More open-source metaverses...?

Jarod, I accept your reply to my point about your assertion being ineffective. Not because efficiency in argument in any way overrides logic, but simply because I'm going to assume that you've made a thorough analysis of OpenCroquet already and that it proves your point. That would be fair enough.

I haven't examined the architecture of Croquet as the basis for a distributed universe of virtual worlds, but I will. I certainly am not put off by Smalltalk, quite the opposite, I've loved it since the start of the 80's and consider that a lot of subsequent efforts took OO downhill. Anyway, languages are irrelevant here, they can always be reimplemented or accelerated as necessary, or indeed replaced.

What *is* important for our purposes here is architecture, and for SL I've done the observation and analysis that lets me know that its current architecture has no massively upscaled future whatsoever, let alone a fully open and distributed one. That is why I do not support (nor do I reject) your call for immediate delivery of the interfacing details of current SL --- I simply think that it bears no major relevance for where both we and LL claim to want to go.

At best, I think those details would show good intentions. At worst, they would allow other sites that work in the same way as SL and are just as non-scalable to be set up and hence compound the current problem. I have no doubt that entrepreneurs would love it, but it wouldn't get us engineers any further.

Ten thousand sites all running LL-compatible code would be 10,000 separate worlds minimally interlinked with teleports with no possibility of resource migration, and with the vast majority instantly dying from overload within seconds of people hearing of them. It would serve no purpose whatsoever in respect of our intended goals, and would instead distract from our work on them and drain our resources.

I welcome expressions of intent of any kind, included release of specs for precursor systems like SL, but they alone are not enough, and may even be damaging if they become the focal point.

What's needed is thought, open design, implementation and testing of the building blocks of a real distributed metaverse. Without that, we'd be relying on solutions to appear magically from nowhere, or from proprietary teams, or from open teams each working on their own thing. It's not enough.

I think there are two separate issues being mixed up in this discussion, so let me separate out "the other one" quickly.

I am a very strong supporter of open-sourcing SL, either wholly or piece by piece so that the community can work on those parts that are currently poor or inefficient or buggy and improve them. LL could then feed the improvements back into its commercial product, and we would soon have a vastly superior SL to enjoy, instead of one that is massively limited by LL's development budgets and manpower.

This would probably also result in a few competing SLs appearing, but that's unlikely to hold LL back since it currently has a reasonable lead in this field, and it might even focus their minds on staying ahead.

That's one aspect of this. It would be good, I'm all for it, yes please. But it would get us nowhere in terms of creating the infrastructure for an open universe of distributed but interacting scalable virtual worlds. And that was my point above. The two issues are, technically, entirely separate. Open-sourcing SL would be terrific for everyone, but it would merely create a far better SL as it currently stands, and that's not the end game.

A FOSS system will likely (though not necessarily) be based on a streaming version of Crystal Space.

Almost by definition accurate mapping of the real world into the virtual is required to be successful. Croquet is a million miles away from that, and it surprises me that it is even being mentioned in this conversation .. unless they are adapting Croquet to use CS.

The real core of any MetaVerse system will be the physics and rendering engine. Crystal Space, in many ways, is doing a very good job at that .. unfortunately they do not have the streaming aspect done very well. I have had conversations with the project owner in these regards a few months ago and it was nowhere near his radar.

I'm also unclear as to the separation of functionality, client versus server, so very problematic security concerns may be present in CS.

For this reason also the project owner of OSMP doesn't stand much of a chance as he's not very interested in using CS. Mostly because the lack of streaming features but also because he's not a VC++ expert.

I had a lot of friends contributing to VOS (based on CS) and it looked very very encouraging. Unfortunately I don't see any project progress there and my friends are all doing other things right now. But, who knows. They stop listening to me because I'm always saying "SecondLife" this "SecondLife" that.

My guess someone else very bright will come out of the blue with a hacked version of CS that streams properly and has a similar arch as VOS does but a bit more practical in terms of real world implementation.

Another possibility is that a Quake or an Unreal or whatever engineer goes down this path and gets their codebase released to open source or rewrites it all from memory. This is another possibility that seems somewhat probable, or at least far more probable than croquet going anywhere.

In the end though it is all about the physics and rendering engine and whether or not it has dynamic update capabilities. All the rest, the scripting language and the client UI is useful but in many ways fairly easy to engineer if you get some really bright guys in a room. wxWidgets for the UI and Lex/Yacc/etc for the compiler. Basic 101 engineering.

The problem of content protection may (or, in fact, may not) be an issue. I'm still undecided on this. I originally thought P2P was the way to go but the content protection in SL made me rethink it a bit.

However, all the trouble Philip and Cory (or whoever) are going to may be pointless if everyone is open sourcing their LSL code, if they end up writing back end code that connects via XML-RPC (like I have and it doesn't *need* their copyright protection) and if people just steal all the prim configuration via the stream and or the video drivers.

In which case the competitive advantages to LL and SL over open source or a P2P solution, content protection, may be a huge and rather unfortunate corporate red herring.

But hey, the status quo seems like its working so no point in thinking out of the box. Well, no point until someone comes along and starts eating our lunch. Though of course by then it may be too late.

And like some poster said above don't underestimate the network effect.

The again I guess TSO / ActiveWorlds / There all had that as well and look where they are now.

Unfortunately, as a consumer, I don't care very much about the potential of a product. Though Croquet certainly shows a lot of promise, your suggestion that I should embrace it because it may become a great product is irritating. Why should I be required to put my faith in a handful of programmers, who only work on the project when they've got spare? Why should I believe that they're suddenly going to bring everything together in an orgasmic epiphany of metaversal goodness? It is just as likely that the project is going to stagnate and die, like the majority of open source projects.

Though you may believe that a product is better just by virtue of being open source, this is not an attitude which I share. I refuse to elevate a piece of software simply because it may some day be good. That's a leap of faith that's too great for me to make. Even projects like GIMP, one of the most vaunted OS apps, are lacking when compared to their closed-source counterparts.

Actually, I think everyone agrees that an open source solution is better. The idea of SL becoming a popular close-source monopoly the likes of Microsoft Windows seems intuitively improbable to me.

An easier discussion to have would be how will an open source solution surmount challenges such as specialized hosting, content protection, and an industrial grade renderer / physics engine.

I don't think physics and rendering are the most important avenues to growth of a 3D virtual space. I think it's connectivity and extensibility. The web was expanding fine before Flash. What mattered was connections: to data, existing systems and to other nodes on the web. And the ability for anyone to create their own space. I believe that if SL just added much greater connectivity and extensibility tomorrow, it would explode.

I think certain systems are more suitable for open source development than others. When you have a large number of users who are capable and motivated to develop the system for themselves, it works well. Like Linux, since every programmer needs an operating system, programmers typically learn how to develop OS's in school, and programmers want to avoid being locked into Microsoft on the PC. I'm not so sure about networked 3D virtual environments. Most people get by fine without one and the skills to make one are rather specialized. SL seems more similar to things like Maya, Photoshop and Quake; software that companies put a lot of expensive expertise into. But that's just my impression.

A definition of virtual reality: an artificial environment created with computer hardware and software and presented to the user in such a way that it appears and feels like a real environment.

So, what you're saying, is we don't want a virtual reality but rather 3D WebPages?

I'm going to have to disagree with this. I think brigadoon (to me, very important) would not thrive in an environment like that. In order to have any real community, we need immersive interactivity which means we need a virtual reality.

I guess it's possible that 3D web pages are the way to go, but I have to admit that leaves me particularly uninspired.

I'm not sure exactly what you mean by 3D web pages. If someone wanted to call parcels in SL 3D web pages, I don't think I'd quibble with them.

My point was that SL's current level of virtual "reality" is sufficient for its continued growth. I don't see the limits of its rendering and physics as serious obstacles to its expansion. I think the main obstacles are connectivity and extensibility. I see an analogy between AOL and the world wide web. But no, I didn't mean to propose "3D web pages".

Blaze, the part I don't understand is why you are so focussed on the graphics engine. SL's graphics engine works fairly reasonably, and while not lightning fast, it's not the slowest thing in the world --- I know some static MMOGs that have far slower engines. It may not have some of the latest eye candy, but our problem here isn't lack of wow.

The main problem with SL's engine is that it doesn't factor out almost anything to the vertex and pixel shaders on our GPUs, so it's messing around with low level detail that ought to be handled by our modern graphics hardware. And because the CPU is tied up in rendering, the whole client becomes sluggish and scales poorly as environments become more complex. That's the only major problem with it as far as the future of the existing SL model is concerned, and this failing really ought to be dealt with by LL as a matter of the greatest priority over everything other than bug fixing. You can't build skyscrapers on sand, it leads to tears.

But that is entirely unrelated to the issue we are discussing here, which is to get ourselves on track for building an open universe of distributed but interacting scalable virtual worlds. Using CS doesn't get us there because graphics isn't the key problem. Just open-sourcing SL doesn't get us there, because it's built on a static resource architecture as I've described in great detail in earlier threads.

I haven't yet done any significant analysis of Croquet, but I have noticed that dynamic and distributed resourcing across an unlimited number of worlds and actors is part of their mantra. That doesn't mean that it works, but at least they recognize what the real problem is.

Let's be looking at the real problem. And if SL is open-sourced on the side, all the better, not only for us but even more so for LL as an object-hosting company that will be running a rapidly-improving product.

I don't think I said SL needs a better physics engine / rendering engine, I said that it's a hurdle for an open source solution to jump if it wishes to compete. ie: it needs an industrial grade rendering / physics engine.

Crystal Space has this, but it's not dynamic. At least not the last time I looked.

Croquet and OSMP do not have these, so I don't see them as viable competitors.

And that's where I disagree.

The key problem is immersive interactivity. I want to feel like I'm in an alternate-dimension, not just in the middle of a CAD presentation.

If it's just another knowledge space, then I don't really see any problems with the current internet.

I think a networked virtual environment is just another space that could be used for a lot of different things (like the web and telephone system and postal system). SL already demonstrates that. People socialize, play games, have discussions, have sex, make art... But there's a reason SLExchange is on the web and not inside SL itself, even though connecting to SL from outside is also a major obstacle.

And I agree with that, but lack of graphically immersive content isn't one of SL's problems except in the sense that people aren't creating much of it. Remember Tiger Crossing's exhibit at the last Burning Man? Just one small enclosure, but you couldn't fault its immersive properties.

Of course, you didn't only say "immersive", you also said "interactive", and in that area SL's feature set ranges from pretty poor to terribly poor, depending on what you want to achieve. Well, I won't say more about that here, except to point out that I've been trying to promote a more powerful structure for interactivity for a long time --- see the first link on my sig. Alas it seems impossible to attract any interest from LL in that direction, so interactivity is one good thing that might improve from opening the SL source.

And finally, the foundation that underpins both graphic immersion and interactivity is SPEED. Well we're never going to have that, unless LL factors out low level graphics handling to the hardware shaders on our GPUs. Why they haven't focussed on that already is beyond me, such a waste of power.

So I agree with you on part of the problem, the need for immersion and FAR greater interactivity/speed. But that still gets us nowhere in respect of creating the open metaverse. That requires looking at connectivity, protocols, resource migration, distribution, security, and above all, scalability.

I 99.9% agree with this statement. However, it is a problem for an open source solution which wishes to compete with SL.

If you can, I'd be very interested in how you think SL could go open source and still stay a viable business:

Post here, or preferably in this thread: /120/d2/42009/1.html

Let me clarify up front, Open Croquet is an example of what I see the next generation of virtual worlds being like. OSMP is another. It's not so much their current level of technology makes me argue for them, it's their organization. It's the fact that when technology was ready, Unix grew out of a bunch of disparate hackers, and into one of the leading operating systems today. It's the fact that Internet Explorer was the leading browser, until Microsoft got lazy; then because Mozilla had so many active developers who could keep it safe from viruses, on multiple platforms, its numbers started growing and applications like Firefox and Komodo began to build atop it.

Right now, Second Life kicks Open Croquet and OSMP in the butt. However, as those develop and gain a grassroots followings, ala Unix and Mozilla, they're going to start having features that Second Life doesn't have. They're going to be installed and used in places where Second Life isn't or can't go. I'm not arguing for what Croquet is, I'm arguing for what it's flexibility will allow it to be, and what Second Life's current, centralist design cannot.

It bears one relevance, it proves that Linden Lab is interested in what we, the user-developers, have to say about what they're doing. Showing us the communications schema wouldn't change anything, but it would open important conversations.

I've read a few articles on cryptography, and the one thing that stands out is the idea that the strongest crypto streams are the open ones that gets beaten on by the crypto community. Secret cyphers that never get tested are the ones that get broken, open ones that people get to test and experiment on are the ones that last.

That's why I want Linden Lab to show us something as useless and cryptic as their communication schema. It allows us a peak into the way SL works that doesn't require them to change a thing. It lets those interested ask questions to better understand why somethings are as they are. It might even help the Linden-developers by having the users ask questions of which they never thought.

I'm not saying you should switch to Croquet. I'm not talking to you as a consumer. I'm talking to Linden Lab as the developer. My point is to say that you, the consumer, likes Second Life best now but that is likely because SL is the only, viable product of its kind.

Let me ask you, Ardith, if two years from now you had a choice of playing Second Life as it currently is or some other system that was more specific to an interest of yours, had fewer people to deal with (re: neighbors who put up spinning cubes so they could run you off), and was much more flexible; which would you use?

There are several ways open source can be profitable. The question isn't, "how can SL be profitable as open source", but rather, "how is open source profitable for people that are putting out the open source product".

I'm somewhat close to this since we (two developers I work with on large flash programming projects any myself) are considering a suggestion to publish the framework we use to build large Flash UIs and the adapters we wrote to connect those UIs to any technology via remoting as open source. Based on some recent proofs, we're beating customers off with a bat (yay new office in a month), but....we know that our little fiefdom can't be long lived.

In a nutshell, and not a complete list at all, profitability appears to come from...

- You still own a company that implements the technology. You are (or should be) perceived as a leader in the use of it. Organizations will no doubt come to you for projects and consulting.
- Partnerships, which again extend the reach of your ability to implement.
- Co-authoring gigs, technical editing, paid workshops, development of training, books, etc. A book by Andrew and Cori on the technical challenges of creating a virtual 3d environment, and another by maybe Robin or somebody on the challenges of the social aspects, would no doubt be an interesting read. Cori himself (I think) has been a guest speaker on the MS campus (DRM I think).
- "Managed" versions of the product (paid for versions of the product that are directly supported by the developers of that particular version...Redhat, for example, was considered a "managed" version of Linux. PLEASE refrain from commenting on your opinion of Redhat).
- Investment from VC and so forth in any and all of these efforts.
- Ongoing hosting of the SL environment as it exists for a fee. YOu may get the code, but I'm willing to bet you don't have the resources to host 20k users.
- Other services, i.e....
-- extending UIs, mods, etc. etc.
-- Real private hosting services.
-- Development services (I have often said that 3d environment authoring/scripting will one day be similar to web page development as a profession). It would not surprise me at all if LL one day started hiring people as subcontractors to develop content for privately hosted efforts. I'm sure they could bill liberally for this service; I mean, who else out there is developing a hardcore bunch of these sorts of developers and making money doing it?

Hardly complete at all, but you get the idea.

I originally was dubious of the open source model for profitability myself...just seemed way, way more starightforward to take our IP and keep making a fair chunk of a living by licensing the compiled libraries without distributing the source (which can, btw, be done in Flash). But when you come right down to it, sooner or later, somebody is going to do something on par with what you are doing (or at least appears to hold that promise) and you will have lost the opportunity to be perceived as a leader interested in the advancement of your genre of computing. Even companies like Microsoft know this (I mean it's obvious now that getting the resources to build even something like an OS are commonplace; the only real barrier is just getting off your duff to do it).

Sure it'd be a shift for LL, but nothing that would break the bank. And as both Andrew and Phil point out, it's almost something they have to do or they'll get blown aside by the first decent open source effort in this sort of collaborative environment, and from what I can see, that can't be more than a couple of years away.

I replied in that thread as requested.

Full featured economy?

IP protection?

Full-fledged physics engine?

25,000 users and counting?

2.5 million in revenues in 2004?

Huh.

LF

Redundant for the development of future, virtual worlds. If we had web-based integration in SL like Croquet has now, we could use Paypal to buy things.

I wear a Kim Possible shirt, we used to have a Green Lantern Corps, pictures are constantly snagged from the web and resold as textures, and I lost count of how many "they got this free from me and resold it" threads are on the forums. Linden Lab doesn't protect anyone's IP rights.

SL wins here, hands down. I don't know if Croquet has earned a dime. I'd counter that by asking how much money Apache earned in 2004, and follow up by asking how many billions of dollars were made via web transactions.

If development of the Internet had been contingent on business people signing off on revenue stream viability and IP lawyers being happy with protection of whatever it is they deal in ....

Yay for XML-structured UIs (jeez just look at EQ) and...Kim Possible!

(answering below...should have put a link...)

http://www.eqinterface.com/

Yep it's all nice, they publish schemas and everything. A little annoying when they repub a window schema that you mod'd heavily, but meh that's xml for ya. In fact I used it in a workshop once to show it could be done with a base of even hundreds of thousands of clients.

Not to mention that the devs just worry about getting the XMLUI subsystem right and leave the actual mods to the community.

I haven't logged in my mage in a long, long time. Insofar as EQ was concerned that toon most definitely was no joke, but my urge to write game and other UI type code finally swept away my desire to interact with other people's...well, with a few exceptions.

It's been years since I got my druid to 65 and woke up and stopped torturing myself.

How is EQ? Fully XML'ified UI now?

People use paypal for "big-ticket" items (when compared to virtual things)... stuff usually costing ten dollars or more.

The linden dollar makes the concept of "micropayments" online more attractive. Why would I use Paypal to pay for a virtual piece of property that costs maybe 50 cents?

Besides...

If you make it Paypal-oriented, then you have to deal with the 100 or so "real world" currencies out there, which means a confusing array of exchange rates and even more confusion over how much that virtual house over there really goes for.

LF

As handy as such swallowing of external apps into in-world objects may seem, it's actually a comparatively trivial and weak feature. The Croquet user's universe ends at that box, a dumb opaque window into "the old world", certainly useful but a barrier compared to the open freeways of Croquet's bidirectional portals.

Or at least that's how it seems from a quick read of the Croquet docs.

Hmm, while the technological aspects are certainly fascinating, I like to analyse things related to sound business models.

Let's take two open source products which are an overwhelming success: Blender and MySQL. Both have been developed by small companies. Both are open source, but mantained by a core team which still works for the original company. Both are also free. In either case, the company releasing the source code has not gone bankrupt. In either case, the company behind those products has a sound business plan which allows them to use their own tools for their own purposes, get income from services developed using those tools, and still keep a developer team which mantains and supports the "core" of those applications.

Other open source software solutions have grown bottom-up. They started with a community of developers, and sooner or later, a "central organisation" which has motivated the project to go ahead. Sometimes this "central organisation" became a foundation, or at least a consortium. Things like W3G spring to mind. Yes, the source code is still open, but there is an "institutionalised" way of dealing with it. "Someone" is in control, to make sure the great ideas around that particular piece of software do not disappear. Apache and Mozilla come to mind.

Now, Linden Lab's own finantial model is not based on software development. If it were, they would charge us a license fee for the SL client, which would reflect it's development cost - up to 4 or 5 years of 3 developers working full-time. Sum that up, divide it by 25,000, and that's the price that each of us should pay for downloading the SL client!

But Linden Lab (unlike MMOGs and MMORPGs) makes money by being a 3D virtual world content hoster. For a fee, you're able to host content on their 700+ servers. The SL client (or the simulator software) are just "side-effects" that they need to provide in order for us to create content to be hosted on their server farm.

LL is not in the software business, or in the content business. So they won't earn money for doing "specialized applications" (ie. software development for specific customers). And, really, almost all content in SL was designed by the residents, not LL. That's why they have just a tiny designer team. Their income comes only from hosting content, not providing it, or providing the tools for designing content.

Now, if you open source the server software, this means that you can threaten their business model, if you are able to compete on the hosting business. So, for LL to survive finantially, and sticking to their core business - hosting content - they have to incorporate "an open source strategy" which doesn't commit their business plan.

One possible solution to this dilemma is described in this article (ok, shameless plug, but I really don't want to copy & paste everything here...). After all, a very similar problem happened with the Internet some time ago, and the solutions devised to deal with those problems - finantially speaking - were even copied from what the telecoms have been doing for a century or so. There is no need to reinvent the wheel here.

As to OpenCroquet and OSMP, they are academic technologies. Like so many others in the world, the major problem with them is how to make them work finantially. Fortunately, we have millions of open-source solutions that have worked well finantially as well, so, again, it's just a question of finding the correct model. I think that currently neither of those platforms have a sound business model behind it. Virtual Universe has a sound business model and a company behind it. You'll see that you can join for free, even host servers for free (like ActiveWorlds), but if you want to set up your own "grid", you pay a lot. And the company doing the software has a core business which has nothing to do with "hosting virtual worlds" or "software development". Anyway, don't bother to download VU It's pretty disappointing, although the written documentation on the technology sounds encouraging.

And BTW and OT, SL will support web pages in-world much sooner than any of you would dream to be possible. I can't reveal more without breaking the Linden's trust. Let them do an official announcement in the near future