

But they have absolutely no idea if they have a good business model or not.

Why the stress on business and not on technology? Mostly because Second Life is not an “awesome technological breakthrough”, although it certainly has quite good ideas. The renderer is by far not the best renderer in the world. The user interface was already obsolete in 2001, years before launch. And as we soon will see, not even communication protocols and server implementation are great. Not when compared to other, more sophisticated solutions.

Linden Lab was actually very lucky. They started with the wrong business model, and, even worse, the wrong market. They tried to sell a subscription-based system to games developers, when clearly Second Life was not at the stage where games could be developed on it (and some claim it will never be). Even worse, they started with all the wrong assumptions. Looking at ActiveWorlds, where almost all content is user-created, they tried to push for a similar model — kick-starting with some content (Linden trees, Welcome areas, roads, bridges, and some decorative elements), they hoped that users would do the rest.

They struck gold. It just barely happened to work. But to make money out of it, Linden Lab had to switch markets (from targeting SL to game developers to digital content creators and now to businesses and education for quite different reasons). They did, however, make some bold decisions: listening to Lawrence Lessig and implementing intellectual property protection on user-submitted content (“the user owns the copyright”), allowing the licensing of that content to others through SL’s interface, is dramatic — not even Facebook allows that!

When you add all that up — a combination of technology; innovation; user-generated content where users retain their copyrights and can sell licenses to it; novel business models (3D content hosting!); and luck — you’ll see that launching a new virtual world is anything but a piece of cake!

As a matter of fact, right now, and unless someone is working in a deep, hidden cave on the Next Best Thing, there is really just a competitor to Second Life that has all its features and characteristics, and adds a few more. Yes, it’s Second Life itself — but not the variety that Linden Lab developed, but, of course, OpenSimulator.

CC BY 4.0 OpenSimulator: The Choice for 2010 by Gwyneth Llewelyn is licensed under a Creative Commons Attribution 4.0 International License.

About Gwyneth Llewelyn

I'm just a virtual girl in a virtual world...

4 Pingbacks/Trackbacks

  • Great post Gwyn.
    I suspect your idea for LL to start to sell it own Opensim grid, will be hard for LL to swallow.

  • Great post Gwyneth.

    What caught my attention was the title. I just got done posting my 3D predictions for 2009. I won’t bore you with the whole thing but here is the part about VW’s

    2009 will not be the year of Virtual Worlds

    Some might say this was 2007. Certainly it was the year of virtual world realization and hype. The problem is that while we have the technologies in this area, we are just not progressing that fast. Don’t get me wrong. I think this is the future (OpenSim progress is exciting) its just on a longer path to becoming mainstream.

    2010 could be fun!

  • Apparently not, if this is true

  • Pingback: Dusan Writer’s Metaverse » OpenSim: The Places We Could Go()

  • Gwyn: So much to say, so I said it at here.

    But just to say: I’ll be circulating this to whoever will take a copy. I think this is the definitive post on OpenSim, at least as a “tick list” of what it is (and isn’t), how it works, and what to consider if you want to host your own (or visit someone else’s). Great links within the post as well – you’ve managed to aggregate some of the top posts and references elsewhere on the subject, including Prok’s summaries.

    I veer away from you a little on the latter parts and the implications for the Lab. Clearly, the Lab is launching stand-alone servers. What we don’t know is what the “estate controls” for those servers will look like: will they include the ability to tweak avatar or prim limits? Plug in your own voice or chat systems? It’s unclear – but for the ‘casual’ enterprise, the value proposition for using a Linden Lab “stand-alone server” (much like a Google search box) will be very difficult to resist. All the talk about modularized code and whatever will be lost: SL will be something you don’t need an engineering degree to plug in and start up (if the Lab does it right, of course, which is unknown – all these new guys over there seem to have SOME idea what they’re doing, at least based on their CVs).

    OpenSim will give you lots of options and, over time, may be more scalable. But if I’m a medium sized business and I’m told I can order a server, have it shipped (or hosted by an experienced 300-person company that is currently supporting 70,000 concurrent users), and I can have controls over things like log-in, identity, prim and avatar limits AND the technology has been proven through X number of user hours (what do you figure they could claim? 200 million user hours? a billion? whatever), and is fully supported by a profitable company with staff in 5 countries or whatever, which would YOU choose? Most companies won’t care that it costs something, or that you can’t code in Java.

    So, I’m not willing to say that the Lab is down for the count because OpenSim is cheap and free and will have as many deployment flavors as ice cream. The Lab has put the breaks on intergrid teleports because, well, they can: and this allows them to at least play catch-up, or secondly to get ahead of the innovation curve of OpenSim….they can worry about that next week, or next year – in the meantime, people like myself, trying to hypergrid all over the place with each server different from the last, with different bandwidth, with no inter-grid commerce: well, that’s a lot of catching up to do.

    Having said all that, you have still given us the post of the year.

    All the best for 2009. 🙂

  • Prokofy Neva

    Gwyn,

    I’m unimipressed. And not breatheless, and looking at this paragraph, amidst all your techno-babble:

    “Even permissions are properly handled. So, yes, you can bring your own content and rez it on the “foreign grids”, and it will still list you as the creator. Of course, remember always that the OpenSim manager will have access to their own databases and be able to copy your content, but, alas, there is nothing we can do about that. You’ll just be careful only to teleport to OpenSim grids you trust.”

    No. They aren’t properly handled. Listing you as the creator is trivial, and even fixed in places like LCO where they didn’t have the same list of names to start you with various kiosks authenticating you between worlds, etc. etc. And the issue isn’t ONLY that the OpenSim manager of the grid can copy your content, like Linden Lab can — but doesn’t. And the issue isn’t only that you have to find “trusted grids”. The issue is that they are not trustworthy. OpenSim has an ingrained wikiculture of communistic “sharing” (theft) and copyleftism and hatred of commerce. That bleeds into all the tools. They say arrogantly you can just, oh, code a module on top of that, but it’s bullshit, when the entire grid and its software have this ingrained in their code.

    Uh, “how” are you to become “careful”? I would have thought, for example, LCO was perfectly fine, it came with a good attitude toward IP, good rules, a seemingly normal TOS, good reputations associated. And whoops, it had a couple-three dramas that seriously called it into question, mainly because that all powerful Grid God or Grid Diva *can* copy your stuff and *will* — or at least keep it if they feel like it. But it’s not even about the grid owner. It’s about all the script kiddies and code bastards all over the new grids who could give a flying fuck about private property, currency, commerce, values, etc. — as long as daddy keeps paying the bill.

    And daddy *is* paying the bill, in the form of gadzillion tier-payers of Second Life, some of them with huge tier bills as rentals managers or big content creators, who are sustaining this entire shakey and idiotic enterprise of LL’s and OS’s to make this “Hypergrid” which is mainly a lot of hot air.

    I’ve TP’d to these OS sims in serveral services now. Where they work — which is a marginal percent of the time — you still face all kinds of shortfalls. No groups. No search. No reliability of TPs. No protection of IP.

    Morgaine Dinova, not surprisingly, with her Extropian and technocommunism views notably on record everywhere, hates the idea of all that owned land out there — private property — which, er, “The People” are not using “efficiently”. Well, uh, what’s the solution then, not to have land that people *pay tier on* always available, even if they don’t log on 24/7, for *when* they wish to log on and use that property *they pay monthly fees for*? Um, hello, Gwyn? YOUR plan? HER plan? Oh, make an algorithm that says, “well, that person logs in only about 27 percent of the time, so let’s keep their sim up only 27 percent of the time and likely we’ll get lucky that they never log in and never notice their sim is down”.

    It will be like the deliberate overbooking of airlines, and the already existing overbooking of sims (Lindens sell land parcelled into 512 bits, but can’t fit 64 avatars who could conceivably each own a separate parcel of that nature on to a Mainland sim at once). You come on, and find your sim has been snatched to feed the needs of some freebie sandbox place, and you don’t have it. You complain, and Concierge says, “oh, sorry sir, here it is, glad you called, we missed you.”

    Re: “This happens dynamically. Hosting providers give you next-to-infinite disk space to use, but charge you by traffic.”

    But…when you make a website and upload your content to it and pay some modest monthly fee…*it stays put*. You can be billed by increased traffic/bandwidth. You can be available based on some dynamic consideration. But your content is there.

    So, how do you manage a sim that is already live and active on the grid with content on it to become “only available sometimes”. Only by *taking it offline,* Gwyn. The Internet doesn’t have to work by taking webpages offline to make them less available; they just have to not have you be able to pull it up into your view once bandwidth taps out. But…how does that work on SL, hm? ‘Splain.

    Re: “All you need to do is to figure out how many sims you’ve got, divide by ten, and buy that amount of high-performance servers to give your users a good experience. The remaining 9 servers can be on “low performance servers” — until they’re needed.”

    So, uh, ok. Let me explain how THAT works, in reality. “When sims we like that are overloaded or that have our friends on it, or with loudly complaining foul-mouthed sailors from USS, we will lay on the CPU for them, and let the ordinary stiffs have a laggy mess.”

    In fact, what you are describing is what the Lindens *did*. They allowed tens of thousands of openspace sims out into the wild, with only a certain capacity, and relied on the system to just apportion the load balance in a sort of “creative” fashion. But they found that in fact, the more you give people, the more they take. All those sims were used to the hilt. Suddenly, the Lindens were faced with bandwidth demand they couldn’t control or assume would remain on its tekkie formulas of “1 out of 9”.

    Either you have land metaphors and a dedicated server/sim model, or you don’t. If it can’t scale, top it off, and start another grid. A lot of the grids simply will never need to talk to each other. If they do, they can use SLIM or Twitter or something — until the day when they can be safely and reasonably linked.

  • I wish I had time to do more than scan this right now – I can never figure out how some of you can be so prodigious on your blogs so consistently! Anyone have small kids?

    I do want to comment on RealXtend. It’s a step forward and two steps back. Why? OpenGL is cross-platform compatible. Direct3D is not. It really pisses me off that they’ve chosen to go a windows only route to enhance the opensource version of the LL compatible metaverse technologies. That’s like the antithesis of what opensource is about, IMO.

  • Dusan, I’m flattered and honoured by both your comment and your own blog post! No, I can hardly say this is the “definitive post” about OpenSim; in fact, I’m looking towards something written by either you, Tara5 Oh, or Tao Takashi that is much better — or that the OpenSim community once more exceeds expectations and totally renders this article obsolete. That would, indeed, be quite good news 🙂

    Like you, I’m pretty sure that the majority of corporations out there — and that will certainly be quite a high number — will just love to buy LL’s solution for “private grids” instead of relying upon the unstability of OpenSim. I can only expect that this might become one of the major sources of income for LL as soon as they launch that product, and this is excellent news. It is also a quite open-minded attitude by Linden Lab — I still feel terribly frustrated for being unable to buy a license of either Basecamp or Ning to run inside my own servers instead of relying on external hosting with a lack of features that my developers could quickly implement if we could just use their code… LL apparently has no problems in doing that, and the parallels you draw with Google are quite on the mark.

    Unlike you, however, I’m quite skeptical about the ability of LL to provide corporations with a more flexible product to run behind their firewalls. It’s just by looking at LL’s track record — they have gone totally the wrong way with their “monolithic” product — you can see, for instance, how integration with Vivox was done using an external application that runs side-by-side with the SL viewer. VoIP servers are also separate from servers running sims. I was pointed to a discussion with Zero Linden during his office hours where Zero, in spite of his flawless engineering background, actually sort of “defended” LL’s monolithic product, astonishing his technical-savvy audience, because of course Zero ought to know better…

    What these announcements seem to indicate is that LL is suddenly doing a 180º turn and recreate their own server software as a configurable product instead? Hmm. 4 years to upgrade Havok, and now all of the sudden they’ll be able to push a server version that can have pluggable external physical engines? Really, I’m quite, quite skeptic about that — I’ll believe it when I see it.

    No, instead, I can only believe that LL will give their corporate customers the choice of getting the server software as it is today and allow them to run private grids with all of today’s limitations and little else — at least not for several years. I’m sure that for many corporations, the technical support, as you say, will be the key selling point, not the “configurability”. Because, you see, if a corporation seriously wishes to develop things with LL’s technology, but capitalise on their internal developer force… they’ll find out that learning LSL is simply not worth the trouble, specially when all their developers will be ‘fluent’ in C#/.Net or Java/JavaScript. So, extending LL’s feature set inside a corporation’s own grid will hardly go via a deployment of LL’s software… and I don’t think that LL will even push their product that way, but just show them a feature list and say: “this is what our product does; here are our credentials (X billion hours of use, etc.); you can get it running inside your corporate firewall with this list of features and nothing else”.

    OpenSim-based grids will, however, be much like Linux distros — everybody will have their own “flavour”, most will be very unstable, most companies providing these services will fail and disappear after a uear or two — but they will all be interconnected, meaning that if something goes wrong you’ll be able to put your content elsewhere and start there from scratch. In a sense, it reminds me the days of 1994/5, where a lot of start-ups asked themselves “can we create an Internet Service Provider only based on open-source software, and run Linux/Apache/Squid/sendmail and compete with the Big Ones which use Microsoft’s software?” The answer in those days was a “yes”, but you surely had to have both the required amount of skills and a solid business model to succeed.

    Actually, I’m a bit skeptical about LL being able to play “catch-up” on OpenSim’s feature list in the middle-term (1-2 years). The reason is mostly because they’d have to drop their current Microsoft-like attitude towards product development, and go back to the days of 2004/5 where development was way faster. I don’t think they’ll do that — unless they split their teams, one going towards reckless innovation, the other towards a Microsoftesque approach to innovation. In fact, the best of both worlds would be to start with OpenSim and tweak it until it does what LL needs 😉

    As said, 2010 will be the interesting year, where you’ll be pitting LL’s obsolete (but field-proven) technology against a feature-rich and stable OpenSim 🙂 LL has two years or so to show what they can do until then… and I’ll certainly be closely watching their efforts 🙂

  • Prokofy, lots of separate issues in your comment 🙂

    First, a point of order: I’m not — yet — telling everybody to drop all their land on LL’s grid and run to the nearest OpenSim grid next to you 🙂 That would, at this stage, be premature and even irresponsible. I think that, at the current rate of development, OpenSim might be close to the 2004 stage of LL’s development in terms of stability — in 2010. The 2004 grid was “stable enough” for people to run business on it, even if obviously we have gone a long way since then. OpenSim is “not there yet” and will not be for at least one year (and likely only in two). And that’s just the software part of it; then you also have the infrastructure part to consider, and that’s where the wheat gets separated from the chaff.

    Let’s start with permissions. A year ago there were none, like there was no way to use a grid currency for anything. Today, we have very limited permissions (and no groups, so no group-based permissions, and without those OpenSim will never be able to compete) and a way to integrate an “economy server” not only throughout a single OpenSim-based grid, but across several grids with agreements between them. Permissions are, however, just one side of the story, the other side is trusting the company/individual running the service.

    Now, from a technological point of view, OpenSim will, sometime closer to 2010, be able to offer a full and rich permission set, at least at the same level as LL’s own. However, as you pointed out, that doesn’t mean that a company is “trustworthy” just because their software allows permissions.

    What makes a company “trustworthy”? I have no clue, besides them signing a contract with you to guarantee that, a contract that you can bring to a RL court and sue them if they break it. Long-term reputation obviously helps. For instance, I don’t “trust” Facebook or Twitter — I have no clue who the techies are behind it, and I happen to know that their shady terms of service actually help them to sell profiling data for third parties, without my explicit consent (on the other hand, I never removed explicitly my consent, either, so both companies are not doing anything “illegal”). When I upload pictures or movies to several social Web sites I cannot prevent the owners of those sites — who almost invariably started as a handful of geeks in their basements — to simply copy my content and make it available on the Web as their own. I might not even be able to track down what they’re doing with my content.

    Start-ups providing OpenSim service are not inherently malicious (“geee, people are such idiots, they register with us and leave all content for us to sell on XStreet SL with full perms, har har har”) but they also aren’t inherently trustworthy, either. They’re just start-ups building a reputation — like LL in 1999. Why should you trust LL with your content in June 2003 when they opened their service to the public? Because Philip and Cory were never involved in content theft before? What would assure you that they wouldn’t do it now?

    Why was under-the-stairs Linden Lab in June 2003 more “trustworthy” than any of the many OpenSim operators today? It’s way too easy to say, in 2009, that LL is a trustworthy company who knows that to keep their customers happy has to abstain from stealing their content. But… the truth is… it’s easy to say that in 2009, but it was impossible to claim the same in 2003. Or even 2004 or 2005. But that never prevented those millions of users to register with LL and trust their content would be safe.

    All start-ups have the same problem, Prokofy. I don’t know if Basecamp won’t randomly disconnect users “because they can”, and thus dumping all the project management task lists my company has — and basically ruining our business, since we would have no clue of what we have to do. But that never happened. I have to trust that the agreement with them was made in good faith. If not, the best I can do is to sue them. If I didn’t trust them, I’d use a different online project management provider.

    That’s what I meant with “caution” in picking an OpenSim provider. I’m sure that many are legitimate businesses and are just waiting to make a happy buck while the insatisfaction with LL grows for a while. A few are not legitimate. Most have no business experience at all and might fail just because they didn’t have a business model. Others might be clueless about providing services in a corporate environment. At this stage, it’s like shopping for hosting web sites in 1993 or 1994, when the Web was brand new. Whom should you trust? (Most of the web hosting providers of that time aren’t around any more, but a very small group actually survived). Even in 1993/4, not all of them were crooks; not all were clueless about business; not all were clueless about the technology (which was so new back then that few could ever claim to be “experts” anyway).

    This is the stage we are now with OpenSim hosting providers: a bunch of energic wannabe grid operators, a few of which are not crooks, technologically savvy, and solid businesspersons with a good track record of successfully ran businesses in the past. Most haven’t got all those skills, and at this stage — I would recommend to avoid them. If it’s too good to be true, it probably is.

    So, to recap — the lack of groups, incomplete permissions, search features (or classifieds), etc. are technological limitations, they aren’t available on OpenSim grids because the software simply hasn’t those features yet. Not yet, but one day they will. Just not today. The aspect of IP protection, overall control of griefers and script kiddies, and trustworthiness of the companies running those grids — are business issues. Like any start-up, they have zero references, and no history to validate their claims of trustworthiness. That’s a completely different story, and one faced by all new companies.

    Your comparison to what happens on Web hosting is mostly true for 2009, but it’s just because the technology behind web hosting has evolved dramatically in the past 15 years! In early 2008, a reputable hosting provider in Florida who hosted part of my company’s content, simply ceased to exist from one day to the next, never replying to emails or phone calls. A few weeks afterwards, even their Web address disappeared from DNS. What happened? I don’t have the slightest idea; but one thing is obvious, web hosting companies come and go, no matter how reputable they appear. You also pay what you get — if I wished to get 100% uptime and my money back, I’d have to host with Verio or Rackspace — and be prepared to get charged 4-10 times as much to what a regular web hosting provider charges. But I’d have more guarantees. With “cheap” hosting providers out there — and there are millions! — it’s always a question of luck and reading reviews and hope that they remain good for a while… Dreamhost was a mess with less than 95% uptime when I started hosting with them, but for US$9.95/month, I didn’t expect more. Today they charge even less and give me close to 99.5% reliability, a huge increase in performance. But… their servers can still fail… if I don’t keep backups, it’s my fault if everything disappears from one day to another, unless I’m prepared to sue them and at least get some money back that way.

    So, no, Web hosting is far from 100% reliable, even in 2009. I used to be happy with web hosting to have 70-80% reliability back in 1995 — for the price I was willing to pay, of course 🙂 I don’t expect a 100% reliable OpenSim-based grid in 2009. I don’t even expect LL’s grid to have more than 95% uptime! You surely remember back in late 2004 when whole sims disappeared, when it took days for a reboot, and when lost inventory became, well, lost forever, because LL didn’t even back up their servers regularly. They still charged us US$195/month a whole sim’s tier 😉

    What makes the difference in “reliability”? Well, the software part of it (ie. OpenSim itself) is just one of the issues. The rest is managing the servers, the bandwidth, and the infrastructure. This requires a completely different skill set than “being a cool programmer”. Granted, these days programmers do some infrastructure management, and infrastructure engineers do some programming, but they are two different skill sets, really. To run a reliable grid you need both, which is what LL has. I have no idea about the other OpenSim grids operators, although I seriously suspect that 3Di, for instance, has both kinds of people working for them.

    For some typical fallacies associated with running distributed systems, I have found this nice white paper. It’s pretty techy, but it will show a bit what infrastructure engineers have to deal with when deploying something like OpenSim…

    So, how do you manage a sim that is already live and active on the grid with content on it to become “only available sometimes”. Only by *taking it offline,* Gwyn.

    No no, not at all. Explaining in minute detail how exactly the process works — how to take a running application from one server to another while not losing any connections to it and within an unnoticeable time frame (ie. milliseconds) — is the subject handled in a semester on distributed applications in a computer science course. They’re subject to complex research papers on advanced techniques like this one. I can’t possibly explain all the details before boring you to death, but I can give you an example: when you’re running a script on an attachment and crossing sims, this is what happens: the sim you’re leaving saves the status of the script (ie. where it was running at the time), flags the next sim to get a copy of the script and of the status, and destroys the original running virtual machine for that sim. The new sim now has to launch its own virtual machine and start the script at the same point. And do that in milliseconds so that the resident crossing sim borders doesn’t perceive the delay to be very great (ideally, everything should happen under 250 ms, which is the threshold the human brain registers as being “instantaneous”).

    In fact, Second Life’s grid is the largest distributed application framework ever, where applications (scripts) migrate to different CPUs (even on different data centres) all the time, hundreds of times per minute as people cross sim borders or teleport elsewhere. This is an accomplishment by itself, one that is never fully grasped and appreciated (when Babbage Linden explains the process to a techy audience, they all open their mouths in disbelief since nobody has ever done this at this huge scale before).

    I haven’t evaluated 3Di’s software yet, but the point is, the process is fully automated, and the ideal “delay” for a sim to move out of an overloaded server and enter a fresh new one is probably very low. Avatars do not get disconnected. The sim does not go offline. You might have a temporary delay — since I’m sure that, no matter how good 3Di’s algorithm is, pushing a 150-MB database with hundreds of connections from one server to another has to have a large enough delay! — but probably it’ll be as short as an intra-grid teleport.

    So it’s not about having people’s sims offline and just sometimes get them up when the admin is willing to do so 🙂 You’ve quite misunderstood me on that. It’s like working with caches: textures get downloaded to the disk cache, from disk to memory (1000x faster), and from memory into VRAM (10x faster). You can only keep a few textures on VRAM, so you’ll just load there the ones being used. The rest stays on slower memory or even much slower disk space, ready to be swapped in if needed. This system of load balancing works a bit like that, although, granted, saving the sim’s state and keeping all connections to the avatars is far more complex than pushing textures from disk into memory into VRAM…

    Those ratios — 1000x slower, 10x slower — is what makes a good programmer decide how to fine-tune their algorithms to push textures around the computer’s many storage systems. A good caching mechanism (say, on a web server) will try to leave 1% of all data in fast memory and expect 90% of all hits to come from there; the remaining 10% of the hits will go to the 99% of data on disk. Those ratios are usually correct for most websites. For Second Life, I have no clue if the 1:10 ratio is correct or not, I haven’t done thorough statistics, but just took a look at the map at one instance and tried to gauge how many events were happening with more than 20 avatars, compared to sims with just 1 or 2 avatars. Under this technique, just the events hosted on sims with more than 20 avatars would be on “exclusive” high-performance servers while the event lasted — while the remaining sims could be on older servers, or have a server hold all 10 low-traffic sims in the same hardware. As soon as the event finishes, the system “swaps out” the sim and puts it back into the slower servers.

    This doesn’t guarantee a 100% use of CPU resources, of course, since guessing the correct ratio is not easy, and it fluctuates over time. For instance, there are far more events during the holidays and during the summer, as people have more time for them. During “peak hours” (say, noon SLT until 6 PM or so) there are also far more events than during the rest of the day. Juggling with the parameters to get the “best” ratio — the one that always gets all sims needing the performance at peak times into the super-fast servers — requires constant tweaking and is the work of a good infrastructure administrator 🙂 Those are not “tekkie formulas”, but simply a result of statistics — metrics you gather to understand how the load in your system is produced, and automatically allocate more resources to alleviate the load. There is some art in getting that right, but the end-results are usually worth the effort.

    Re: “All you need to do is to figure out how many sims you’ve got, divide by ten, and buy that amount of high-performance servers to give your users a good experience. The remaining 9 servers can be on “low performance servers” — until they’re needed.”

    So, uh, ok. Let me explain how THAT works, in reality. “When sims we like that are overloaded or that have our friends on it, or with loudly complaining foul-mouthed sailors from USS, we will lay on the CPU for them, and let the ordinary stiffs have a laggy mess.”

    That’s just the way you see things, Prok 🙂 You can argue that LL might have designed their texture cache algorithm so that Linden textures or FIC textures load faster (they might never get saved to disk cache, but always remain in memory). Or that their routers filter out IP addresses, and raise the QoS settings on them if they detect the IP address from a FIC and give them a larger slice of bandwidth. If you start thinking that way, gosh, everything is possible if you’re mean enough and your clients don’t suspect of what you’re doing 🙂 Oh yes, I worked for companies where the CEO had in fact better performance through the corporate VPN, so that he could show off how good his Internet access was during business meetings — while thousands of users had to grumble with “network problems” during those shows. Sure, that happens in real life too.

    On the other hand, that’s not the point at all 🙂 Let’s get back to web hosting for a while. If your site has no access at all, the CPU doesn’t need a slice of time to process anything — it’s just idle, waiting for requests. Idle? Not really — it’s serving requests for all other web sites on the same server. When it comes your turn, you’ll get your slice of the CPU — as big as you need to serve the requests. Now imagine that your site is so popular that a single CPU cannot handle the load — if you have a multiprocessor/multicore machine, it’ll be slicing time off other CPU’s as well, in an effort to better handle the load that way. However, during off-peak hours, it won’t need to allocate slices to your website — it can happily run other people’s websites during that time. System administrators will hope to get a mix of sites on the same server that don’t have all peaks at the same exact time — or, if they do, that they have enough CPU/memory to handle all requests simultaneously. If they did their maths wrongly, your site will be slow (but so will everybody else’s).

    High-end web server hosting providers use load balancers to use the same techniques to deal with an increased load. You might just be using one server’s resources if your load is light. Suddenly, your site gets slashdotted and the traffic increases a thousand times — a well designed load balancing system will immediately allocate as many copies of your webserver as possible, to handle all those requests. In fact, this is the type of service provided by Amazon’s AWS. It’s a quite fair system, since you just pay for the resources you actually need for your websites — if you need more, you can immediately — and automatically — get more CPUs to run your website, and fall back to just one server on off-peak hours. That’s the whole principle. Amazon, of course, charges for the resources you use — it’s not as if “only Amazon’s friends will benefit from more CPUs for their websites” while the rest of the paying customers have to endure lag on their webservers during a traffic spike. Not at all. Everybody can get this service, and pay only for what they use. If your website is idle most of the time, you don’t need many resources (as said, disk space is cheap), and you pay little. (Katharine Berry’s Ajaxlife uses AWS).

    In fact, what you are describing is what the Lindens *did*. They allowed tens of thousands of openspace sims out into the wild, with only a certain capacity, and relied on the system to just apportion the load balance in a sort of “creative” fashion.

    Linden Lab doesn’t have dynamic load balancing technology on their grid 🙂 and to the best of my knowledge it’s impossible to implement it on the current codebase, although if I were LL, I would certainly consider developing it…

    No, LL just did some maths based on their statistics, and calculated that, on average, one of their four-processor Class 5 servers is able to manage 60,000 prims and 400 simultaneous avatars, at best. This allows for 4 regular sims or 16 openspace sims per server. What they missed is intra-region communications. 16 sims don’t require 4 times as much intra-region communication as 4 sims, but probably 64 times as much, if the sims have the same usage. LL relied on openspace sims to have far less need of intra-region communications (based on their own usage of void sims in the past) since they would be “almost always empty”. Well, they totally miscalculated things. This is supposed to be the “official” reason for the price change, and technically, it’s a plausible one, although IMHO I think that there were far more “political” reasons than technical ones behind the decision…

    Putting just one sim per server is not just a political/business issue. Rack space has a cost — the more servers you have, the more you pay, obviously. Increasing density is always the trick to lower costs. This means that buying a server with twice the performance (but with the same footprint) is a solid business decision, as it allows to keep the same rack lease costs but serve twice the number of users. LL’s move from one sim per server to one sim per CPU is purely arbitrary, but somehow fundamented on their internal metrics, of course. Overloaded sims would probably be far better off with more CPUs (and more bandwidth) per sim, not just one.

    This, of course, is not related to the land metaphor at all. In spite of many disagreeing, I still think that having a metaphor instead of charging per CPU and for bandwidth is far more reasonable for a virtual world. In fact, I cannot understand how possibly a different model would work — how much bandwidth ought a resident hire from LL if they have a 300-prim house? How do you “translate” building sizes into CPU/bandwidth? Ah well, I guess that the same issue will arise when meshes are introduced later this year, and I see no other option than moving from a prim-based economy into a polygon-based one… but still keeping the land metaphor intact, of course.

  • A very interesting blog post on the Opensim software. Good to see it tickles your interest and more and more people start to see the importance of it. Opensim will give the people of the Metaverse a choice and they will no longer have to abide the monopoly abuse by a single company in order to have a virtual life. 2009 will become an interesting year for the further development of the Opensim project which is going very well.

  • Wonderful points Gwyn – I’ll come back for a read of your response to Prok…I’m side-stepping some of those issues right now, although I touched on policy in my own post.

    Regarding the Lab – I totally agree with you. I’m not proposing that the Lab’s solution will be flexible, nor particularly configurable. What I’m proposing is that in the near-term when enterprise makes the choice in which they need to balance out stability, service (for what it is) and “proven” versus configurable and codeable they’ll go for the first: for now. Many of them won’t care about that coding bit you mention as long as they can save travel costs or whatever – it’s only LATER when they start saying “OK, now that we’ve held virtual meetings, what ELSE can we do” and start wondering about product prototyping or data visualization or project architectures or whatever that they’ll bump up against the limitations of monolithic code.

    But I’d still like to hear a clear list of use cases that OpenSim currently supports that LL can’t. Physics, sure. Anything else? I’m not being snide, I’m sure there’s tons, I’d just like to see a good list started somewhere.

    The Lab will offer sufficient configurability, they’ll add bells and whistles, HTTP-In and, who knows, HTML on a prim, whatever else. This will add a new layer of innovation, and then look for far more development Web-side over the next year. Take Justin’s work, the Web-side integration, packaging up mobile phone SMS, calendar functions, all that stuff and expect to see more. I’m actually of the belief that there’s a sufficiently high innovation ceiling on the Lab’s platform that hasn’t even been tapped yet – and if Mark is true to his word and we see “new tools for commerce and content development” then the ceiling will rise slightly higher.

    Having said all that, I alluded to this in my post that I think the innovation on OpenSim won’t be because it’s a “deeply configurable” SL imitation Grid, but when they break loose a bit more (not that they haven’t already – what was that Intel project with physics?) and demonstrate configurations that are TRULY different – not how it’s scripted or deployed, but what use cases it can support that we haven’t seen before.

    OK. Now, back to reading your discussion with Prok.

  • My goodness. OK, just read Prok’s comments and your reply – who knew New Years Day would be so, um, exciting!

    Now…I’m not going to get into a parsing here, but first Gwyn, your reply to Prok is pretty astounding, clearing up some of the misconceptions I’ve always had and reminding me, yet again, that crossing sim boundaries is in fact one of the great technical miracles of our time. It’s also one of the great barriers to concurrency – it’s not having 100 avatars on a sim that’s the problem, it’s having them arrive around the same time, what with all those virtual machines or other fancy codey things that need to happen. (Thanks to Sidewinder for that last bit, btw, he gave me a cool little lesson on this).

    In any case, on many of your points, I agree, as best a non-techy can with those techy bits, and the difference between the technology and the business issues. But I still take issue with the general thesis, I’m afraid. And the thesis is something like this:

    – Code is code
    – Policy and business issues are complicated and will require code, once its written, to enable different expressions and approaches to this
    – Therefore, code will align with policy over time, because code is flexible, and modular, and business will demand it, therefore it will BE.

    Now, I brought this up at the VW conference in LA at the OpenSim session, and I felt like I was being hectored and lectured by Frisby who, sure, is a little boy genius and whatever, but it made me feel like one of those people who just can’t get their kids to listen to them, I’ve BEEN THERE dammit, but they need to learn for themselves, I suppose, and take us all down one more time, history doomed to repeat itself or whatever.

    My point in LA was the same as it has always been: yes, code is separate from policy, but that doesn’t mean they should be decoupled. I use the example of e-mail. I got the Frisby lecture in return, which focused on protocols and a lesson on how e-mail works, which was totally irrelevant to the main issue, which is that e-mail was constructed in a way that doesn’t allow us, now, to somehow figure out how to get rid of all the SPAM I get. It’s maybe a tangential example, but it’s an understandable one: spam is a policy issue. E-mail is delivered because of code. If the code had been developed with an acknowledgment of this policy issue, (I’m not saying it wasn’t, I don’t even know), or if they had been able to PREDICT it, we would have better control over spam (whether that’s good or bad I won’t debate).

    Now, content protection: we KNOW, we can PREDICT, that this will be an issue. That it will be a policy issue, a legal one, a “trust” one. It’s a cop-out to say “sure, but digital content in GENERAL is an issue, why should WE solve it?” Because the answer is: we CAN. And virtual worlds can demonstrate that they’re not only a platform for innovation related to 3D space, but for innovations in content protection as well. And this means going deep into these business and policy issues NOW, at a time when the code is still sufficiently flexible to allow a deep think on what innovation around this COULD look like, before the philosophy of the people building what they’re building gets so far along, and so entrenched, that it’s too late for innovation of this sort.

    I mean look – what’s going to happen will be like e-mail. Once it’s out there and people are using it, you think they’ll swap out all that installed technology to make room for, what, identity and IP protection systems that benefit the USERS instead of the platform owners? Doubtful.

    Imagine a version of the Internet where the copyleftists could live happily beside the publishing companies or music companies or whoever else. Hmm. Can you say iTunes? Maybe not perfect, but it shows how innovation can be derived from rethinking business models for content in a world in which, supposedly, all content is digital and should be free.

    What about Better Places to go even further afield – forget cars, think about how cars are powered, build your business model on that. The received orthodoxies are simply that: received. It doesn’t make them true, and this supposedly highly innovative community is simply receiving orthodoxies instead of pushing the envelope a bit more.

    I mentioned this on my own post: look at Raph Koster, he didn’t start to code Metaplace until he had thought through all these business issues. His model for avatar rights, content protection and commerce is as deeply embedded in his architecture as the mind set of the current OpenSim’s coders is embedded in it. They’ve decided these issues are for future generations to figure out, or business, whoever – and they know better, why, they’re even smarter than Raph Koster I guess.

    My feeling is that OpenSim, by continually deferring these issues of content protection and policy, because we’re “not there yet” are missing one of the most meaningful opportunities we may ever have to deeply innovate around issues that are far more central to our times than prim counts or concurrency. Issues related to identity, trust and content and with them the opportunities for collaboration, product development, licensing of creative output – these are THE hotbeds for our era.

    Companies will be destroyed and empires built by those who can solve these issues. OpenSim, by generally putting to the back burner discussions of content, commerce and policy are denying themselves one of the key innovation opportunities presented by virtual worlds, as so ably demonstrated by the success of the Lab which, as you point out, wasn’t built on TECHNOLOGY but built on the policy decisions they practically made by accident: why, this time, does it need to be accidental again?

    So, you argue:

    “So, to recap — the lack of groups, incomplete permissions, search features (or classifieds), etc. are technological limitations, they aren’t available on OpenSim grids because the software simply hasn’t those features yet. Not yet, but one day they will. Just not today. The aspect of IP protection, overall control of griefers and script kiddies, and trustworthiness of the companies running those grids — are business issues. Like any start-up, they have zero references, and no history to validate their claims of trustworthiness. That’s a completely different story, and one faced by all new companies.”

    And I respectfully disagree. Or, to quote someone who’s thought about this a bit more than I have:

    “It’s about all the script kiddies and code bastards all over the new grids who could give a flying fuck about private property, currency, commerce, values, etc”

  • Tish Shute

    Wonderful post Gwyn. Really shows that OpenSim has “arrived!”

    Something to keep in mind, I think, when thinking about OpenSim is that: OpenSim, while it looks like Second Life, acts like Second Life, and even quacks like Second Life in some ways; it is actually a VERY different beast to Second Life. Crista Lopes the developer of the hypergrid code for OpenSim said this to me recently (as have other core OpenSim developers at various times).

    Anyway I think just how different OpenSim is will become more and more apparent as it matures. As you note the modular architecture is a key difference and this will have big implications as the community of developers expands.

    So while Dusan would like to see a list of “features that OpenSim supports that LL doesn’t” more to the point, perhaps, is that OpenSim will be able to support features/integrations/mashups that cannot even be anticipated yet.

  • Tish – I think we’re in agreement. We’re already seeing applications, mash-ups, and wonderful experiments. I suppose my point here was to make sure that we not confuse OpenSim with SL when we discuss feature sets and advantages. If we focus on prim limits and concurrency we get stuck in discussions about whether the hosts are more or less trustworthy than the Lab. If we talk about OpenSim as a place to test new physics models, opportunities for real time prototyping, the advantages to shared inventories for co-creation – then we start worrying less about whether OpenSim “worlds” look like the SL world.

    I use the term “features that LL doesn’t” to draw the analogy. Having said that: is there such a list? Forget about comparisons to the Lab – and I’ve found bits and pieces on different blogs, but I’d really love to see more documentation and sharing around this. Or a point in the right direction.

    I keep checking your blog Tish to find out the latest, of course. 🙂 But maybe there’s an aggregated summary of different experiments and tests that I’ve missed.

    Now – what about the policy issues Tish? As I’ve mentioned before, your post with the interview with Eben Moglen was one of my turning point moments in virtual worlds. It pointed in the direction of really deep and compelling opportunities around identity, trust and so on – are we seeing ANY movement in these areas? (Other than OpenID…ick).

    And PLEASE refute my claim that policy does not HAVE to follow code. In fact, that discussions of these issues can facilitate deep innovation ahead of code being executed, distributed, hosted, and ubiquitous. I keep hearing the arguments, and then I hear these things about standards and a growing sense that they are starting to emerge – but someone please set me straight on all of this, because I still haven’t heard anything that sounds like a valid argument for what harm there would be in having more meaningful discussion around these issues – unless, again, I’m missing the really cool meetings where this stuff is being talked about without the room being dominated by folks solely interested in code.

  • Hey all. I THINK I’ve read much of what’s been said. Some serious foot shooting going on here and there, but over all, a great discussion starter, obviously. Thanks Gwyneth for this.

    Gwy and I spoke recently about OpenSim Grid platform. My partner and I, who just happens to be a copyright, IP and patent attorney out of Washington, DC., have discussed these issues of content protection … a lot! Especially since we’re preparing to open our own grid and have a lot of interest from bran name designers. I’d like to shed some light on this subject, if I can.

    CONTENT PROTECTION

  • Dag NabIt! My long nail hit the wrong key and made this submit before I was done. 😛 Let me try again.

    CONTENT PROTECTION:

    Its a bugger of a topic and can be the cause of endless drama anywhere there are permissions – probably one of the reasons the code boys … and girls, don’t wanna think about it. Lets face it, there are no easy answers. Everyone is going to have to accept a compromise of sorts for anything to trying be done about it.

    The content designers are going to have to accept that no matter how much you try there will be some deviously smart coder who will find a means to circumvent any and all methods to protection content. Why? Because that’s how they are wired. Make it hard and its a freakin’ challenge to them.

    Does that then mean you don’t even address the issue? No. If we did that in real life, banks would not have sophisticated security systems, stores would not have those detection thingies at the doorways. If such measured weren’t in place and stores made no efforts to compel honesty, i.e., store clerks and security guards, video cameras and round mirrors, then honestly folks, the vast majority if the population would be inclined to pinch things when their pockets got lean. Its just human nature and we go to great lengths to justify it in our minds. There’s a whole psychology to this.

    I steal cause Walmarts runs small businesses to the ground and they deserve this!

    I steal because they have insurance to cover it.

    I steal because they make millions and I make $6 an hour. Why shouldn’t I be able to own that expensive sweater?

    I steal because I hate big business. Its a political statement.

    In reality, people steal because they want stuff they CAN afford, but don’t WANT to spend the money for. And, as long as they can do it undetected its a big thrill!

    Now, lets put all the psychology to work. Knowing this, how do you combat this? In the real world designers and retailers have to estimate the rate of loss in sales due to theft. It runs anywhere from 10 to 30 percent! And that’s real world product. Those stores that have the better security systems and vigilant and honest store clerks are the ones that keep the percentage lowest, but even then they lose revenues.

    Content providers in our digital worlds have to accept that some loss is going to happen, just as real world designers and retailers do, and price accordingly. Us grid owners and coders have to come up with a means to limit that risk as much is feasibly possible. HOW? EASY!

    Yes, I said easy. You just make it so frustrating to try and do that the MAJORITY walk away. Again, there are always going to be those few individual who have the knowledge and expertise, and frankly, the moral bankruptcy to steal – no matter what. They don’t consider the person behind the product. That what they do virtually may be their sole income to pay their bills and feed their children. There are a lot of us out there. That message needs to be sent out to the user community in a palatable way that kinda shames those people that would steal and maybe compel them to be honest and pay for what they want.

    For others, its a sense of risk of being caught. Granted, for some that’s the whole point, but if the risk seems imminent or ruins what they want – in some way announces that the garment they just ripped out of the code was not gotten properly, then it becomes tainted goods. Its like wearing a billboard saying “I am a digital thief” and they’ll be less likely to do it again.

    I think we also have to look at what the person’s intent after the dirty deed is done. Do they rip a skin to learn how to make one themselves – find all the sweet spots and seam edges of a skin and learn how to make those disappear but looking at how another artist did it? I personally have no problems with this. Its when they cut and paste MaryJo’s boobies to Mark’s skin with the abs, and resell it under their own name that I have a problem with it.

    That’s not learning. That’s not even being creative. That’s just doing the great shuffle to get something different. Now, if you bought 5 skins cause you liked a part of all five and wanted to make a custom skin, melding them all into one, that again I have no real problem with this concept. You’ve respected the artists work by buying them all. But for gosh sake, don’t go telling everyone how you did it, do it for others or resell it as your own work. Pushing a bit of smudge around to make disparaging parts meld together does not constitute artistic talent or skill. It just means you know how to use a tool.

    This is why, when I do play Frankenstein with other’s textures, like TRU’s stuff, I rename them with the original name and add UnReal’s onto it. Example: TRU BrickWall1a. I torture it to the point that even the original artist might have a hard time laying claim to it, but STILL I rename it like this: TRU/UnReal BrickWall1a2. This gives both artist recognition for our efforts and … well … its honest! I didn’t draw that original work. I changed parts, added parts from my photography, added lighting and shadows and tinted. I spent a considerable amount of time making it work for my needs, but that doesn’t take away from the time the original artist put into it. That is one of the main reasons I’ve never sold my textures. I am a morphing fool. It simply would be wrong to do such a thing.

    Ok, enough about defining what is stealing, the psychology behind it, and the intents afterwards. Lets get down to content protection. It can be done and we’re working on it right now. The easiest way to do it immediately is to keep your grid closed, which we will be doing for a while till we can reasonable assure our designers they have tools to use to protect their work and let it out into the other grids. But that will demand that the other grids establish and/or follow some agreed upon standards on how to deal with content protection. They’d have to adopt and respect any tools used to protect content, such as, integrating them into their grid system, if needed, all the way back to the server side. They’ll also have to agree to POLICIES on how to deal with content theft.

    We’ll be establishing a community panel made up of users, designers and IP/copyright experts to hear cases of suspected theft so both sides can be heard by members of their own community. This panel’s objective will be to get the facts out on the table in a open format that will encourage QUIET observers, look at the situation, and then make a recommendation to both parties as to whether, in their opinion, it has legs in terms of real world legalities. If they find it has, they will offer a low cost way for the offended party to file a legal claim – we’re hoping this will run about $150 usd per claim – and point them to a directory of able and digitally savvy IP and copyright lawyers to either help mediate or if necessary, take it all the way to court. This panel can also recommend suspension or banning from the grid and OpenSim Grids have wide breath of options when it comes to banning. I think, in the face of that kind of inspection, and direct access to legal options, most people prone to theft as a means to pad their virtual wallets will seriously reconsider coming to our grid, or trying to steal content and bring to theirs. The difference is we’re not standing up and letting it happen. We’re making it clear, via these type missives, and in our TOS that we’re not going to be hands off on this topic. We feel content is half the reason for the success of Second Life. Without original well designed content we’d all be sitting around in our sad slider hair and clothes and looking like we just migrated from ActiveWorlds with only the primy clothing on our backs *snickers* We won’t become judge and jury from a legal point of view, as grid operators and developers, but we will give encourage our community to become a part of the solution and support their efforts in this, as well as provide sim property and tools to make it happen.

    We will allow other “Trusted Grids” to link up to use eventually, but they will have to sign a TOS with our grid and pay a monthly fee to do so. It won’t be a huge fee, but large enough to cut out any single home grown grids that might use such a connection to us as a means to grab content illegally, then unlink to us and sell on their grid. The revenues will help to make continued advances in the content protection tools, public awareness marketing and support of these panels. These grids will also be required to setup these same anti-theft panels, or seriously encourage their community to participate in ours, when matters of theft get reported.

    These are just some of our ideas, mostly policy in nature mentioned here. Obviously we aren’t going to discuss the code tools we’re working on until we have something definitive to show, but rest assured there are ways to frustrate those who would wish to get free stuff instead of paying to much that they give up. That is our real objective – to make 80-90% of the population go … man! No way! That’s harder than making it myself! Never mind the risk of being hauled in front of everyone to answer questions about how I got the stuff? I’ll just go buy it instead.

    Would love to hear CONSTRUCTED critics opinions, but please … no need to curse like a sailor as you go at it. Just sounds so very angry. Plenty of things to be mad about today without us inventing new things to curse and storm over. No reason to dip into that tone when we’re just throwing ideas and around here and trying to brainstorm solutions together. I do want to hear about any holes ya’ll might see in our ideas. After all, 20 heads x 2 are far better, huh? *-)

  • Pingback: Rating my predictions for SL over the last year - Part 3 « Pavig Lok()

  • Nice post Gwyn – you’re probably even more optimistic about the future of OpenSim than I am (and I’m a core OpenSim developer!). I don’t think anybody really knows the destination or even the waypoints for the OpenSim road, but the journey is going to be mighty interesting.

    I do have to pick up on a few things, though

    * A good proportion of bug fixing is much more difficult than feature coding. I have to disagree that the bugs will be picked up by ‘less experienced OpenSim developers’. OpenSim is a compliated beast and some of these bugs (especially race conditions) are things that only really really experienced developers can fix – feature coding is easy-peasy by comparison. If OpenSim is going to succeed then I believe that we *must* have good unit, functional and integration test suites.

    * Some of the modules you mention are probably not quite ready for prime time. I believe the Asterisk module has issues right now with the unavailability of a drop in replacement for the proprietary Vivox voice code. The SVN and ContentManagementSystem modules are there but may have suffered from some bitrot (they haven’t been worked on too recently and I haven’t heard from anybody who is running them). This isn’t meant to disparage such modules – it’s just an attempt at expectations management (alpha, alpha, alpha!). The OpenSim ‘forge’ (http://forge.opensimulator.org) is also an excellent place to look for modules (or to start your own).

    * You almost seem to imply that there are already LL employees working in the OpenSim core. To my knowledge this is not the case. I’m guessing you’re really saying that it might make sense for Linden Lab to contribute people who can do quality development to get them into OpenSim core 🙂

    On that note, I have to express skepticism that Linden Lab would rebase on OpenSim any time soon. As you say, they have hundreds of employees and some very clever software engineers, working on a platform which has had various bugs and scalability issues beaten out of it. This is no small feat.

    I think what’s really triggered the growth is the presence of a known viewer that OpenSim can be coded against. This means that lots of developers can work in relative isolation without having to decide where they’re going – the future is mapped out for them. As OpenSim matures, it’s going to be very interesting to see how we can progress beyond simply filling in the blanks.

    One thing which may really spur this is if work took place to make another 3D viewer using a completely different protocol work with OpenSim. Obviously, things can’t be too different (going to an OpenCroquet model is probably too much of a stretch), but there’s potentially a high level of flexibility available for a completely different viewer – after all realXtend have shown that it’s possible to contort OpenSim enough to support stuff such as meshes rather than simple prims. One candidate viewer is that used by Sun’s Wonderland, although with that we face the GPL issues which continue to make it difficult to code OpenSim for the Linden Labs Second Life viewer.

    Having a completely different viewer use OpenSim would greatly encourage the genericization of the underlying code and make OpenSim a much more universal solution rather than filling in the narrow niche of being an alternative Second Life server. Indeed, this is one of the project’s explicit aims. And if this starts to happen in 2009 then I think that we are really going places.

  • Wow, I’m quite glad that the discussion turned from the “pure technology” behind OpenSim to what can actually be done with it, and hints about its policy…! Thank you guys for your very insightful comments and explanations.

    Dusan, you’re quite right about all your points regarding policy. From my point of view, of course, OpenSim is “just the technology” — e.g. I have as many worries about it as Adobe has when Photoshop is used to allow people to steal pictures and change them to look as their own. Adobe doesn’t have a “policy” for dealing with copyright content, that is, if a known photo pirate is caught, nobody can sue Adobe because they didn’t care that their technology has been used for illegitimate purposes.

    Similarly — and this goes for lots of technological solutions — you can’t sue Nokia or Motorola for making mobile phones used by terrorist cells and drug dealers to coordinate their criminal efforts; or Microsoft for developing server software that allows extreme-right or extreme-left groups to publish their blogs. Technology is “neutral”; the use of it, of course, requires legislation/protection/policies.

    LL has access to 2.2 billion items in their asset servers. At any point in time, any of the 300 Linden employees could, theoretically, download them and upload copies of the content “as if done by Lindens” and, say, give them away as freebies on the Help Island. But we all know they won’t do that. Why? Well, mostly because LL is a reputable, trustworthy company, and has a contract with all their customers guaranteeing that LL will not copy our content but allow you to retain your IP licenses on it. Forever. Without any charges.

    So it’s just a matter of trusting LL on good faith — because if they break that good faith, they’re liable in court.

    Now, OpenSim is not a “company providing a virtual world”, nor an organisation of people that are building a competitor to SL. It’s just — code. Technology. Tools. A platform. What will happen is that some companies providing services using OpenSim will naturally have to include their policies in the agreements they have with their own customers — just like LL does with their residents. The option to “trust” these startups is as hard as the option people made when trusting the 20 or 30-employee company in a garage in San Francisco back in June 2003. LL was not “more reputable” back then as any of the start-up OpenSim providers are in 2009 (well, Philip had a reasonably good track record… but he could have “turned evil” 😉 ).

    So when we talk about “interoperability” we’re actually talking about two levels here. One is purely technological — your comparison with email is incidentally quite on the spot, that’s exactly why we have spam today! — and there are limits to what “code” can actually protect, specially if it’s relatively easy to subvert code, and if we want a totally decentralised model (ie. anyone can interconnect to anyone else, and it’s just up to those two parties interconnecting to forge an agreement; no third party acts as middle-man). The other is simply an issue of business and company/consumer relationships, where policy becomes part of the agreement.

    What this mostly means is pretty similar to what happens on the Internet. For instance, we have seen all those neat mashups where you log in to one service and grant another service full or partial access to your data. Typical examples are logging in with OpenID, adding applications to Facebook, using your Yahoo Account to enter Flickr or your Google Account to enter YouTube, and so on. When in these cases you’re crossing the company boundary (granted, the examples about Flickr and YouTube are “inside the company boundary”) things become interesting — since you have to trust both companies, the one where all your data is stored (you’re assuming that only the data you wish to push over to the other company is going to be visible and nothing else), and the one you’re logging in to (where you have to assume that they won’t just get all your data — including the password — and use it maliciously). What “policies” are there to safeguard the process? In fact, absolutely none — except two companies’ terms of services, with which you have to agree, or, well, forfeit the easy integration and interconnection.

    At this stage, that’s all we need for the metaverse or intergrid.

    When I say that I recommend “caution” is just because at this stage, many small OpenSim grid providers are virtually “unknowns” and you have no way to validate if they’re going to be honest with their own terms of service and policies. However, why do people automatically trust anything on the Web but, at the same time, automatically mistrust anything on the virtual worlds? Right now, I have thousands of pictures on my (other) Flickr accounts that might being used maliciously by thousands of people, selling those images as if they were theirs, using them inappropriately on third-party web sites without my permission… and I have absolutely no idea of what goes on! Sure, I have posted those pictures with an Attribution CC license, but how do I track down anyone who ignores the license and refuses to comply with my requests? Even if I managed to find out who’s doing things with my pictures, how would I even start the whole lawsuit process against the offending parties? Should I file a suit against Yahoo because they allow Flickr to display my pictures and anyone to copy them?

    Similarly, if someone develops a game and sells them online (through one of the many websites that sell microgames, for example), how can you track down if someone is doing multiple copies of your game and giving it away to friends? Granted, you can invent some kind of licensing scheme, but how will you track down crackers and key generators that are a keystroke away in Google? The answer is, of course, that you can’t. And, of course, I won’t even get into the discussion about copying music and videos… We took 15 years to get used to the idea that everything posted on the Internet is there to be grabbed and used maliciously by others in one form or another — and, surprisingly, the crime rate and the malicious use is actually negligible in most cases. Music, as we all know, is the big exception that almost brought the worldwide music industry to its knees — but in spite of that, musicians are surviving, even if possibly the labels aren’t making such a profit as during the pre-Napster years.

    Of course, everybody will be saying that “this is completely different” and people know that posting their content on a Web page is “unsafe” (to the degree that it can always be copied) unless you have a host of lawyers working for you, some lobbyists at the many governments, and are thus able to enforce your rights.

    Second Life is not different from that. The message to be passed is simple: when logging into another OpenSim grid, beware. You’re going to take a risk. A serious risk. Try to figure out the reputation of the grid operator you’re logging in to before you regret your choice. But in reality, I can imagine that with few exceptions, all other grids out there are “as safe” as LL’s own — which, as we all know, is not “safe” from content copy, and very likely will never be, unless LL reverts their decision of releasing an open source client and close down the access to their grid to official SL clients again. This is not likely to happen.

    Code rarely can replace law. And, in some cases, it shouldn’t even attempt to do that. If enough people are scammed by a grid operator, it’ll be quickly the target of someone who can afford a lawyer.

    Tessa’s model is precisely the way I see it happening — pretty much as so many other things have been created on the Internet in the past. A “closed model” where the best and largest grid operators establish a board of trust, and only accept members willing to enforce a common set of policies. If you abide by the overall set of policies (and paying a fee is obviouly a good idea), you’ll remain interconnected. As soon as you start violating the rules, you’re out — and no amount of public drama will get you back in, while, on the other hand, abiding by the common set of policies and enforcing them will definitely allow that.

    In fact, if I read Tessa’s proposals correctly, her suggestions will probably be way easier to enforce on non-LL grids (where content theft is a quite low priority), since all those grid operators will wish to maintain their reputation and reliablilty as trustworthy operators, and will go through pains to keep high ethical standards in order to attract more customers. So I believe that this is the way to go ahead.

    I’ll reserve my opinion about Raph Koster’s ideas for another day 😉

    @Justin, thanks for your many corrections, and your very insightful comments. Oh yes, I know I’m quite an optimist 🙂 Still, even if some of the modules are at a pre-alpha stage, the good news is that they’re there and can be “seen” (even if they totally and completely crash your grid 😉 ), which is far more than what we can expect from LL at this stage…

    I’m also aware that I tend to mix up in my text OpenSim as it was in late 2008 and how I expect it to be by late 2010. For instance, Asterisk integration is something I haven’t tested yet, but I can only imagine — from the comments in the configuration files! — that it’s obviously not production-ready yet. Or not even at an alpha stage. But… the point is… it will be. All it takes right now is to get a company very interested in the possibility to use OpenSim for exactly that purpose, download OpenSim, and start to develop the appropriate module starting from what’s already existing 🙂 Well, it might take a year or two, but… at least it will be available. One day.

    I found it amusing that you consider the lack of “your own client” a limitation of OpenSim and that “replicating SL” is a “niche market”. Well. Looking at the current population of virtual world users, I’d say that SL has perhaps a quarter of that market (taking into account IMVU’s 20 million or more registered users), so it’s not quite a “niche market”. If we go further and look at all virtual worlds with user-generated content, persistence, and contiguity, well, SL has 100% of that market (or, well, 99.999%). That hardly qualifies as a “niche market”. I understand that being able to use OpenSim to emulate, say, World of Warcraft, Sony Home, There.com, IMVU, realXtend, OpenCroquet, and Second Life, all running simultaneously and being served from the same software would truly be the ultimate goal — the “Universal Virtual World Simulation Software”. That would be insanely cool, of course, but for the next 5-10 years, I’ll be more than happy with something far less ambitious but with higher practical use: full compatibility with Second Life with lots of plugins to do what LL will never manage to do with their software. That would be “good enough” for me.

    A last comment. Since the past two years or so, using the pretext of “we’re so bad at deadlines that we’ll never tell what we’re working on”, Linden Lab pretty much stopped letting us know on what they’re working on. Allegedly they have 200 developers — surely not all are fixing bugs and doing quality assurance tests?… So… what are they doing? Turning LL’s sim server software into a modular, plug-in-based solution? Programming Second Life 2.0? Well, I can hardly believe any of that… so I’ll be stepping back and wait to see what they’re really doing…

  • Hiri Nurmi

    Another one of Gwyneth’s stunningly interesting articles. If there’s *one* SL blog I follow religiously and quote widely this is the one.

    But..

    I’m not at all convinced yet by the OpenSim alternative for widespread general use. Not because of any technological issues – I’ve been loading OpenSim builds onto my own hardware and exploring them for the past year, and as Gwyn says the progress has been amazing (although I would point to the known rule of software that the last 5% of development always takes up 50% of the time). I’ve even recommended it as something to explore to a client interested in producing an interactive archaeological model.

    More my issues are that I don’t see any feasible mechanism for a substantial migration of any community from SL. Quite the reverse.

    To explain my context. RL I’m professionally involved with graphics software, and in SL I operate as a designer specializing in prim clothing for Roleplay – about half Gor, the rest medieval, SF, cyperpunk etc. I’m not a ‘big name designer’ by any means but I do nicely enough to have good brand recognition (within the communities I operate in), support my own sim, and take a useful RL income back – enough to the point where I would think very carefully before I would risk it by needless exposure to copy protection issues. I approach running my business in SL with the same degree of seriousness that I run a business in RL.

    Now, in many ways the RP communities are perfect candidates to migrate to OpenSim grids. Gor for example had over 300 dedicated sims on the main grid at the last count, several of which are in small continents (and wouldn’t want to be connected to the main grid anyway). Roleplayers tend to be at the more dedicated ends of SL residents – it’s generally something you don’t find until you’ve been around the grid for a fair while and most people dip their toes in it a few times in different environments before settling into it. It’s also highly community orientated so people tend to make friends and stick around.

    On the face of it a RP community would be a perfect test case for migrating a group of residents en-masse off the SL grid onto a set of dedicated OpenSim servers. Commited, experienced SL users, good community, and they operate on the grid like they were half off it already.

    Except two factors run heavily against…

    Firstly all RP communities need new blood. Sometimes is people new to RP, sometimes it’s people moving between communities. In most RP sims it’s quite common to see normally dressed avatars wandering around with ‘Observer’ tags. Or often people will set go and play in a different environment for a while as a change – the Arabian Night’s or Medieval Worlds sims seems to be a common diversion from Gor for example – or I’ve know people swap between Vikings and Jedi within the hour. The drag against OpenSim here and for SL is that in SL the barriers between environments is just a teleport, whereas use OpenSim and the potential player has to both find the alternative grid and set up the client to go to it. It may not sound that much to a techie, but the difference between the success of Flickr and the failure of previous photosharing sites was reputably one mouse click.

    Secondly, and even more important, is content. RP sims are often among the best constructed in SL and the general standard of avatar clothing and equipment is also very, very high. This isn’t accidental of course – good builds attract players and Roleplayers are basically playing Amateur Dramatics online – so they like to play the part and show off. An OpenSim grid without great content would simply not attract players, you have to get the good content providers across first, in sufficient numbers, to support the new grid community. And this isn’t something where you can half succeed – a RP sim needs a critical mass of players or it fails completely.

    So to make your OS RP sims work you need to get the content providers in. And sure as one myself I’m interested, but to be worthwhile I have to know

    1.That the amount of effort that I put into adding items into your new OS grid is going to be financially worthwhile – as compared to spending the same amount of time releasing new lines in SL. That means I have to be confident you have a working, trustworthy, payment system in place and there are going to be enough people present to buy with it.
    2.That I can trust your permissions system and your integrity. Gwyn says this is true of any new startup website, but in the cases she cites all the new user is exposing is the new content they put in the website. Unless I specifically create new content for the new grid (hence upping concerns with the above even more) I’m also risking my income stream from the content I already have in place in SL – and even with new content I could end up seeing it release C/M/T around the SL grid or sold in direct competition – no way would the Lindons pursue DRM issues with that.

    So what’s to tempt me as a content provider across to this new RP OpenSim grid? If I go at startup I risk wasting time creating/moving content AND my existing income stream in SL, for which I might gain a groundfloor place on a new grid – but only if everyone else – content providers and players move with me. On the other hand I can wait to see if it does succeed and trustworthy, let others take the initial risk, and when it’s growing I can join and I’m still sure to be welcomed – except in that case it’s likely that most other designers won’t have taken the risk either, no players will have followed, and it’s fallen flat.

    So now the barrier to success starts to look alarmingly high. It’s like an stable equilibrium – you can have a successful RP community on a set of sims in SL, and you can imagine having one on OpenSim, both take about the same amount of ‘effort’ to maintain (if the recruitment issue can be solved), but the process of moving SL to OS? That’s a whole different ballgame – there’s a massive fence there that’s only apparent when you look closely.

    Myself I suspect that the only way that an OS RP grid can significantly succeed is if SL itself fails – either completely, or relatively if the technological advantages of OS over SL become so great that people prefer OS because of the greater realism. Myself I would follow the income stream.

    Now RP communities may be a special case, possibly the barriers to transfer are less in ‘generic’ SL. But I feel that many of the points the RP situation highlights about the practicalities of moving a society are just as true, just a little more hidden.

  • Hello Hiri …. FYI …. There is a quite successful tool for migration of content from SL to OPenSim called Second Inventory. Please forgive me if you know of it already. Just didn’t sound like you did.

    Its not perfect by any means, and can be somewhat frustrating for a big builder like myself, but I think clothing designers will not have any real problems. It doesn’t yet recognize nested objects, so you can’t just bring in stuff that is boxed up, but its a small price to pay to have your stuff backed up and able to then turn around, when you feel the time is right, and rez it in a trust OpenSim grid.

    Tessa

  • Score 98 out of a hundred, Gwyn 😀

    Excellent article!

    The other 2%?

    Goes against the (mis)use of the term ‘reverse engineering’, which in typical use, implies that the reverse-engineered software was taken apart and analyzed in it’s original form.

    While the client *may* have been reverse-engineered to obtain the basic protocols, the server software itself, as you say, has never been in the posession of opensim core, and consequently could not have been ‘reverse-engineered’.

    The opensim software is, in fact, designed and built from scratch from the protocols up 🙂

    That said, this post is a clarion call of the highest calibre!

    Cheers! and keep up the great work!

    Hiro Protagonist
    OSGrid Admin/Region Operator
    (OSGrid is the reference grid implementation for the Opensimulator Project)

  • Hiri Nurmi

    Thanks Tessa, yes I’ve seen Open Inventory, and if I was moving inventory to OpenSim that’s the way I’d go.

    But that’s not the concern. The issue is that if I did move my contents so they were available on ‘Spanking New FreeOpenCyberGorPirateCapture Grid’ then I’m trusting whoever is running it with my current income stream on the SL grid, and for what? – well with the current number of OpenSim users effectively nothing. I just doesn’t make sense to do it.

  • Thanks, Hiro 🙂 I was using one of the two popular uses of the term reverse-engineering, namely “clean room reverse engineering” where the actual source code is not known.

    Hiri, you’re right that the number of regular OpenSim users is quite minuscule compared to SL, of course. Then again, it’s about the same number of users SL had in early 2005, and there were quite a number of content creators getting an interesting revenue from it. The key aspects here is trusting the grid you’re connecting with; in my opinion, if Tessa’s project goes ahead, “complying” grids will actually give more protection to content creators than non-complying ones (ie. like LL’s own grid which doesn’t recognise any organisation protecting content creators’ designs).

  • Thanks for the mention Gwy … And FYI, we are going forward, no matter what. We’ve just spent about 15K on our server setups and our prospective investor is already granting us a co-location rack space for it with a wide open bandwidth for a ridiculously low figure per month.

    We are very fortunate to have this investor interested in us and willing to put his money where his mouth is before even the ink has been laid down on our contract with him. It is a testament to his belief in our business model, and probably more importantly, us as business people. This alone should allow us to remain very competitive and stable.

    We are waiting for a few more irritating building bugs to go away before we open, and we have a lot of building and organizing and programming to do, after we get our investment funding, and before we can open.

    If anyone wants to volunteer time or builds, please let me know. We’ll also be selecting stellar volunteers to help us with newbies after we get funded and open up. We’ve budgeted in funds to pay them something for their contribution. Won’t be much, but we feel its important to give back something for the effort.

    We’re shooting for March 1st and welcome any other grid owners to get in touch with us after the 15th of January, so we can share ideas and thoughts on grid standards for the HyperGrid protocols. I’ll be annoucning a meetings soon that to be conducted in Second Life to field questions and get your feedback and ideas. I’ll also have our Senior Grid Admin there to help answer any really techy questions about OpenSim’s core.

    We’re just hoping we can give the business and player class clients a viable place to more to. I know we won’t get it right 100% of the time and we won’t be able to meet everyone’s idea of utopia 100% of the time, due to all of us having to work with an ALPHA software. We’ll be contributing code to core as much as possible, once funded. We think that’s just what you should do if you use this software.

    We also realize there will be a lot of people coming over who feel they’ve been burned intentionally through their experiences on the other grid. Its natural that they will may come at us a bit hypercritical and skeptical. We understand. We’re going to try our best to be a more responsive management that listens and then acts to the best interest of the grid population. I personally just would like to ask for everyone to give us a little benefit of the doubt and remember we are NOT LL. In our minds, sure bills gotta get paid, but HOW you go about getting them paid is just as important. Having said that, if anyone becomes overly divisive in meetings and such they’ll be asked to leave.

    Please forgive me if this sounds like a commercial. Just don’t know how to reach out to the others who many be contemplating opening their own grid or looking for a reputable alternative to Second Life.

    Theresa Johnson aka Tessa Harrington in Virtual
    * everyone calls me Tessa in real too, so that’s kewl*

  • Pingback: I absolutely love Hiri Nurmi’s insight … « Must … Write … Something()

  • Hiri Nurmi

    Gwyn – yes I can see the OpenSim grids have around the same number of users as SL a while back and content providers were doing ok then, but at that point SL was the only game in town, so there was no additional risk. That’s not true with any new OS grid ventures.

    Let me try and quantify it. Say, for the sake of argument and round figures I’m making $200 a month from the SL Gorean population – estimated at around 20,000 avatars active to some degree. Now if ‘spankingnewOSGorGrid’ opens up and does stunningly well and attracts 5% off the main grid – widely optimistic in my opinion – then by being on ‘spankingnewOSGorGrid’ I can make myself an extra $10 a month. I suppose I might do better than that initially because it’s new and everyone’s buying clothes and not all designers are there, so we might get as high as $30 or $40 even.

    So I’m hardly going to develop new lines for ‘spankingnewOSGorGrid’ – I’d take stuff over from SL with Open Inventory. But every item I put up on ‘spankingnewOSGorGrid’ opens me up to the possibility that I might be ripped off and my stuff appears back on the SL grid. So I’m risking a $400 pm income stream on SL (factoring in Gor at 50%) for a marginal increase of 10% at the most.

    Hence my choice is I stay in SL only, risk nothing, and maybe see a marginal decline of 1 or 2% in my income because of the migration to ‘spankingnewOSGorGrid’, or I risk several thousand dollars in total for the possibility of a few hundred. And I can always change my mind later to add ‘spankingnewOSGorGrid’, but once I do it’s an irreversable exposure.

    So for a decision to sell on the new OS grid to be realistic I don’t just need to trust ‘spankingnewOSGorGrid Ltd’ the same as the Lindons, I need to trust them more, much, much more. And I need to know I’ll be able to carry on trusting them and their successors for the lifetime of my products.

    And as I said before Roleplayers like quality content. Wander around any RP sim and you’ll see the degree of care and attention that is lavished on RP avies. If you don’t have the content you won’t attract the players, but if you don’t have the players you won’t attract the content providers – and of course by definition almost the best content providers are the ones with an income in SL itself they need to protect so the most likely to be cautious about moving.

    Finally let me emphasise, it’s not that I’m against the arrival of alternative OS Grids – by all means let a thousand grids bloom – and if there is a viable model that addresses my concerns then I would support them – and my particular take is very RP sim orientated so may not apply elsewhere. It’s simply that I see an assumption made that just because the technology is rapidly approaching then an OS equivalent of the LL grid will be on it’s heels and I don’t think the social and business aspects of how that arises has been thought through given that the LL grid already exists and is dominant.

  • Pingback: Vig does OpenSim « Second Arts()

  • Hiri, point taken, of course 🙂 In fact, that’s the same argument why people design digital content for SL (potential customer base: 16 million consumers) or IMVU (20 million potential customers) and are leaving Renderosity (where only another 100-150,000 3D designers are potential customers).

    You’re just plainly stating pure and simple economics, the effort of attracting another 5% of the market with uncertain risk is too high for a regular designer to take that risk (since the use of CopyBot on the main grid is by far less than 5% of your risk).

    I couldn’t agree more with you. And since by 2010, the probability of OpenSim residents surpassing Second Life residents in size is effectively zero you’re quite safe in developing just for the main grid, at least until, oh, 2015 or so 🙂

    Still, I’m pretty sure that some content providers will follow Tessa’s endeavour to create a federation of grids that actually protect more user-generated content than LL currently does, and will not hesitate to launch their content there — for just one reason: there is high risk, yes, but also a higher margin for profit, as designers in SL are reluctant to move over to OpenSim, and for a while (let’s say 5 years) you can offer your products and services without fear of competition and for the price you wish — pretty much like low-quality products made a huge profit in 2004/5, since there were few choices available. When Namssor Daguerre invented the concept of avatar skins in 2005, he sold them for L$6000 each — and he sold thousands 🙂 (today, you get them in packs of reasonably high quality for a few hundreds of L$) The first person who devised a way to make animation overriders sold the script for L$400 without any animations inside. These claims sound absurd today, just because we’re used to competition to lower the prices for the consumer. OpenSim is still virgin, unexplored territory; but, of course, it requires first for people to trust it. Nobody made any money on the Closed Beta SL in 2002, when permissions were introduced and the L$ was a novelty 🙂 — and nothing really worked well yet.

  • Well, I have to kinda disagree with 2015 being the day OPenSim is a real contender. I think that’s truly right around the corner, not so much because they want it to be so, but because circumstances are pushing it that way.

    Why? Well, because LL seriously needs a makeover. You say LL or SL to a corporate client you get scoffed at. Not a good thing. And its not that they don’t fully appreciate the concept. They love it. What they don’t love is LL’s apparent inability to get out of its own way – See past their program-centric ego, and listen to at least some of their client base for how they can heal their mostly self-inflicted wounds.

    Or maybe they ARE listening.

    Ask yourself some of these questions. Why is LL not wringing their hands over the competition that OPenSim is or will eventually present to their business model? Why are they seemingly encouraging companies like my own to push past their own boarders and set up grids? Why isn’t LL working on Interpolarity – teleporting back and forth between SL and OpenSim? Most of the crowd that was working on that have been laid off.

    Now, take a gander at the LL economic statistics on the SL website. The “Islands OWNED/ADDED” section is disabled. Why? Because it has been showing a negative total in the latter for about two months. That alone should make LL worried, but they are not.

    I think LL has been struggling to find their way. What, between the meteoric rise of 2005-7, with all its ouchies, the media humiliation of the rain of penises, and then all this economic mess and media brow beating they are constantly taking, they are lost in some ways. They truly need to reinvent themselves – find the silver lining to having over 500 CPU’s sitting ideal in their co-locations, due to a mass exodus over the price hike.

    How does a company like this take advantage of a mass exodus? I’ve always contended that LL was pretty clueless as to how confrontational the 67% OpenSim price hike would be to their user base. Its pretty clear Jake Linden certainly had no idea the doo doo he was stepping into when he blurted out his speech.

    So maybe they are finally thinking outside the box and going to use those freed up resources to go in a new direction – possibly to become the ultimate hosting service for fortune 500 and military gov contracts. LL has recently hired Washington/White House insider, Scott Secher, as their Manager of Government Accounts, as well as cuddling up to some heavy hitter military veterans. And we’ve all read about how they are doing a major server side improvement on he blog by that FRANK guy. Some think he’s Jack Linden, the infamous giver of bad news about the OpenSpace sims. Word is that he’s heading up the initiative to run some serious dark fiber, (dedicated fiber optics) to all their co-locations around the world to give them the best internet bandwidth and highest level of security possible. Why? Something tells me their not doing this just of us, the grid users.

    If my guesses are right, a year from now you will see a vastly changed horizon for the 3D Web, Linden Labs and OpenSim. IMHO its all good – Not a bad thing at all, IF everyone works hard to accept and prepare for the change.

    How could things pan out? Well, LL would probably end up taking on the big bad dog-eat-dog Fortune 500 boys. Go for it, I say. I’ve worked with the big boys and I have to tell ya. Its much nicer working with smaller and more agile business, so you won’t hear me being angry at LL anymore over all this.

    And who knows. Maybe LL will even throw us a bone and offer an affordable product those to those who want to host their own OpenSim grids. Some people will probably laugh at that idea, considering LL’s loNG LONG LONG history of memory leaks,lost inventory items and just genera bugginess. Kinda hard to charge for a service you haven’t perfected, but I’m betting the dark fiber will help.

    Truth is, this probably has to happen for things to progress as they should. Maybe we should consider putting our own positive spin on all this and work towards moving our communities to populate and financially fund the progression of the 3D World Wide Web through virtual land rentals, the retailing of virtual goods and doing designs for the small to medium sized companies using OpenSim. Otherwise, we might just not reach our full potential. Lord knows I can’t sit around and wait for LL to grow a brain and go the way I want them to, so I’m just stepping out and turning the situation upside down and trying to find that sunny side.

  • I think Hiri brings up some good points.

    Under the current OpenSim architecture inherited from the Linden Labs Second Life viewer, it is difficult to visit other grids and worlds. One could argue that the Hypergrid is actually a work around for this – the connections between worlds are achieved at the server level rather than the client level. The disadvantage (or conversely, advantage), is that you can only visit other grids that your home grid has explicitly chosen to connect with. Though maybe one can then travel from that grid to a completely different grid – I’m not sure if the code currently allows this).

    A more long term alternative may be to visit other grids just by typing a different url into your viewer, in exactly the same way that a web browser currently works. Then the architecture of independent virtual worlds becomes much more like that of the existing web, with all the advantages (and disadvantages!) that this may bring. It would allow you to go and visit any virtual world you please and make it much easier to bring in new blood.

    For anybody interested there are a few more details at http://justincc.wordpress.com/2008/08/15/could-there-be-a-future-without-big-grids/ (which was written pre-hypergrid).

  • Dear Gwyneth:

    I want to compliment you on sorting out so many of the issues so well. Your writing makes me think about our obsession a bit with IM and where we are going with OpenSim development and deployment on grids such as OSGrid.

    Personally, my goal is to help us get to version 1.0 of OpenSim by the end of the year. Currently we are at about version 0.6 and are consequently about 60% to having the features an average user of the maingrid might expect.

    OSGrid has been running for a year and a half and is the oldest continuously running OpenSim grid. We support both the development with a close relationship with the developers and have been building a community. Both the OGP and HyperGrids first appeared on OSGrid, so we will continue to be in the forefront of OpenSim development and deployment and help all the newer and younger grids work out deployments in UGAIM and sims.

    Again, thanks for the article. Its a really good one.

    Charles Krinke
    OpenSim Core Developer
    Director of OSGrid

  • Hi Gwyneth,

    I appreciate the level of expertise and knowledge you bring to this HIGHLY interesting topic. I will be keeping any eye on your blog and OpenSim as I am keenly interested to see how things shape up.

    Thanks for all the great work,
    Avarie

  • Ranma Tardis

    Not sure if this will be the answer to Second Life. The technology used is based directly from Linden Labs creation. A problem I see is the content providers trying to secure any possible rights to things made in Open Sims. Then there are the freeloaders in second life who provide nothing to the experience. Sighs I do not see an easy solution and being international makes things so much more complex. Will the different laws bring second life to its knees? German law is more harsh than American law. I mention “age play” which is allowed under American law but forbidden by German and European law.
    What does this have to do with open sims? The problems encountered by Linden Labs will be shared. I am still wondering if this means we can create our own grids completely separated from each other? One might ask why make another. My answer is that it would be run at the whim of its operators. I suppose one could make a grid for the purposes of ageplay, gorean, furries, democratic, nobility, etc. Any grid located within the United States would be much more free than if located in the EU. The only thing the EU could would be to block such. They would be helpless to enforce European Law on an American system operating within the United States.
    I do believe that the people behind open sims will solve the problems with time. I have a lot of faith in the ingenuity of people and we are not all driven by the profit motive.
    Would like to know how I can help? Also I wonder what is going to replace the current and when? What is holding back the development of the next generation?

  • Raddick

    Nice article Gwen. I pop in on your blog from time to time, and you usually have something interesting to say.

    I have been investigating OpenSimulator for about a month… I have a grid of my own, running on hosted servers, and since I am an IT guy in RL naturally I have been digging into the code and am starting to contribute back to the project.

    For me, OpenSim is a very exciting project. What it does is astounding, considering the small team of people and the timeframe used to get where we are now. I dont think Linden Labs is quaking in its boots just yet though; as you have pointed out, they have a production system running on the largest public grid in the world. I suspect that they will start reaping the benefits they secretly and not so secretly expected when they put the client into open source. What I mean is they have a good team of unpaid developers in the wild, cranking out new functionality at no expense to themselves. They can afford to sit back and see what works and what doesnt in OpenSim, and cherry-pick the best features for future incorporation as they see fit.

    Yes, the Linden codebase is creaky and difficult to update. That is the penalty for being the first explorer to land on the beach… on the other hand, they have a lot of experience on that beach, and have a strong position and control that the OpenSim community will never have.

    I think you gloss over the IP rights issue far too much. I am for sure not going to hypergrid to some Bulgarian server, provide hooks to my content, and hope that the grids I go to are ‘honest’. With HyperGrid, if you can see them, they can see you.

    The LL money system is another advantage which is seriously underestimated by almost everyone I talk to. SL has done a great job of locking down their money system. I am very skeptical of the trustworthiness of open source solutions when it comes to my wallet. This is not to say there arent smart people working on the issue; rather that LL is running a closed matrix, and they control all of it.

    In any case, a year is a long time in the OpenSim development lifecycle. it will be interesting to revisit the topic, come January 2010.

  • Lance

    The smartest thing for LL to do would be to get out of the hosting business entirely.

    Providing two major service (currency/billing services and inventory server services) would address their strengths while allowing them to avoid where they keep falling down (grid server performance, customer service, etc.). The fact that LL is refusing to see is this: nobody in SL cares that they can walk/fly from one mainland sim to the next; new locations are found by Search and from bookmarks from friends, and traveled to by teleport. -There is no need for unrelated sims to be adjoining-, therefore no need for them to be on the same grid.

    It would also get LL out of the “content approvals” business, which is one which no sane company wants to touch.

  • Hmm so the idea is that people would pay for storing their inventory and the LindeX on LL’s servers?

    That requires some thinking before saying if it’s a good idea or not. My current first thought is: what would the impact of such a slimming down of LL be? They earn over US$5 million/month just from hosting sims; where would all that come from?

    Premium service? That would be a possibility: to be able to use LL’s central servers, you’d pay, say, US$60/month per avatar. There are around 80,000 Premium accounts right now (who pay a tenth of that) — would they be willing to pay that much to have their inventory centrally stored? If yes, well, I’d say that would be a possible solution.

    Then the next question would arise: how would the migration work? Where would the 30,000 grid regions be hosted from now on? How would LL handle it? An auction system where grid operators would bid to get a region to host? I can imagine that this would be very, very tricky to handle.

    Note that, unlike what you claim, the visual contiguity of Second Life is far more important than it seems, as I have explained on an earlier post 🙂