Should Linden Lab Open Source Second Life?

And along comes Philip.

At the birthday bash today there were a lot of troubles and it pretty much devolved into an open discussion about whatever came up.. Philip mentioned that evendually Linden Lab would no longer be able to be the only developer working on Second Life..

He also said Linden Labs is moving towards open standards and eventually open source. There will be a time when Linden Lab is not the single server host for Second Life and the goal is MILLIONS of servers. Open source and open standards are the only way ;0..

Yeah.. So the questions are when? and What are the issues we will face?

I tried installing open source on my computer once and it ruined my registry and I had to reinstall! Also isn't open source illegal? I heard you could get fined for running Linus on your computer. This doesn't sound like a good idea at all, I don't want people spying on me through my webcam.

Very funny ;0

...particularly funny from someone who advertises a "God Mode" for sale

Actually Baba stuff cant actually go client side that would ruin the purpose of SL as a whole and i doubt LL is willing to take that dent to the content providers. How SL works requires that there isnt a client side thing and everyone would be having to add a patch every time they purchased something or something was placed in world. Yeah like i really wanna do that. Seriously though there are alot of problems open sourcing SL.

For starters its a call made by the people using the Linux client i knew when LL announced they'd be working on a linux client that this question would arise since alot of them are free software junkies. Development of SL in the right direction requires LL to keep it the way it is. Open sourcing it just makes stuff weird. I can see adding more open source projects possibly like the Ubrowser thing, mabye some open sourcing of the UI etc but aside from that ugh.

If you dont like how SL is now and you want an open source game like it you can pull the plug on SL and go get a degree and start making a game yourself. They have people to pay for development liscenses, investors, etc. The idea of them going Open source is pretty out there simply because of the investors part. They are going to want to see a profit or a return on their investment and if LL makes things open source well their investors go bye bye and they become some open source company that really cant do to much. There are some client things that could be opened up but they need to be limited to what they can access.

I dont think it really neccessary to open source SL if you dont wanna pay for it dont pay for it, if you want to pay for it pay for it but stop whining about people providing a service for you that you chose to play. LL may be able to open some things up client side but i doubt they will give out server related information for people to connect with it and in the next year they expect competition so ultimately open sourcing wont happen for years if it ever does.

The way things are right now, Linden Lab is one of the few players in a niche market. Second Life can only grow as fast as Linden Lab does, and while they are growing pretty quickly, the market for Second Life in it's current state is small.

The call for open source is more about open standards than it is about free software. It's about expanding the scope of this project into a future platform for web services. Second Life is not meant to be a game, but it's it's built like a game; as a closed system. Second Life needs to become something more like Apache, and less like World of Warcraft.

What complaint could investors have when Linden Lab is the largest web host and main software developer of the new web?

I know that's Philip Rosedale's dream.

What possible benefits could become of this that are worth the likely bad things that will occur?

Ever heard of Apache? Runs 60% of the web.. ;0

Anyway, open source is meaningless without open standards as well ;0 So Second Life needs to develop or adopt open standards for Second Life as well.

If standards were open, anyone could develop a server or client that works with any other server or client, open source or not. Open standards is another topic I'm keen on ;0 ...

I want open source because I think it could fast-track the move to an open standards model in Second Life. A quicker way to SL as 3D web.

Sure, it'll change the whole sales/value dynamic, but the protocol has already been reverse engineered in large part. Might as well officially open source it and get the advantages that go with it. Like low resource clients, wider hardware support, better ui, ai, etc.

With the move to REST architecture for asset requests and such, we could move away from a single central asset server. An object could have a uri that could send a request for an asset at a specific location.

asset://storage12.yahoo.com/request?id=sp9uwout0esoaz4uw
riaspoe5r2u93o5driejleprlufr82rlet8a1lacrou8lec&loc=secondlife://
joe.slhosting.com/Utopex/24/44/38/9le7oecriavlat9a2oadr4agluflucl4

My views are outlined in the second half of this blog post and (mainly) in its comments. (Kei Mars = Miss K, btw)

Any move to open source, has to be followed by a commitment to open standards..

An example of an open standard is HTML and XHTML. Every browser should be able to render the same image from the given HTML. This allows for competition between browsers. We have Mozilla, Opera, IE and many others.

Open source would not be neccesary to implement an open standard, but as I stated above, I think it would speed development into that area.




Read the comments... Miss K put it well:

I think part of open sourcing something like Second Life (which is by no means a simple task) is to define some Open metaverse protocols and standards (much like the W3C continue to do for the web), which means that grids and clients on different branches have some ability to talk with each other and interoperate It might be that Linden Lab act as the guardians of this, or form the consortium to do this (a little bit like how Netscape seeded the Mozilla foundation).

So it would actually lead to better and more consistant standards rather than less standards.

Ok there was a time when i would be against this topic, and SL being open source, BUT those days are behind me now.. We all see how bad SL is doing... Atleast if some intelligent programmer got their hands on it they might be able to repair all the problems and listen to us about what we want changed/removed/added.

I'm really shocked at how evenly split this poll is..

Umm, what were you expecting?

What went wrong with OSMP, the Open Source Metaverse Project?

What would keep whatever it was from happening to SL if SL went open source?

I don't think the OSMP was backed by investors and supported by a revinue model... It didn't have a community already tied to it... It was in an early stage of development...

Thanks very much Baba. A lot of my original comments, including several arguments as to why open sourcing would actually *help* Linden Lab's business, were hidden on that post for some reason - I've *unborked* it now, if anyone cares to go back and have another read:

http://www.thedragnet.org/blog/musings/secondlife/relay2006.html

The bit that was concealed is the comment that starts "deep breath..."

... it's just a question of time.

Right now, two parallel projects have sprung up and revealed a few months ago that will make SL's becoming more open (even if not open source yet) inevitable:

The Second Life Protocol Wiki

libsecondlife

The first is an attempt to fully document the whole communications protocol for Second Life, by reverse-engineering it. If you look at the Wiki closely, it already has way over 90% of all the data decoded and classified.

The second project is an implementation of the communication protocol in C# (used to be in C++).

What this now means is that things like the Open Source Metaverse Project can simply get the code and the references, compile their browser with the in-built prim & mesh renderer, and connect to LL's servers; some work is under way to do precisely this. The OSMP will be in a position to become the first open-source, non-LL SL client. It might lack some features — but it'll be oh waaaay fast.

The second thing will be fullfilling many users' wishes: getting all the UI elements out of Second Life's client, and making them run in parallel. We all open the SL client as big as we can manage, just to have some window space for all the UI elements — inventory, chat, IM, scripting dialogue boxes... and still we'll have just a tiny, broken view of what goes around us (and keep apologising with "sorry I didn't see you, my window is full of opened thingies).

Now imagine you'd have your favourite IM client (say, something Jabber-based) opening outside the viewer. As well as something to move Inventory, and map the Inventory folders to a disk folder on your computer. A LSL compiler running in your favourite text editor, compiling things out of SL but saving them to your inventory/object. A text processor to edit notecards. All of those running in parallel, with easy TAB-switching and minimising that works.

Hmm. A distant dream? Far from it. We know LL will never give us that (they simply are in love with their own UI), so users will build them with the above tools. Both the off-world IM system and the offworld LSL compiler (or really... it could be a completely different language, it'll only save to LSL bytecode...) are under way.

And in a few months the protocol will even be simpler — moving from "common" RPC to XML-RPC, where at least all the structure will be easier to code/decode with a few library calls.

So what this means is simply that "the cat is out of the bag". The information is there; people will use it; the open source crowd can start to programme these new tools, as well as new clients, whatever they wish, right now. They do not need to wait!

Curiously enough, this mimicks what I've seen when the Doom game was released in around 1994 on the Internet. It was pretty much close and proprietary. After 3-4 months, the first discussions in the forums explained how bits and parts of the map could be changed, as well as some textures. After another month, everybody was busily creating map-generation tools, and texture manipulatores. Another month comes out, and the first "full-mod" patches are released — effectively turning Doom into another game. And by that time, id Software gave up, and released the specs to the public. It was silly to pretend that the whole specifications weren't already publicly known.

I think that the same will happen with Linden Lab's Second Life right now. A few LL employees even have given those two projects some hints and made some helpful comments — which is the right way to go. They still don't have much exposure, but at least they are public, they have discussion lists, they have wikis, they're referred from other wikis... it's now a question of rolling up your sleeves, and start coding!

Maybe for this "first stage" this is enough (I remember the words quoted from elsewhere: "Linden lab has no real direct competitor, so they can afford to set their own pace, and the advancement has to be done by outsiders".

There are no more excuses now. So grab your favourite compiler and start coding

Wicked, Gwyneth. Thanks for that!

We are all concentrating on the client we run our own pc's and I applaud all the work that is going on to open up the client and provide all those mouth watering client side features.

However, the real genius of the SL design is in the sim server software. It is responsible for executing all scripts, managing rezzed items, havok physical engine, and a heap of management functions. A question I am still considering is whether each sim should be responsible for its own inventory store, whether each sim should have a companion asset server or whether the current system is the best way. Whatever, it should be possible to move from a Linden world to a Microsoft (sorry) world or an Apple world. I dont think this is likely to happen anytime soon as LL needs to recoup its investments.

I have no doubt in my mind that as 3D gets more pervasive, both open source and propriety worlds will emerge over the next few years. LL will need to pull its socks up if it is not to be surpassed by these new worlds.

Ed

Yeah, libSL and the protocol wiki are coming along great ;0 Linden Lab has been very supportive of this project.


I really don't see how the UI in Second Life is different from a UI outside of Second Life.. In fact when you're full screen Having a UI inside of Second Life is more intutive. I guess I'm more into the idea of Second Life as a desktop. With the new XUI system coming along and libSL projects like SLProxy we will have a huge ability to customise the official client. If Linden Lab open sources the client, it will be even easier to start hacking it up.


How about having all of that in the client full screen. Taking up just as much real estate on your screen as any window outside of the client?

I'm looking into a method of hacking Jabber directly into the client right now using SLProxy. If LL releases XUI documentation and takes the bindings for elements out of the executable, we could see an ever greater level of customization with far less effort.

Linden Lab is working at the XUI system for UI customisation. We can already import custom UI elements into Second Life, but the blindings are still locked into the executabe. Hopefully we can get our hands on it sometime soon.

For compiling outside of Second Life, Mono and CIL will make this a lot easier. Linden Lab is looking into ways of allowing CIL uploads. I've been following the mono development very closely.

Actually the move is away from RPC, and towards a RESTful architecture for asset requests and other data such as online status and messaging using HTTP. The problem is that RPC is not very cachable, so it leads to a HUGE overhead in common requests. HTTP has a much greater cachability.

This is the technology that I think could make Second Life a viable 3D web, because it's the same architecture that the 2D web is built on..

It could turn a collection of widely disparate servers into a single system for asset creation(POST), retrival(GET), Updates(PUT), and removal(DELETE). A virtual world can be built with a collection of URIs. A URI references data from some other location and the data is requested for viewing.

The language used to describe REST is pretty damn confusing, so I reccomend reading "How I explained REST to my wife..." By Ryan Tomayko.

The cat is indeed out of the bag ;0


Linden Lab would like to be more helpful, but they're looking at a complete overhaul of their network protocol right now, and anything they give us will totally different in 6 months or so. With the new RESTful architecture, we can hope for more documentation and an API for using the available services.

As the asset system is now, no it's not scalable. Things are changing already on that front with REST. See my above post ;0

Linden Lab need not control all assets through their own central system. Assets could be data that is represented in a simple XML schema when requested by a client or server. They would use a URI, that could map to any location, to represent the location of the asset, and when requested the reply would be an asset in the proper format.

I have been using Yahoo as my theoretical third party.

You have an account with yahoo and you keep your inventory with them. When you want to use an object you pull it out of your inventory, but in reality that object in inventory is just a URI that references yahoo. If your request is authorized, the object is transfered to the simulator and appears in the world.

Still, all scripts and physics running server-side does limit the number of users each server can handle quite a bit. Webservers normally do a lot less work.

I think scripts will go clientside eventually.. I hope

Personally I think the possiblity of the server side going open source at this time isn't gonna happen even though it'd probly do wonders for the asset server. What I *can* see is them making the client open source, however, one thing will need to be changed for that to happen. And thats the removal of FMOD since thats not opensource. QT has a public SDK wich is most likely what LL used so I don't think it'd be an issue. Honestly I can see them changing out FMOD with OpenAL and QuickTime with VideoLan (aka VLC) if need be. They're open source and do their jobs really well. Chances are if they release the client as opensource it'll probly be listed under a modified BSD liscense or the Mozilla Public Liscense (MPL).