Not yet a party but soon?

Inara Pey did a wonderful review of the new “kid on the block” (her words, not mine!), Cloud Party, which made me try it out. Specially because on on of the snapshot on her blog I saw the Facebook picture of my PhD supervisor who had created an island on Cloud Party 😉 so I guess I was allowed to try it as well and write it off as, uh, “virtual world research”…

After seeing so many VWs out there, most of them utterly failing, I had very low expectations. And I’m glad I had them, since clearly my laptop is completely underpowered to run Cloud Party (CP) — I get about 5 frames per minute under Chrome, and about 10 or so frames per minute under Firefox. That’s at CP’s lowest settings. It makes it pretty much unusable. Granted, SL runs at the lowest settings at perhaps 6-9 FPS, but that’s 50 times faster than CP 🙂

So this is what I get when doing a snapshot. This gets posted on one’s Facebook album, from which you need to download it if you wish to keep a copy:

Not very inspiring, is it? Well, taking a screen print at least shows my avatar:

Since I didn’t expect much, I wasn’t very disappointed. I know I have an underpowered laptop (my so-called “main computer” is so old and broken that it’s been endlessly attempting to repair itself), so no wonder this doesn’t look as nice as what other users have been seeing in Cloud Party.

Registration is easier than creating a new account on Second Life. There is just one screen, to create an anonymous login, or to use a Facebook login. I don’t have a Facebook account (Facebook kicked me out long ago) but for doing experiments I’m fine in registering with Facebook once more until they delete me again 🙂 One assumes that CP might allow different registration methods when it officially launches…

The tutorials, as Inara Pey reports so well, are actually well done. The interface is minimalistic and easy to understand. If I could actually get a decent amount of frames per second, I might even be more impressed with it. It’s easier to use than Kaneva, IMVU, Moove, Blue Mars, or anything else I’ve tested in the past years. It’s crammed full with SLers. Well, not really — there is a 25-avatar-limit — but you know what I mean: when a new VW pops up, for months you will have nobody else around but your familiar faces. Or, rather, unfamiliar faces — since there is practically no avatar personalisation, unless you know people from their Facebook accounts, it will be impossible to know who’s who. But I know a few. They’re all eager SL residents, trying to escape the clutches of LL’s tier monopoly 🙂

Now others have extensively reported on Cloud Party’s functionality, features, looks, and so forth. You can get a list of more to read on avatarium.se; it’s a Swedish blog, but the links are mostly to articles in English. Of particular interest is the one describing the technical aspects of Cloud Party written by the peerless Nalates Urriah. These will be important for all the “comparisons” in the next few weeks.

It’s interesting to see “Cory’s hand” behind CP. IMVU, who has more than twice the user base of SL (or more), focuses on avatar personalisation, and that made them financially rock-solid as dozens of thousands of content creators design the awesomest pieces of clothing using mesh — years ahead of SL. Still, the engine doesn’t do wonders, and at the end of the day, IMVU avatars just look slightly better than SL, even though all clothes, hair, etc. have always been meshes. You can create your own rooms in IMVU but it’s painful, hard work, best left to professional developers. Blue Mars focused on stunning graphics and allowing professional developers to deploy their models there. As such, things looked much nicer than SL, but it was hard to personalise everything. Kaneva tried to create a social networking tool first — you had to participate for weeks on their forums etc. before being allowed to register — and then gave you the doors to a VW which looked a bit better than There.com but not that much better. CP is true to Cory’s beginning love in SL, in 2002: give them great building tools, and who cares about the avatars or the remaining things that make up a VW — let the builders build, let the designers design, let the programmers programme, and everybody will be happy! And, indeed, the reviews from SL content creators have been all highly positive so far. They’ve noticed one or two limitations here and there, but for something that allegedly has been done from scratch by a handful of developers in barely nine months, it’s impressive work.

Now it’s impossible to predict what will happen with CP in the near future… but be can apply a few of Gwyn’s Rules for VW Success:

1) Content creators are just 1% of SL residents.

The remaining are consumers. A builder-friendly VW will quickly attract content creators (good for any launching company) but little else besides. So expect hundreds of SL content creators on CP (like they hopped onto Blue Mars) and nobody else following the trend…

One might argue that new content creators would come just for CP, who would never consider going to SL/OpenSim. Well, after so many years watching these social VW popping up and disappearing again, I have noticed a few trends. First of all, there are a certain (limited) number of types of content creators. There are the Hollywood-class 3D modellers — they’re all employed and have little time for playing around in VWs. There are the Hollywood-wannabe-class of 3D modellers: competent professionals who get routinely employed to do 3D logos or architecture movies to promote condos. Of those, perhaps half think that VWs are not “serious” enough, so, for fun, they create models which they sell in Renderosity or similar “pro” sites. The other half is all doing content for Second Life and IMVU. And then there are amateurs — people that have learned to use 3D tools for their own fun and have no intention to get a job out of it, but who naturally enjoy a few dollars if they manage to sell any of their content. These people are all in SL or OpenSim, since even on IMVU, being an amateur doesn’t pay off.

One might expect that we’re talking about “hundreds of millions of users” that would fit in those categories. Actually, no. It’s more like hundreds of thousands. In truth, the fun issue of SL is that it actually enabled amateurs to produce — and sell! — content, at least during the explosive growth phase of SL. Once that phase went away, only the professional content developers who do not work for Hollywood (yet!) but have pretty much the same skills and talent are in SL… and they’re few. Very few. Many of them also develop content for IMVU. Many will very likely develop it for CP as well, assuming that they can get a new market that is worth the effort. Only a handful have started to develop for OpenSim, for instance, even though the time spent in converting content from SL to OpenSim takes a just a few minutes — because with 30,000 or so active users, the thousands of (mostly unconnected) OpenSim grids are still a tiny market. CP will need to grow a lot to start really capturing the attention of the pro developers. IMVU got the first few pros when it already had millions of users..

2) Facebook “integration” is being completely over-rated on the CP reviews.

It’s totally pointless: the application runs on a Web page, yes, but it’s completely detached from Facebook. Currently you can only post pictures to your FB albums and/or invite friends from an easy interface. That’s the only real “advantage” (but you can do exactly the same within SL/OpenSim as well, it just requires an extra step) that mandatory FB logins give. People are not playing games on FB any longer; it’s getting out of fashion; Zynga is losing market value on its stocks, and all their games have always been perfectly integrated into FB, while CP is not. So “mandatory” FB logins are just great to avoid people logging in from scratch (LL should do that for http://join.secondlife.com/ as well — it’s incredibly easy to do so and would stop people from complaining that they have to type so many things to create an avatar) but it’s not “sitting on the doorstep of the biggest on-line user community in the world” like Inara Pey says. It’s just a convenient way of typing less things to create your own avatar — and making sure that everybody has a profile with a name (which cannot be changed) and a picture (which cannot be changed either).

CP, right now, cannot “leverage” on FB ads. It doesn’t even integrate FB chat within its VW environment. It’s just a quick login solution. It might become one of many when CP launches. They might feature ads from Google, Facebook and Microsoft at some point, inside the virtual world, but… the point is that it’s not using Facebook like, say, Zynga does.

I don’t know if this is a deliberate attempt at over-rating Cloud Party or not, but drop in to the landing zone and this meme is being repeated over and over again — perhaps it’s part of CP’s marketing policy, to try to give users a false sense of “immediate success”.

3) Web-based VW applications are only good for lazy people.

The choice of WebGL is not even necessarily a good one… On a Mac, I’ve managed to use CP with Chrome (slowest) and Firefox (slow); Safari is supposed to support WebGL with a few special tweaks but I couldn’t get CP to connect to the server. Opera doesn’t work even though it’s supposed to. And IE apparently doesn’t, either, not without some extra plugins, which requires users to be computer-savvy enough to know how to install them. So… one wonders how wide the audience actually is. While Firefox and Chrome users are now starting to outpace IE, and I suppose that most Android users will be able to use CP on a tablet or even on a phone, well, I’m not so sure that this will work as well as expected. Why browser-based apps are the “magic dust” that turns a VW into a successful product is beyond me — after all, the majority of Android or iPhone games are native applications, and that has never hurt them 😉

There is always this idea that if you place something on the Web it will be a huge success. It isn’t. Native applications always outperform Web-based ones — at least, there is a bit less overhead. It’s just good for restricted environments where you cannot really install an application on your computer (like on some campus networks) or don’t want to, or where you have firewall restrictions (which will filter pretty much everything out except for HTTP port 80). On the other hand, it’s good to have a Web-based version for lazy people or the ones who don’t know how to install applications on their computers. But even this argument is quickly getting out of date — I seriously suspect that Windows 8 will have something equivalent to the App Store, and once that is in place, it will be as easy to install anything on a PC than it is on a Mac, an iPhone, or an Android tablet 🙂

4) Look at the social aspect.

That’s what will bring new users easily (even though they might leave afterwards). Kaneva is the best example of an attempt to catch people easily and tie them to a social environment first, and push them into a VW next. Sadly, Kaneva did “their own social networking tool” while Facebook was already on the rise, and so pretty much missed the point. Google Lively was launched without a close integration with Google Talk and it naturally failed; Lively 2, if Google ever thought of entering virtual worlds again, would probably be fully integrated into Google+ and be a much better success.

The key aspect to get new users is to make sure that as soon as you log in to an environment, you can quickly find your friends, and invite new ones. SL suffers from “too much anonymity” — this means that most people, even if they wished, are unable to tell their Facebook/Twitter/LinkedIn/Yahoo/Google friends what they’re up to. IMVU is as bad as SL in that regard, even though it has its own social environment, just like Kaneva (aye, I know that SL technically also has my.secondlife.com, but it seems to be one of those eternal “work in progresses” that LL is so fond of starting but never finishing — why isn’t it fully integrated into the SL Marketplace, for example, or why doesn’t it allow sharing your timeline on Facebook, Google+ and Twitter?).

This is where Cloud Party has some room to expand, as all their competition is shy about integrating with social networking. They can — if they’re willing; some already claim that this “beta” version is pretty much the pre-release, with only minor bugs to fix before the VW launches officially and CP starts leasing islands — grab this opportunity. Will that truly make a difference? Perhaps, to quickly attract new users. But the problem is converting them into paying customers.

5) Have a solid business model beyond venture capital.

I must have written this a billion times, but it doesn’t hurt to repeat it again.

Tech companies still live in the pre-dot-com era, and what is more surprising is that their venture capital funders still think this can work! Their assumption has remained the same one since 1997: give access for free, grow the number of users into the millions, and “somehow” it will work out fine.

Well, let me give you an example. Suppose you’re looking at Facebook’s share prices on the stock market and wondering if it’s a good idea to invest in them, now that they’re low. Facebook is certainly going to stay around for another decade or so, and they are the largest online community ever, with (allegedly) 800+ million users and still growing — not as fast as before, of course, but growing. Surely they’re a solid investment, right?

They also make about a billion US$ in advertising per quarter. That a bit more than a dollar per user. Not bad! Or is it? Depends how you compare it. There were 350 million Gmail accounts in January 2012. Google made US$42 billions in ads. So that’s around 120 dollars per user per year — thirty times more than what Facebook makes! Now do Facebook shares look so good?

“Wait!”, you’re now saying. “This isn’t right, because you cannot tie users of one system to the other… it’s thanks to Facebook’s global community that they make money from ads… Google makes money from billions of web pages which are not related to Gmail accounts at all… you cannot do the maths that way!”

I know 🙂 but tech journalists don’t. So, to make sure the media got it right, Apple’s new CEO, during the keynote speech at WWDC, made a quite astonishing comment. Besides all those millions of gadgets out there, Apple has 400 million paying registered users on iTunes/iApps. And makes $8 billions out of that. Yes, it’s twenty times better than Facebook. But that’s not all the story. That’s just 6% of Apple’s revenue. What they’re saying is this:

“Look, we all have heard about that cute little company from the dot-com era known as Facebook, right? Sure, the guys are great, and they even made a billion in ads last quarter. To reach that number, they had to grow to 800 million users, and were unprofitable during most of their time. That’s their single source of revenue: ads. The ad giant Google fares better: they made $10 billion in the same quarter. Both have no easy way to link ‘users’ to ‘revenue’ except on the media, who will publish any kind of junk. More users doesn’t mean more revenue; it’s just the media that thinks so. But it does mean higher costs to support more free users which consume bandwidth, CPU and memory. How many of those 800 million Facebook users and 350 million Google users are actually paying anything to those companies to offset their costs of running those services for free? They’re not saying! Well, we are. In the last quarter only, our 400 million paying customers bought $2 billions of digital content on iTunes/iApps. And that’s just for our spare change, since 94% of our business model is based  on selling hardware and not digital content…”

That’s why Apple shares are hot, Google shares are not too bad, and Facebook is a scam.

When we try to apply this to virtual worlds, we have to analyse what each company is doing. Linden Lab is profitable. They have a million active users. These users transact half a billion US$ yearly. LL only gets a tiny slice of that on the LindeX, but because transactions happen mostly in-world, it means leasing land, and that’s where the bulk of LL’s revenue come from. And at the end of the year, LL gets 50 or 60 millions in profits after taxes. That’s not too bad. It’s not Facebook, nor Apple, nor Google. However, if the business model could grow indefinitely, and LL managed to get, say, 800 million active users, they would, from their tiny slice of the proceedings, get $20 billions in profits — half as much as Google but twenty times as much as Facebook.

Where is the problem here? There are not enough people on virtual worlds and never will be. This is a very hard fact to swallow. More on that later, but the point here is that LL cannot grow — it has reached the upper limit. Facebook can, at most, grow to 2 billion users — which is theoretically possible, if 100% of the Internet users in the world had a Facebook account — but that wouldn’t mean anything in terms of ad revenues. At best, they would get twice the profit. Which is nothing. Google, by contrast, doesn’t care about growing “users” — all they need is that the Internet grows in sites — more sites equals more opportunities to sell more ads, and it’s unimportant if Gmail accounts grow or not (except for the media to post nice numbers on their articles). Thus, they have a much safer business model. Apple has a trickier one: they have constantly to release new gadgets 🙂 and redefine the tech world each time they do that, which is way harder to do than the sit-and-wait approach used by Google…

IMVU also has a very interesting business model. When they started, they put out some ads. Each time a user registered through an ad and paid something, they would take part of the money received and invest in further ads. They did it over and over again until the whole of the Internet was swamped with IMVU ads. But the strategy paid off: the more ads were seen “everywhere”, the more likely they would get a paying customer, and the more ads they could buy. This trend seemed to have stopped, but at least they have turned IMVU into a profitable company — mostly thanks to the ability of allowing content creators to make money as well, of course, and taking a commission from sales.

Now, can you point me out what was Lively’s business model? Right. You can’t. How does Kaneva make money? I have no idea. The only guys that tried to replicate LL’s model so far were Blue Mars — and they nevertheless failed, since the price difference was not so big, and the amount of content and users was limited. Apparently Cloud Party is going the same route as Blue Mars and compete with Linden Lab on island tier and selling currency, too. The challenge is to grow the userbase — the paying userbase, which is the only one that counts! — and become profitable until their venture capital runs out. That might happen in 6, 12, 18, or 36 months — at best. That’s what they have, as a realistic time frame, to prove that they can beat Linden Lab at their own game — not in technical improvements or social features, not even in number of users, but in generating revenue using the same business model.

This won’t be easy. Why? Mostly because of the next rule:

6) There aren’t that many people interested in social virtual worlds with user-generated content.

In the Golden Days of SL where growth was exponential, the sky seemed to be the limited. Optimists like myself believed that “everybody would want to be in a 3D virtual world ” — it was a natural evolution from outdated, Web-based, 2D social environments. Guess what? We were all wrong. There are just a few millions that regularly enjoy social 3D virtual worlds with user-generated content. How many millions? Well, I’d say about… two. These are the regular users faithful to SL and IMVU. There aren’t more!

On top of that, there are perhaps 20-30 million casual users, but they won’t become paying customers. If one stretches this to reach out to 2-1/2-D VWs like Habbo Hotel (who has failed in spite of everything), there might be, perhaps, 50 or so millions. But there simply aren’t more. It’s not a question of more marketing, easier interfaces, simple setup, or available content. It’s simply a question of mindset, and you can’t change mindsets easily. It takes generations. It’s thanks to the current batch of educators working on virtual worlds — numbering dozens of thousands, perhaps hundreds of thousands at best — that the next generation of users might be more familiar with 3D virtual worlds and eventually join them; but that won’t happen in the next decade. It might take a whole generation.

In the mean time, Second Life, OpenSim, Unity3D-based VWs, Kaneva, IMVU, and whatever “survivors” are out there — as well as new start-ups like Cloud Party — will have to ferociously compete among a very small user base. Well, small, yes, but valuable. Half a billion US$ annually in digital content sales is not bad. It’s not as good as Facebook ads, or Skype minutes, or applications, or e-Books from Amazon.com, or music… but it’s not bad either. But there isn’t more: it’s all there is.

What can Linden Lab do to survive?

Nalates Urriah and Inara Pey, among many others, correctly point out that, in the past, even just the possibility of competition made them improve their virtual world, which benefitted us all. Ironically, LL is so slow in responding that once they finish their development, their competition has already failed 🙂

When IMVU became very successful thanks to their very easy to use Web-based marketplace, LL had to buy one and desperately try to compete there as well. They still have problems with the SL Marketplace. There is a big problem for LL if the SL Marketplace ever becomes “perfect” — if it starts seriously undermining in-world shops, it means that shop owners will drop their land, which means less revenue for LL. And they don’t recover that lost revenue from SL Marketplace sales. So SL Marketplace has to be “as good as IMVU’s” but not much better.

When Blue Mars started to show a much better rendering engine with fantastic mesh support, and a different model of paying tier (one BM “island” could have 500 MBytes of mesh data, but it was up to the content developers to “fit” all the content they wished on that upper limit), LL started working on a better engine and a new viewer. Blue Mars collapsed before LL finished their mesh-based engine, and only now, with Land Impact, are starting to review how tier is connected to content. Lucky LL! Their competitors fail too easily 🙂 But then Blue Mars created an iPhone application and made LL wonder if that wouldn’t be a good idea to copy. We still have no clue if LL is doing a mobile SL version — even a “lite” version of SL — but there have been rumours around it. Blue Mars Lite might fail before LL releases their own mobile app, though, and, very likely, like many of LL’s projects, it will remain semi-abandoned for years.

Unity3D-based VWs like Jibe from ReactionGrid have been around for some time, and they target primarily educators and corporations needing simulation and training facilities, where the ability to launch a VW from a Web page is crucial for them. Cloud Party might have turned that feature into a mass-market product. This will mean that it’s likely that LL will need to finally develop their own Web-based viewer as well. It’s not that it’s too hard; there are already a few one-developer-projects which create conversion tools between a C++ application using OpenGL (like the SL viewer) into JavaScript applications with WebGL. Or you could do a new renderer from scratch, like Radegast which runs on .NET/Mono or Lumiya for Android. Whatever approach LL takes, it’s a technical possibility — it just needs developers to be assigned to it.

But is that enough? Hamlet Au, the eternal doomsayer, reports that Second Life might lose 10% of its landmass by the end of the year, based on a report by Tyche Shepherd (who is a RL statistician). This would mean that LL would barely make a profit in 2012, if at all, and their execs will be pressed to come up with solutions.

Well, let’s try simple solutions.

First of all, the meme that “Cloud Party will be an immediate success because it allows Facebook integration” has to be stopped. Quickly. But in a positive way: all Linden Lab has to do is to enable Facebook registrations, and do it quickly. That’s so easy to do and will take just a week of development time. If they wish to do it even better, they can simply subscribe to Janrain Engage, which will handle authentication and “integration” with a dozen different identity providers. With a few lines of extra code they can start sharing my.secondlife.com timelines on Twitter, Facebook and LinkedIn as well. And then they have to bring out a press release that they have completed “integration with a dozen major identity providers world-wide, including Facebook, Twitter, LinkedIn, Yahoo and Google”. That should cool down a lot of overhyping around the “importance” of being connected to all those social sites and identity providers. To understand how stupid this claim is — my own blog also allows people to “register” using half a dozen identity providers, including Facebook. That does not mean that I have “an expected audience of 800 million viewers”! Actually, over the years, as my readership has declined constantly, integration with so many identity providers has done absolutely nothing to improve my own site’s traffic! And so it will make no difference to Second Life, either — except for the media.

Then they might be able to integrate some of the functionality that some third-party viewers have already provided, like the ability to post pictures directly to Flickr. Or even Facebook. Or even Twitter. Most of the code is already published somewhere else, and it wouldn’t be insanely complex to integrate it into the “main” viewers. Again, it requires very little development, and has a huge impact on the media.

Development of a full-fledged Web-based/mobile viewer is something that either LL has been developing for long, or it will simply take too long to make an impact in the short term. It’s also only hype — the amount of people actually using a Web-based viewer for SL will be small. LL, in the past, were obsessed with the concept that Web-based/mobile viewers needed to convey “the full range of SL experience”. That’s pointless to aim for: go for something simple that works, which shall please the media, see the reactions, and only then work with something that is “perfect”, which might actually never be. In the mean time, the few thousands that might be in desperate need of a Web-based or mobile viewer will be happy and write articles about it.

Of course it’s impossible to figure out the millions of features that SL needs and fix the trillions of bugs that SL needs to have corrected. But, in truth, they don’t affect residents that much.

What affects everybody is… tier. Even if it’s just indirectly: a free account might not worry about tier, but they will worry that their favourite spot, which is being paid by someone else out of their pockets, disappears because the owner cannot afford to pay any longer.

Well, the good news is that this requires little development, but merely a policy change. And here I have to borrow the idea from my good friend Scarp Godenot — move to a model based on sales tax on L$ sales.

Before you all start to collapse on your chair and yell “NO!!!” consider the following: LL already does that on the SL Marketplace, and people are fine in selling there. Why? Because they don’t need to pay tier on the SL Marketplace. The more content creators are on the SL Marketplace, the less they pay tier — since they can close their in-world shops and move to the Web instead — but that means LL is going to lose even more money. Trying to “replace” in-world sales with SL Marketplace sales is not a good strategy, long-term.

Now imagine the following scenario. Suppose that someone rents a whole sim for US$295 for their super-mall. Currently, they pay zero taxes on sales. So what that means is that as soon as a content creator makes more than US$295 monthly, they have covered the costs of tier, and everything else is profit (well, minus labour costs in creating the content first, as well as, ads, and so forth — but you know what I mean!).

For some content creators, reaching that amount of sales is easy. But for most it’s not. So here goes an alternative model. Instead of paying for the full tier, you pay a sales tax on every item sold on your region. The more you sell, the less tier you pay to LL, who will deduct the amount of sales tax from the region tier cost. So if you don’t sell anything, you will pay LL the full cost of the region; if you sell a lot, you might be able not to pay LL anything at all. If you sell so much that you have paid everything to LL for that month and you’re still selling more and more, then LL is getting a lot of taxes just from you. If that annoys you, you have a good choice: get a second region.

Perhaps it’s best to give an example. Let’s suppose that the sales tax is 10%. If you do zero sales on your region, you pay for the whole region — US$295. If in one month you sell the equivalent of US$10, then LL applies a sales tax of US$1, and your region is just charged US$294. But if you’re one of the top content creators and make sales equivalent to US$2950, your region is free of charge — all the money collected in taxes have been used to pay for your region. If you sell US$6000 in a month, then you get your region for free, but LL also gets an extra US$305 on taxes, beyond the US$295 they have received and which went to “pay” for your region.

Now this doesn’t automatically mean that LL is going to earn gazillions from taxes. What it means is that a  clever region owner will try very hard to pay as little for their regions as they can, by adding as many regions as they can, as their income from those regions increases. The result? Successful business owners will start adding landmass again, just to be able to cut down on tier costs. Think of it as a “tax return on investment”: the more you invest in landmass, the less total taxes you pay.

Notice that this model has two advantages. LL is not going to lose money just because they “give” you a region for free, if you happen to be very successful in generating sales — they collect taxes, after all. People not doing any sales will not be hurt — they won’t pay anything additional, but the cost of their own non-profit sims won’t be raised. They will remain at the same price. Land barons won’t cry havoc because of a sudden reduction in prices. Instead, they know that the more rent they collect on their own regions, the less they will pay for them. The same will apply for mainland region ownership as well, of course; and possibly it might be even feasible to create a model for private regions with individual parcels to apply a two-step tier collection process, where sales from individual parcel owners contribute “tax” to the region owner, who in turn will get a cheaper region — and a free land market will make region owners pass the reduced amount on their clients. Things get trickier with group ownership, and items distributing money among several people, but I’m sure the clever people at Linden Lab can design a model to deal with those cases.

The theory behind this model is that it allows Linden Lab to announce “reduced tier” for anyone who actively participates in the SL economy, while not needing to lower the tier costs (which would reduce LL’s profits even further) nor apply a “general tax” on everybody, which would be unpopular. Note that buyers don’t pay taxes. It’s only sellers that pay it. But sellers will need land to offer their products, and the sales tax will contribute to pay for the tier they have. So people are not really “losing” much that way; the only ones losing anything would be non-profits. But since for all purposes a “donation box” is just like a “vendor”, money received by a donation box would also pay taxes and contribute to lower tier. In fact, this system is really not much different, except that it psychologically means that LL can announce “a programme to progressively own land cheaper — potentially get it for free if your business thrives!”

But that’s not all that LL can do. They can also work on the side of dramatically reducing costs, and by doing so, offer new kinds of regions for a much lower price.

I call them “blink regions”, but they’re really pretty much the kind of service that Kitely already provides, just done with LL’s own infrastructure and integrated into a visually contiguous virtual world.

What are “blink regions”? When you blink, the whole world disappears. But when you open your eyes again, nothing has changed — it’s all there. That’s why nobody notices when they blink their eyes 🙂

Take a look at the SL map. How many regions have green dots on them? As we all know, green dots accumulate on popular spots, but the rest of the grid is mostly empty. Now this is rather stupid from the perspective of LL, because so much CPU power and memory is being spread over the grid, useless for all purposes, because nobody is there to use it. Since LL needs to have the simulator software running all the time, they have to charge them in full — even if nobody is visiting them.

Kitely correctly identified this issue, and does something clever on their OpenSim “grid”. Every time someone logs in to a region, it pushes an image of that region to an Amazon server hosted on their cloud, and launches it. This takes very little time. Once the region is up, it works pretty much like a normal region, and users are charged for the time someone is online. When you log off the region, and nobody is there any longer, the region is saved to disk and removed from Amazon’s cloud — and you don’t need to be charged any more. This is all done automatically and is great for “temporary” setups which will be just used a few hours per day or so.

There is really no reason for LL not to implement something very similar, with the only difference that there is visual contiguity on the grid. When you open the map, it looks just like before — but actually the only simulators running regions are the ones with green dots. Empty sims are unloaded from the servers. As soon as someone teleports to a “blink region”, its image gets loaded by one available server, and instantly restored — the resident will not notice a big delay, since they will be watching the teleport screen anyway. If a lot of people pop in, and a sim is full all the time, well, then it will be using CPU and memory around the clock, and LL will charge the region owner the full price. But if someone buys a whole region and just uses it one hour per day, then the region will be charged at merely US$12/month. Think of that! Having a full region of your own, with all the 15,000 prims, with all the quality that LL guarantees on their grid, for less than most OpenSim operators charge. Sure, you will only be able to use it one hour per day (to keep the costs low) but that might be enough in most cases. Even if you use your region several hours per day with your friends, it might still be much cheaper than anything else out there. And from LL’s perspective, they’re just saving resources, or rather, using them more efficiently.

In practice this requires some more development: effectively this pretty much means running the whole grid as a cloud service, which is something I never understood why LL hasn’t been doing for years — as there are lots and lots of open source and free cloud management systems out there. LL might very likely be able to run the 30,000 or so regions they have on just 50 or 60 servers, instead of the 4 or 5 thousand they allegedly have (probably less these days, as they’re running more and more sims per server, thanks to an increase in core density). This is because cloud technology also uses virtual servers, which can get allocated more or less CPU and memory depending on usage, and we know that LL already runs their simulator software on virtual servers anyway — all they need is a cloud supervisor software to dispatch simulator images to free CPUs across the grid. And under such a model they could follow Kitely’s example and just rely on Amazon’s cloud service instead, and just keep the handful of central asset servers on their co-location facility. The costs savings would be tremendous, even though I still believe that the biggest cost that LL has is with their staff, so this option would really need to be carefully evaluated in financial terms. But the end result would be the ability to cut costs dramatically for everybody that doesn’t really need “always up” regions but just “blink regions on demand”. Obviously large malls or event areas which are in use 24h/day will not see a tier reduction that way — but there is nothing to stop LL to adopt both strategies: sales tax with tier refunds plus a cloud-based grid. So the more you sell, the more likely your region is always up, but the more likely your sales will pay for tier costs. Smaller shops with little traffic will raise little sales — and thus little taxes will be gathered — but since they will only cost a fraction of a total sim (because they will be little used), this might be enough to get even small shop owners their land “for free”.

Of course there is a catch: ‘bots. ‘Bots roam the grid in search of vital statistics, and by doing so, they will automatically instantiate a region, which will get charged by the time that the ‘bot is active. Since you cannot prevent ‘bots from teleporting to your place — since ‘bots are only voluntarily labeled as such — there is a problem here. Even if SL — like OpenSim — provided a lot of statistics via web services, thus rendering most ‘bots obsolete (and one wonders why LL has been so reluctant in opening up more sim statistics via web services…), there will always be “legacy” bot sytems roaming the grid, as well as griefers and abusers. So I can imagine that this model would require some thought. Perhaps LL could just start offering “blink regions” as an optional service, at a lower cost, while still keeping the remaining services just as they are. They also might offer the service in “chunks” — say, owners could preset their regions to be on at most 72 hours per month (1/10th of the time) and commit to pay a tenth of the price. This would give them two and a half hours per day or so. But if a griefer starts adding ‘bots to force the region to be “always on”, the owner would just get 3 days of service, and then the region would be placed offline. Before that happens, the owner would get a notice, and could look into things to see what happened — and notify LL about the abuse before the time runs out. I know this is a complex situation (they might demand refunds and so forth) because LL is so reluctant to deal with abuse reports (as so many are fake) — it would mean that LL would really have to develop better tools to track down who had visited the region and for how long, which would be part of the Estate Tools, so that both LL and the region owner could track down potential abusers. As said, OpenSim includes all those statistics and makes them available through easy-to-understand web services in multiple formats. LL’s technology is quite different but I’m sure they have a way to get the same data as well.

A variation to prevent abuse would be to limit usage to non-paying customers, finally giving Premium users some real benefit. Imagine that non-paying accounts would be limited to be in-world, say, one hour per day (or perhaps a total of 30 hours per month). That way, a griefer ‘bot could not do much damage on its own, and LL has ways to figure out if someone is launching a lot of ‘bots from the same network and prevent that from happening; for example, they could restrict one non-paying account per unique IP address/CPU. This would not prevent distributed denial of service attacks, where hordes of abusers coordinate their attacks from multiple locations, but it would go a long way to prevent the simplest attacks and make the others harder (there are not so many large groups of griefers coordinated together any more; they have lost interest in SL).

Premium users, by contrast, would get unlimited access to SL. But those who abuse the system by forcing other people’s regions to be up all the time are easily charged — LL has their credit data online. Abusers could be blacklisted or even sued, since LL knows who they are. Giving Premium users some real benefit — the ability to be around all the time! — is also a good idea. Let’s be honest: Premium is cheap — US$6 per month if you pay it annually, and you get L$1200 free to spend every month on top of that, so it’s really a bargain. There is little reason why a regular user shouldn’t pay a few dollars per month just to be able to remain around much longer than the non-paying users.

To summarise, what I welcome most with Cloud Party is how it will push Linden Lab in reflecting a bit about their options and come up with ways to compete better with a worthy competitor. At this stage, Cloud Party is just a “SL killer application” for a handful of content creators (aye, I know there can easily be a few thousands of those around). Cloud Party is not SL, but it’s close, and a lot of meshed content can be exported with some ease into CP. It might be easier to setup and use than competing technologies like Jibe and the many other web-based virtual worlds launched by SL exiles, all of them looking better than SL in certain aspects, but quickly forgotten by the media as they are drawn into oblivion — except by their owners, who sell their solutions directly to customers (mostly educators and corporations looking for simulation and training environments in 3D) which might not even have heard about SL. SL die-hard exiles are happy on OpenSim, because it’s a much “closer” clone, even though the reliance upon a SL-inspired viewer means that there are no technological breakthroughs on the visual side, which Cloud Party might be offering now. Cloud Party also might be drawing some IMVU users, because the IMVU content creators will have easy ways to port their content to CP, and CP offers easy building facilities, not unlike SL, which IMVU currently lacks. Nevertheless, this is an old story — every other year or so (albeit less in recent times!) some start-up creates “the next SL killer app” by targeting content creators and programmers, because that was how SL was launched in the first place, by appealing to them with a modern renderer with lots of features and high visual impact. But these companies all shut down quickly after burning all venture capital because they haven’t found a valid business model to become sustainable long-term.

The big issue will not be if Cloud Party offers more and better features than SL. Some are already much better and not easily duplicated by LL in the short term. But in the past, a lot of start-ups have developed better virtual worlds than SL. Many of those allowed user-generated content. These days, I don’t even remember their names, and need to browse through old articles and YouTube videos to remember how cool they looked like. The current batch of virtual worlds that are still around all have companies with valid business models behind them. Except for IMVU — which has a larger user base than SL — the majority of them have a tiny amount of actual users, but they have companies behind them with sales forces which push their solutions to educators and corporations, and are more “development platforms” than independent virtual worlds for average users. I really have no idea how most of the generic, mainstream virtual worlds (like Kaneva) are still able to be around and afford paying for all costs — the few that survive are oddities.

So to be able to predict Cloud Party’s success we will need to understand how their business model is presented and how the company will promote its product, and what their target is going to be. Right now, it’s a tempting beta application, gathering as much attention as Blue Mars did when Blue Mars was in beta, too. The real test will come when it gets officially launched and has to struggle to survive in a very small market.

Because even the OpenSimulator grid operators have learned a tough lesson: it’s very easy to launch a virtual world (specially if the software costs are zero!) and announce prices to undercut Linden Lab. But as soon as a grid starts attracting people, the costs will skyrocket very quickly. The largest OpenSimulator grids are charging prices closer and closer to LL’s own, but offer a worse service — for residents disgruntled with LL, it’s the only option, but there aren’t that many as one might have imagined from the forum discussions and blog posts. A new technology with new features which radically departs from LL’s technological platform might be very welcome, but one also has to wonder how well it will scale. For example, when a Cloud Party island reaches the 25-avatar-limit, it gets sharded — a clever way to allow many more people to see the same content! — but obviously that means that the next 25 avatars will not see the first 25. They will be on separate islands. If you’re planning to do a mega-event with 400 avatars on the same location, you won’t be looking for Cloud Party to host it. Second Life might just barely manage it, assuming you’re willing to fight the lag. Intel has developed a technology called Distributed Scene Graphs to allow “hundreds of avatars” on a modified OpenSimulator version (it’s currently publicly available on Dreamland Metaverse) which is perhaps the only way to get huge regions with lots of content and lots of avatars share the same experience — but it’s not (yet) in widespread use as it’s still in the alpha development stage, although it sounds promising.

To conclude, it’s not that easy to compete with Linden Lab’s Second Life as most people would like to believe. What has Cloud Party and that makes it so special that Blue Mars, Project Wonderland, etc. and all the “missing” virtual worlds had not?

That’s ultimately the big question. Because for Linden Lab’s Second Life we know the answer: one million active users, billions of content for sale, a solid financial model (both for the company and its residents eager to sell digital content), and a lot of luck to have been at the right place at the right time. It’s never been “features”, “cool visual effects”, “ease of use”, “advanced social networking tools”, nor even “being dirt cheap”, because Second Life has none of those and still thrives…

[UPDATE: Thanks to graymills for pointing out that Intel’s Distributed Scene Graph (DSG) technology for allowing hundreds of avatars in a single OpenSimulator region crammed full with content is publicly available and can be actually downloaded and tested. I wasn’t aware that the source code was public or even that some OpenSim grid operators were already providing that service. My sincere apologies to Intel for the mistake, which was amended on the above article. Keep up doing your fantastic work on OpenSimulator, Intel! 🙂 ]