A CNET article was published today written by my pal and fellow ‘Burning Man’ participant Daniel Terdiman;‘Second Life’: Don’t worry, we can scale. I am quoted as follows; “My understanding of (Linden Lab’s) back-end requirements are that they’re absurd and unsustainable,” said Daniel James, CEO of Three Rings, publisher of the online game Puzzle Pirates.” “They have (about) as many peak simultaneous players as we do, and we’re doing it on four CPUs.”
This is mostly accurate, but of course out of context. This is an excellent opportunity for me to blog about what I think about Second Life.
First I should correct the above; each Puzzle Pirates Ocean runs on one dual-xeon class server. We have five Oceans. So we have a total of 10 CPUs supporting peak simultaneous users of around 7,000 (which I believe is about what Second Life has). Currently the database for each Ocean runs on the game server hardware, but we are migrating to shared database clusters. We’re not sure how many oceans we’ll be able to support on one database machine. So, let’s say we support our 7k pcu on 18 CPUs or 8 physical servers. Please note that Puzzle Pirates is of course a much, much simpler service than Second Life. The EQ2 example given in the article is a much better comparison, as 3D requires a lot more server-side work, and even EQ2 is not doing many of the clever things Second Life has to do.
Secondly, I wish to preface this rant by saying, as I did to Daniel, that I love Second Life. It is teh awesome. I enjoy talking to Philip and Cory a lot, they are terribly smart chaps, and I think that what their team has made is incredible. They are easily the market leaders in the most exciting part of the MMO space. Indeed, they are the only practical deployed people in the most exciting part of the MMO space, if you discount early and niche folks like Alphaworld and Furcadia.
As I’ve blogged about before, I think that Second Life is the flag-bearer for the Burning Man side of the war for the future between manufactured entertainment and player-created worlds. I’m very much in favour of enabling player creativity. People have done extraordinary things in Second Life and continue to do so.
That said (you were waiting for it, right?) I naturally have issues. Most of these stem from one of the strengths of Linden Lab’s culture; Second Life has been built and operated more like a religion than a consumer-oriented business. I like to joke that Philip went up the mountain and came down with stone tablets describing how to build ‘The Metaverse’. Indeed, when I’ve seen him speak, he likes to refer to these immutable laws.
I’m reluctant to flippantly recite the sacred scriptures, but grossly paraphrasing some of them might yield; ‘It must be 3D’, ‘There must be contiguous landscape’, ‘Thou shalt not Teleport, thou must walk to where thoust would go’ (this one just got scrubbed, which is an encouraging sign!), ‘There must be arbitary code runneth over on the server’, ‘There must be physics runneth over on the server’, ‘ ‘Thy business model must be based on real-estate as a limited resource’ and so forth.
My problem is that I think that most of these beliefs are basically wrong, or at least, not necessarily the case for a successful player-created world. Moreover, I think that they have absolutely nothing to do with what the player wants to experience in what is most definitely an entertainment product (even though the religion forbids calling Second Life a game, it occupies that entertainment time the same way that games do, to the exclusion of things like TV). Often, as with teleportation, they are fundamentally and directly in conflict with what the player wants to do (go where they want to go, right away).
It’s my belief that to succeed in building an entertainment service, the player’s experience must be front-and-center the most important thing you consider. As a game developer, I’m all about creating fun for the player. That’s the main thing.
One of the next most important things for us at Three Rings (note: this entire blog does not reflect the opinion of the company etc. blah blah blah) is that we build such player experiences in a manner which is practical, both in terms of the time and efforts it takes us to make, and in terms of the practicality of delivering the service to what we hope will be a large population of players. Again, it’s my opinion that some of Linden’s ‘commandments’ are in direct conflict with this; it would be my suspicion that it’s not practical to do physics and arbitrary scripting on the server in the form they’re offering, as their back-end requirements indicate. Naturally they could (and hopefully will, for the sake of the A/C at the colocation facility we share with them and many another San Francisco startup!) optimise this stuff and make it much less onerous on the back-end. And, of course, I don’t have any particular insights into where the cycles are going — it may be a user-facing task (like culling the player’s view, or somehow handling their interactions) that put the real strain on the servers. All of this could be optimised into reasonableness. I will go on the record that 3-20 concurrent users per CPU is not reasonable for a service that hopes to scale to become the defacto Metaverse. To run with 500k concurrent users in China, like Audition, you’ll need all the servers in Shanghai, and more.
Another aspect of their business I find puzzling is the real-estate side of things. I understand that physical space is the point of greatest cost of goods for Second Life, so it makes sense to charge a lot of money for it. But as a long-term business model it sounds pretty… well, crap. That’s renting servers, otherwise known as a more complicated kind of web-hosting with fancier proprietary code. Perhaps the exclusive supplier aspect means that the margins are sufficiently great, but I still can’t get comfortable with the idea that Land Barons are going to be pivotal to the future virtual utopia. Instead, I’m going to call bullsh*t on the belief that ‘physical’ virtual space should be in any substantive way a restricted commodity. Sure, it’s another way to make money — on Habbo Hotel or Puzzle Pirates you can buy a bigger hotel room or palatial estate — but it’s a minor part of the overall item-based business model.
I can’t get my head around why Linden would not operate with the same model that we, Sulake, and every Captain Insano Korean game uses; mint currency and sell it to folks, have them buy virtual goods from us (for example land) and, in some cases, other players. When I asked Philip this a while ago he said something about treating Second Life as a ‘country’ and managing its money supply responsibly, as governments don’t just print money willy-nilly. I don’t know enough about economics to really say whether this is sensible or not; I know that a simple faucet (time/money) -> drain (taxes/’state-sold’ goods/item decay) economy works out well enough in Puzzle Pirates and we don’t carry an problematic float of either pieces of eight (time-based currency) or doubloons (cash-based currency)… but we also don’t support cashing out of Doubloons and there is no secondary player market for them, as yet. Perhaps if we were trying to manage a dynamic currency market between the ‘loon and the dollar, we might think about things differently. Or, if we were smarter. My suspicion is, though, that Linden has evolved to their currently business model via subscription and found it to be lucrative, so they’re sticking with it. I believe, however, that it will hamper significant adoption in the long-haul as it puts the emphasis on the wrong aspect of things. Of course my opinion here is influenced by my belief that the physical, landscape-based metaphor has absolutely nothing to do with how people will want to interact with ‘the metaverse’, and that 3D is an attribute, like the colour blue, not a requirement for satisfying player experience.
So, having said all of this, I will repeat; I lubs Second Life. I lubs the Lindens. I don’t particularly want to rain on their parade (and oh my! What a parade! Being in the Economist the same week as a cover in Business Week gets them a huge prize… the Economist is the only print media I read these days.) What they’ve done is awesome and they are in an incredibly strong position to continue to execute, with lots of money, a great team, and some traction with a tremendous community of players^H^H^H^H^H^Hresidents. I love what they’re doing to web-ify Second Life, it’s very smart. If they keep moving in this direction, sort out the back-end, and keep scrubbing those commandments off, they have a reasonable crack at being the metaverse.
However (you didn’t think the Capn was going to end on that kind of up-note, did you?), I believe that the metaverse, should a ubitquitous platform for online entertainment ever emerge on top of the the web (which I would argue already *is* the metaverse) will have lots of properties that we cannot begin to foresee, right now. It’s only through tremendous iteration that we’re going to get to the point where something metaversal emerges. To think that we really have any definite ‘rules’ yet is frankly absurd (I do like that word).
I also believe that such a successful platform will have to be open-source in a pretty deep way (which conflicts, it seems to me, with the land-rental model, though not necessarily with the ‘not-a-b@nk-but-a-bit-like-one’ model).
And, in closing, I am going to put my money where my mouth is (instead of just my foot, as usual), as Three Rings will develop something oriented towards player-created content. Ya’ll will have to wait and see that, but hereby I grant open season to ridicule me appropriately when I come down from the mountain with my tablets… I think those hooded robes would look pretty good on me.