Massively

Interesting analysis! The one thing about it that makes me scratch my head a bit is why, in the cases that you outline, one would want to teleport from the SL grid to one of these disconnected other grids at all, rather than just logging onto the other grid in the first place, without going through SL at all. If all that's in common between the SL grid and the disconnected other grid is your AV name, what's the point?

Reply

I had wondered about what the Lab was planning to do to earn money with their efforts to open source. So far it seems they are expecting to be able to continue as they have been. I don't see how that can be a good idea.

As you mentioned old spare hardware will be able to run light simulators and a lot of subcultures really would be more comfortable without the oversight of the Lab. If it were an option of turning on the old pc beside my desk for whatever added cost in electricity versus paying a block of cash and upkeep for an SL sim and having more control over the former its not a hard choice.

Linden Lab is continuing to sound detached and unaware of what is actually going on under their noses. They had best wake up soon or they'll be left holding the gradually dwindling bag. Projects like Multiverse and Vastpark are banking on the idea of being a hub for worlds. LL might want to look into that idea too, but maybe thats just me. Perhaps I underestimate the stubborn loyalty of the SL residents.

Reply

I believe this is where everything is going.

One thing that is so appealing is you should be able to run OpenSim on any one of these http://us.shuttle.com/kpc/buy.html
You can get a well equipped sim for 229 - 299usd! Not to mention the power requirements for these shuttle PC's will be much lower then most old clunker PC's laying around.

One thing I would love to see is if a giant portion of people go out and buy extra hardware to host there own world, at 229usd it is not much of a drop in the bucket.

Reply

Very insightful article, and a great read.

Personally, the three main things that keep me tied to SL are my hobbyist content creation, my friends, and my work. And surprisingly, none of those things are necessarily dependent on LL's grid, except for the lack of an alternative.

A grid with no upload fees where I can collaborate and share with other artists would serve my first need as well or better than LL's grid. Plurk and Gtalk/Gmail already serve my second need better than SL. My company's corporate clients would find the idea of hosting or resting a sim at a lower cost to be quite attractive and sufficient for their needs, so my third need would also be served.

The main things I would miss would be the variety of clothing, hair, and other content available, and the expansive collection of useful bits and bobbles I've accumulated in my inventory over the years.

Then again, most of my inventory is junk, and anything I really needed I could either re-upload or rebuild on the new grid. It would be particularly nice if I could export my builds to my hard drive (libSL can do this) and then reimport them on the other grid (libSL can also do this).

And honestly, I could live without the clothes and the fancy hair. Plus, the commercial content creators might eventually expand to other grids, and offer sales in USD via something similar to SLExchange. (*fingers crossed*)

So, while an alternate grid wouldn't be such a good option for everyone, it's a rather exciting prospect for me and quite a few others.

LL's estimate of 99% staying only on the Linden grid is rather optimistic of them. I think they'll be find their role as a company shifting to "one of many sim hosting providers", and they'd have to compete to stay relevant with other companies adding new features to their OpenSim grids. The biggest advantages they'd keep would be the "extra" grid services -- search, events, LindeX, central asset service, and so on.

Those, and the fact that LL would be the "trusted solution" would be enough to keep the dough coming in, but it would definitely be a major transition for LL.

Reply

yay! SL is getting closer and closer to actually become "web 3.D" !!!

regarding content, since most content can be saved to disk (at least techinicly, not focusing on the contractual side), it will probably be an analogous to what image files are for the web now

Reply

quote: "The biggest advantages they'd keep would be the "extra" grid services -- search, events, LindeX, central asset service, and so on.
"
With the lag at many live performances, which are a big draw in LL, I would suspect someone will figure out a way to have large concerts off the grid that would allow for less lag, more people to attend and thus more lucrative for the performers.

Just my thought.

Ansel

Reply

quote: "The biggest advantages they'd keep would be the "extra" grid services -- search, events, LindeX, central asset service, and so on.
"
With the lag at many live performances, which are a big draw in LL, I would suspect someone will figure out a way to have large concerts off the grid that would allow for less lag, more people to attend and thus more lucrative for the performers.

Just my thought.

Ansel

Reply

I'm afraid I'm a bit more skeptical than you are, Tateru. Before my words are read wrongly out of context, I should mention first that I'm an eager enthusiast of the OpenSim software and the LL-promoted OpenGrid Protocol that allows quasi-seemless integration of these OpenSim-managed grids with LL's own grid. This is something we've all been patiently waiting for since at least late 2004, and certainly since 2005 when the commitment of LL towards this goal became a reality and not "wishful thinking".

Yes, OpenSim runs quite well on old hardware; and yes, people have managed to run around a hundred sims simultaneously on their own PCs at home. This is nice, but just half the story. You also need bandwidth to deal with the incoming avatars. You can count with at least 30 Kbps (on average) per avatar; on a home-based PC with an ADSL connection, you'll have at most 1 Mbps upstreaming bandwidth (if your provider is very, very nice), and that will allow you to get 30 avatars on your hundred sims — total. That's not much. And most small-scale corporations also have similar ADSL connections.

Granted, large university or corporate campuses will have lots of more bandwidth available — and so will people hosting rented servers on co-location facilities. I can very well imagine that you can get a 10 Mbps Ethernet port on such a server for as little as US$100/month (if you shop around, you can get prices like that on low-end hardware and third-rate hosting providers), and this will realistically get you 30-60 avatars on a sim (a 10 Mbps port can usually handle a constant throughput of 6-7 Mbps). Not too bad. But if you wish to get more sims on the same server... you need to have a 100 Mbps port at least, to handle, say, 300-600 avatars on 4 sims, or pretty much close to what LL manages. Suddenly the prices skyrocket when you demand that kind of throughput from hosting providers (since this slowly enters into the domain of huge websites with millions of hits per second, or streaming video service). I haven't found none offering it for less than US$600/month, and often much more — as they start to charge for monthly traffic on top of the available bandwidth as well.

Again, this allows you to get "cheaper" sims than Linden Lab provides, but then you have to figure out the costs of maintaining those servers running 24/7 for your users. For hobbyists and universities, this might be more than enough. For casual use — to test out some new device, script, or technology that you're developing, before deployment on LL's own grid — it should be enough too. Jacek correctly pointed out that libSL allows you to copy content to and from OpenSim: so I can imagine many builders and metaverse development companies starting to do the development internally on OpenSim-run servers — thus saving the cost of hiring a sim from LL while it's in development.

All that seems quite clearly to be what is going to happen in the next few weeks, months, and a year or two — and I'm naturally assuming that a lot more in terms of integration will happen (eg. a new permission system on objects allowing them to be "teleported" as well between grids), or so it seems by looking at the awesome ideas posted by the Architecture Working Group.

The question is if people will ultimately be mostly on their home-run sims, or on LL's grid. Here I have my strong doubts. In spite of many comments to the contrary, the grid is inhabited by a vast majority of residents that want to socially interact — although, granted, the few dozens of thousands that create and build things are quite often solitary creators and don't miss much the crowds around them. But these days, the "solitary creator" is quite in the minority. This naturally means that most residents will gravitate towards the place where you have the biggest crowds, the largest and most exciting events, and the more intriguing architecture to explore. For many, many years this will naturally be Linden Lab's own grid (and thus I can very well believe their "optimistic" claim of keeping around 99% of the residents, even with microgrids proliferating all over the place). OpenSim-based grids currently attract a lot of new user registrations — it has been rumoured there are already at least 15 to 20 thousand users on OpenSim-based grids, or pretty much the number of SL residents in mid-2004 — but they have even higher churn (lower retention rate). I speak for myself and the ones that have "visited" OpenSim-based grids: they're (currently) as boring and uninteresting as, well, Lively (whose largest "room" has merely attracted 52,000 non-unique visitors in a month, or about the same number of page requests that my humble blog gets in a week...).

So what I think that will happen is that the vast majority of OpenSim-based residents, at least in the next year or two, will be university and corporate campuses, where "glorious content" is secondary, but conferences, teaching, training, workshops, and other similar activities will benefit from a closed use of SL, behind firewalls, in a Puritan no-nudity universe, and without griefers. The casual resident will never hear about those places. I can also imagine that huge, private initiatives like MTV's "Virtual Laguna Beach" or CSI:NY might attract their own private crowd in a sanitised environment — who might welcome the total control exercised by the owners of those environments and the pre-developed content by professionals — but, by and large, these will be mostly exceptions. Linden Lab is also "coming back to the mainland" (as reported elsewhere on Massively) and exercising much more control, so there will be a hard choice for prospective multisim projects — either run it on their own, benefiting from a lack of Linden Lab interference and sloppy prioritising, but with high running costs; or just pay the maintenance costs to Linden Lab and "trust" them to start to do much more aggressive enforcing on a more stable grid (which will be the task of LL's new Senior VP for Global Technology, Frank Ambrose).

Is this ultimately bad? Not at all. I still think that at some point Linden Lab will drop their own server efforts and switch over to OpenSim when it's stable and feature-rich enough to provide a solid infrastructure, neatly replicating all that we're used to have in SL itself. It thus makes all the sense to promote its development as strongly as possible. After all, its developer crowd are pulling code from all sort of open-source projects to manage a contiguous grid: using DNS for the asset server, OpenLDAP (or Windows Live ID!) for authentication, and IRC for intra-grid group chat. All these are clever efforts to duplicate LL's code on the central servers by using popular tools that are readily available, and in some cases (like DNS and IRC) have decades of stable code at its core. I'm quite sure that Linden Lab is looking very closely at what the OpenSim-based grids are managing to come up with, and learn from their experience in using readily-available open source tools to enhance the user experience at a very low cost of development.

OpenSim will thus become very quickly the forefront of research & development, where innovative uses of SL will be explored free of the restrictions of LL's slowness.

Another use, of course, will be for escaping the confinement of the Puritan US laws. Thus, as soon as intergrid currency is available (although I can also easily believe that people might simply use real world currency at some stage), anything that the California State forbids LL to run on their grid might be available on the OpenSim-based grids: starting with casinos and ending with escort services. For Linden Lab, this might mean that they'll be able to enforce a PG mainland, and get the 20% or so of residents merely interested in gambling and sex pushed into a distributed, owned-by-nobody, OpenSim-based grid. As soon as the teleporting is not only seemless, but also the content can be transferred between grids, this option might have a huge appeal for a substantial part of the SL residents — while keeping Linden Lab "virginally pure" on their own grid, but smugly having LL employees teleporting over to the "dark side of the grid" to have fun with the rest of the residents, too. If that becomes true, I might concede that at least 20% of all residents will always remain on the OpenSim-based grids — and not only the 1% estimated by LL.

More than that? I find it quite hard to believe, and mostly for one reason: Linden Lab has five years of experience in managing a huge grid, and they still have problems doing so. Everyone starting right now will have to experience the drama of complex networking and infrastructure maintenance: it's easy with a handful of sims; it's manageable for a few hundred sims; it becomes an utter nightmare when you have a few thousands, and requires expertise beyond what the common self-taught system administrator knows. And those that have the required experience are hard to find, since they already work for huge infrastructure providers — or even for Linden Lab.

Reply