I and most of the people I interact with in Second Life, find the community (i.e. friend interaction), the intellectual stimulation and the creative outlet to be the compelling reasons for being there. The level of interaction is far greater than any 2d application, and I am SO TIRED of people saying that SL is another chat group. They are beyond clueless.

Face it people, SL is the best most compelling as well as entertaining PLACE, (and yes Pavig is spot on as describing place as critical), in cyberspace.

Everyone loves to want the ideal that satisfies their ultimate desire of what things should be, but I would ask you to take a step back and see what is here now: something that is a far more intersting interactive replacement for passive entertainment of the past. (i.e. Television)

OK, that said, here is a concrete suggestion of how to make premium membership grow. Double the number of groups available to the premium member and uncap IMs. I also suggest to add more profile picks, allow unlimited (or substantially greater) text in profile picks, about, 1st life etc. Everyone I know who uses SL regularly wants this stuff and will pay for it. Having a couple of hundred thousand premium members is NOT trivial economically.

OK, Good work and excellent analysis as usual Gwyn!

Everyone was making computers and no one was getting anywhere. IBM had this new idea, we will create a computer but instead of guessing what everyone wants in it we will just create slots that prople can make things to slot into and we will make the architecture public.

This they did, and the PC did take off but everyone soon cottoned onto the fact that the IBM “clones” could do exactly the same job as an IBM PC but were far cheaper. How many branded IBM PC’s do you see nowadays?

I think LL has done similar by making the source code for the viewer available.

I don’t know which geek category I fit into but being an IT manager I was immediately intrigued by the way SL worked and OS gives me the chance to look under the hood.

The service provider I use for my servers btw won’t sell to the US as the transit links are not that good.

Oh yes, OpenSim has a bright future… in about two or three years For now, what amazes me most is that it works at all. My team has just completed a major building project, still using OpenSim 0.5, and after a month of work, it was presented at a public conference. We never told anybody that we did it on OpenSim and not on LL’s grid (the client didn’t have a budget to pay for LL’s servers’ monthly fee, so this was the alternative) The fun is that the whole grid just required 5 or 6 reboots in a month or so — not too bad, for this early generation of software. And it allowed the builders to stay long stretches in-world on a poor, low-powered server (a mere 512 MB of RAM on a low-end dual-core, running 4 sims with about 10,000 prims among them). I wish I had asked you for the name of the provider that gets you four times the RAM for pretty much what we pay

Still, SL is by far not “doomed”. To be honest, and although it doesn’t look like it for most of the residents, SL has barely started. Granted, in my view of the future, the MetaGrid will encompass sims from both LL and a whole host of different providers. OGP might be seen as “dead” because LL moves ever so slowly — a patch of two or three lines of code that was submitted to the JIRA can take up to half a year to be implemented, due to staggeringly complex Quality Assurance testing. OpenSim doesn’t require that — patches are applied immediately, and if something breaks, a new patch will be out after a few hours. That allows for an insane pace of development, and it’s fair to say that OpenSim today resembles Second Life in 2003 (in terms of stability), but with half the development time. It has far more developers than LL had in the 1999-2003 period, though, and, if you add up the OpenSim residents on all the different grids, they’re about 4 or 5 times the number of residents of SL in late 2004 — and they all accept far worse performance, far more glitches, and less stability than we had in 2004. So, yes, the future for OpenSim looks bright, and OpenSim improving way faster than I actually believed to be possible!

What LL has — and will continue to have — is a momentum. If right now they fail to attract more residents (or, to be honest, fail to keep those 10-20,000 new daily subscriptions in-world), it’s mostly because they find SL too cumbersome to use, too flawed, too demanding on old hardware, too slow, and, for many, too boring. OpenSim is all that — and worse. So even if tomorrow 10-20,000 new users joined OpenSim daily, the ones remaining would be far less than the 1% that LL still manages to capture in spite of everything. OpenSim will thus grow much slower.

It is an old adage of the software industry that any software application requires about a decade of heavy use by hundreds of thousands of users (as opposed to a handful of beta-testers) until all bugs are ironed out and the application is robust and stable. That’s why these days Unix users don’t laugh at Microsoft as hard as they used to — Unix might be 40 years old (next year), but Windows is around for a bit more than two decades, and it’s becoming stable enough for regular use. Second Life’s first lines of code were written back in 1999 — so by the end of next year, it’ll become stable, too (well, it’s getting there!). While OpenSim still has some 8 years ahead to rough out all bugs. And it’s not a question of having more people working faster (which they have) — Microsoft also used to think that way in the late 1990s, and found out that it’s not the number of developers that count, but the complex mix of years of use by hundreds of thousands of people reporting bugs and having them fixed. Software gains maturity like a good wine.

However, I can imagine that by 2020 Linden Lab will have no choice but to switch over to OpenSim as well

Just take a practical example. If LL’s server software implemented OpenSim’s hypergrid, they wouldn’t need the insanely expensive fibre deployed between their two co-location facilities. They’d just run two asset servers on each co-location data centre, and use hypergrid to teleport between both. Simple. So, even in 2008, bad implementations are already hurting LL financially. At the end of the race, in a decade or so, the winner can only be OpenSim, and as soon as LL figures this out and actually uses OpenSim internally, the better it will be for their business as 3D content producers.

Alas, it’ll take a few years until they realise this Nothing at LL ever happens “quickly”.

I now use OSGrid and will am now moving to a grid of my own.

OGP seems to be pretty much dead in the water. From the OpenSim side of things a team at UCI has created an alternative called hypergrid which took around three weeks to do. We already are able to TP beteween each otheres grids and you have access to your inventory and assets whulst you are in a “foreign” grid. You can even pick up freebies and return them to your own grid.

Bandwidth and server space is getting cheaper all the time. I have two servers at the moment, one is a dual core with 2GB RAM and unlimited traffic, this only costs me around $40 a month.

I currently run 11 sims on it, granted only around 10000 prims in total but I have the flexibility to configure it the way I want etc. If it gets too overloaded I can move a couple to another server.

OpenSim is still quite young but is moving along at a rapid pace, nearly daily, another new feature is implemented. It is only a matter of time before OS reaches the same levels of funcionality and stability of SL.

I know I will never return to SL, everything that SL has will come in the end at a far lower price tag. In the meantime it is exciting to be part of a project that is moving with such momentum.

Robin

There are a couple of problems with your flocking to mainland idea. One of which is that people who have a bad experience remember it for a hell of a long time, and the openspace experience has been a bad one. People are doing something with their openspaces, whether they’re abandoning them or converting them, there has been a significant drop in the number of islands on the grid, according to economic stats -1977. As growth was generally over 1,000 a month during the boom this is a large reduction.

Two things will happen, people will either abandon now whilst there’s no resale or hang onto January when a new homestead will cost USD$375 which may or may not introduce a viable second hand market as there’s not one at the moment.

As for mainland, a new private product, that joins the mainland content but has no waterfront could be viable. This would need to be less expensive than a private island but more expensive than standard mainland. A third party managed mainland sim, giving owners estate tools on the understanding that they must always have public access. That would have potential.

We don’t have any reason to insist on a pure p2p solution. A hybrid solution where servers will supply registration, tracking and other services like money, groups will be good for me. Servers should also be used as “support users”; persistent users with large upload speed to enhance performance.

What we need is that most of the static content will be stored in the clients ahead of rendering. That all the computing related to scripting and physics will be in the client. That most of the communications and coordination between avatars will be p2p. Lag will be minimized, and the number of servers will not have to be proportional to the number of sims. (I would scrape the sim concept altogether.)

I don’t have the full specification of how to make SL p2p, and even I had it, this is hardly the right place. But since you mentioned storage, lets consider some guidelines for handling storage.

User created content does not mean that any user is able to create where ever he wants and the content will be always persistent. This is definitely not the situation in SL as it is now. If you are not a group member then almost always you are not permitted to create anything, and if you are permitted to build, usually your creation will persist for 10 minutes or so, and then returned to you.

There should be two tracks for content loading.

1. Just In Time – Similar to what happens now. p2p and client cashes could help here.

2. Ahead of Time – Each community or developer group owns a big content file that contains the terrain buildings and any other object or clothing that is part of their environment or game map. The group members create and publish updates. The group behaves like any game publisher that allows its customers to download its games from the web before using them.

Everyone can create huge spaces, residential, commercial or game maps. Model builders are free to use any technology and create very detailed and heavy models.
In this regime land is free (save the fee to be paid to the developers), and the company running the world should get its revenue from services and VAT on any inworld transaction.

We are pragmatists. We want a fast world, user created content, an economy that will make us rich fast , and most importantly an advanced game engine. On all else we should compromise.

Two comments, though. First, those do not represent the totality of all possible gaming environments — although, granted, definitely a large slice of the market (I’m old enough to remember that “3D games” and “action games” where just one tiny part of all the possible games being played on computers ).

Secondly, you claim that the “vast majority of online players” are “games fans”. Well, I don’t know about that — statistics tend to show that there is quite an overlap of both “gamers” and “socialites” (as well as articles such as this one), and that both have dozens of millions of users world-wide.

Granted, after 2003 Linden Lab gave up on SL as a “games developer platform”. Indeed, as you claim, five years ago, LL really thought that game fans and game designers “should have been the natural end users of SL”. But they weren’t. LL quickly gave up on them, and let them find other platforms to design games (specially, as said, 3D action games — other types work reasonably well in SL).

It’s also true that “content designers” create content for other content designers, or, well, at least for the ones willing to pay for content (who are most often users that happen to earn money through SL, and that mostly means being either in the content business, the land business, or the event hosting business). The problem here, however, is different. Most (and that means a bit over 99%…) of the registered users have no interest in spending money in SL to buy content. It’s a very small economy. I could certainly agree that if SL was more appealing for game design, it might have paid games working inside SL, and these would be another source of revenue in SL’s economy.

Granted, it won’t be thanks to p2p networking (that’s all very nice to say, but virtual worlds with user-generated content require persistence, which is something that p2p networking doesn’t provide), but possibly, as you mentioned, by pushing more of the work done by the servers into the clients. However, there are some things that definitely will always have to come from the servers: textures and prims.

WoW can get rid of all that since all content is installed/downloaded by the clients — including pre-generated and pre-rendered scenes. WoW servers do little else but tracking avatar data. You simply cannot use a similar model in SL. It doesn’t work that way.

Extie: oops, you’re sooooo right. I’m sorry!

Extie- She chairs Thinkers you know.