Sadly, as most of you have noticed, there has been not enough time for me to keep the blog updated… and just when everything in Second Life® is about to change:

Linden Lab is now profitable

In an unexpected side remark, Jeska dropped the hint that Linden Lab is now a profitable company, and has been as much for quite some time. No, they’re not making billions; but finally they’re able to rely on their income to plod through the next stages, without relying upon external sources of investment.

Will this make them more attractive for a potential buyer? Hardly likely. The business model will not allow anyone to “make money fast”; now they have an even better incentive not to consider the buy-out option; and open sourcing the technology makes Linden Lab uninteresting to potential buyers (would, say, Microsoft buy a company with an open source product? Never!).

This means that Linden Lab will not disappear shortly, as people have been predicting for, oh, perhaps three years now :) — but will continue to grow and expand their operations. Slowly, of course, but on a solid and positive cash flow, a good income, and a marginal profit. Just what it takes to continue the Road to the Metaverse.

Voice in SL

Due to launch on the next release (May 23), I expect eagerly the impact this will have on people’s lives, when one more layer of anonymity is finally revealed. No more pretension of being someone you’re not; the mist of illusion about your true self is dispelled and Second Life will slowly abandon “the place where you wished to live” to become “the new communication medium of the 21st century”. It’ll be a huge jump into unchartered waters – specially because, as so many point out correctly, there will be no alternative. All Metaverse wannabes these days include voice, even the ones that are just vapourware right now.

Sorry, immersionists — you’re out of this game. It’s augmentism from now on that will dominate the shaping of the Metaverse.

The next step will probably be some sort of face-morphing based on a picture on your Webcam, that will allow your avatar to assume some of your real expressions and present them in-world. Physical avatars, probably to be launched this year, will be a first step — doing gestures without the need for animations — and face-morphing is available from Logitech as a popular, mainstream product. Remember that Linden Lab now employes FX specialists that have worked on movie productions like 300. They know all the tricks in the trade.

Bye-bye, anonymity — it’ll be really you in-world.

Resident Validation

The hottest issue in SL since the prim taxes were abolished. Linden Lab has thought it the following way: if a Windows user downloads Internet Explorer from Microsoft and connects to a site running Microsoft’s IIS, and views child porn, who is responsible — Microsoft?

Of course not. Internet users are responsible for what they watch, and content producers are responsible for what they put on the Web. Microsoft, in fact, warns on their Terms of Services that their products are not to be used for illegitimate means.

And so does Linden Lab. But to go a step further in forcing residents of SL to accept that now they’re responsible for what they do, they’re introducing (probably also in May 23) a new, opt-in, validation system. You’ll be able to send your real life data — name, address, and an ID card, depending on where you live in the world — to a third party company, Integrity Services, and through an API established with Linden Lab, they will validate your real identity, and LL will be able to flag your avatar as “validated adult”.

In return, you’ll be forced to flag all your content and your parcels as being adult or not. Only validated avatars will be able to hold adult content in their inventories, and enter (or own/rent) adult parcels. No more fear of teenagers seeing those naughty sexy ankles.

LL will rely on the community for flagging content that is mislabeled, and abuse-report it — just like it happens in, say, YouTube or MySpace. So most legitimate users will flag as much content as mature as quickly as possible to avoid being shut out of SL — and expect their close friends to do the same.

The issue is polemic at several levels. Many residents have established a relationship of trust with Linden Lab — and their laissez faire mentality — and are positively uncomfortable with third party companies validating their real life data. In the past, companies providing validation systems used by popular pornography sites have been targeted by the FBI as potential sources of many sorts of frauds and of “leaking” information to spammers (emails), marketing agencies, and even US government agencies. Speculation or fact? It depends on what newspapers, e-zines, or blogs you read. A better approach would have been to give residents a choice of company to work with.

So, what will this impact SL? Sure, a few will leave — so what? There are always people leaving — SL has as low as 10% of retention rate, and that is not an issue for LL.

The impact, however, will not be in how many people will leave (my estimate: about a million in a month, but we will hardly notice that, since SL almost grows by the same amount… it’ll just be a “flatter growth curve” for a while), but how they use content. The ‘adult’ market in SL — specially the one pursued by amateurs — is huge. How many percent of all transactions in SL — items, land, services — is on adult content? We don’t know the exact data. We only know that itmust be huge, or the landscape would not be crammed full with shops offering all sort of kinky clothes, escort clubs, casinos, and all types of very mature items, animations, and attachments. They’re so ubiquous as to come to us as “second nature” — they’re part of the landscape, like the Linden trees.

They will not disappear overnight — but their customers will. To enter these shops, you’ll have to be validated, and this means sending your data to a company most don’t trust. Even if the fuss of going through that process is not much, many will never bother with it, and leave a “puritan” SL for something else… well, mostly, pronography on YouTube or MySpace, which is easily accessible and doesn’t require even a credit card.

I expect a “transition” period where people will offer to buy ‘adult’ items, unflag them, and give them to friends (thus making transferrable items fashionable again, hooray!); and others who will validate themselves, flag a whole sim as adult, and then allow people to rent plots nearby and allow them to use their cameras to close in to the desired adult content. Residents of SL are very creative in bending the rules to make a profit out of it — but this will be mostly a transition phase. At the end of the day, we’ll have a mostly Disneylandish landscape, all PG and politically correct, and “ghettos” where the few validated adults will enjoy themselves to the fullest.

For companies coming in to Second Life, this will be a boon. The landscape will be stripped down of prnography, illegal gambling, prostitution and even probably violence — the perfect environment to show off your corporate exec that this “metaverse thingy” is what your business has to be in.

Also, professional content creators who do not fear the conspiration theories will now have a huge advantage in the adult content production — the competition from amateurs that pop in under different alts every day will simply disappear. So, in a sense, the market will shrink, but the quality will increase, and probably the ones that are willing to get validated will be available to pay much more for high-quality adult content.

So it’s not “a huge loss”. It just means things will change a lot.

“Governance Tools”

At the beginning, this will be something quite simple: the ability to “subscribe to ban lists” directly from a tab on the “About Land” parcel dialogue box. What will this provide? Well, a BanLink “clone” that does not require in-world scripts/prims on your parcels, and that will be easy to operate. Just like BanLink, once LL releases the protocol (ie. what an external webserver needs to provide — probably just a simple yes/no query if an avatar is banned), this will be opt-in, and anyone will be able to create their own ban list systems, and subscribe to any list they want. Linden Lab will also have their own.

What this means is that Abuse Reports will be quite differently handled. First, they will go to Estate Owners (LL on the mainland), and Estate Owners will do whatever they please with an Abuse Report. They might simply add the offending resident to any lists they’ll subscribe — and this will propagate instantly to all parcels which subscribe to the same list. So, all you need is to subscribe to lists being run by people you trust, and you’ll be safe from griefers forever.

Obviously, this also means — no appeal and no recourse. LL will do nothing about that. If Rude Avatar hates your guts and places Jane Doe on a ban list that blocks her out of half of the world, that’s it. Jane has no way to appeal that decision. It’s done. She can obviously complain to the list owner, but very likely get no satisfaction there. A very few lists will have attached a whole judiciary system, but very likely they won’t be popular — people in SL want the immediate exercise of personal power (or trust friends), not complex systems.

Actually, this is not fundamentally different from world-wide anti-spam and anti-hacker lists; they work under the same assumptions, perhaps with a single difference — they usually are incompatible among themselves and require different software. In Second Life, there will be a single mechanism that will work for all lists.

Although this will make griefers in SL very short-lived, the question that is always raised is how you’ll get fairness and justice — how can you prevent one person from destroying your enjoyment of Second Life by placing your name on all lists, thus effectively banishing you out of SL except for your parcel? Without appeal and recourse, and LL effectively out of the loop, ostracism will reign supreme in this libertarian world — not justice nor democracy or the right to a fair trial.

It worries me, although I expect that most “unjust” cases will simply be wiped out of existence. Let’s be fair: if you’re ostracised in SL, you won’t log any more, ever again, and not waste time with SL any more. You’re out. And the number of the ones who will never log in again will never raise a huge drama, simply because they won’t be around to raise it.

It’ll be a huge change of mentality in SL, since, after all, people hate others for all possible reasons, and now they’re able to exterminate others with a click on a website. Give atom bombs to children and see how long they can refrain from pressing the button. Again, a blessing for the ones surviving the “nuclear winter” — no griefers, only happy people around — but at the expense of justice, fairness, and a violation of your human rights. I definitely have some issues with this — the idea is good from a technical point of view, but I fear the social consequences, and would have preferred to have LL as a “last court of appeal” in every case. Like on “abusing” Abuse Reports, people that abuse the system should, in turn, be subject to sanctions by LL themselves.

However, LL has stated very clearly that they wish to have as little as possible with resident arbitration and moderation — for legal issues mostly, but also to allow international growth under different laws — so I guess this step was unavoidable.

For myself, you’ll be certain to always get a fair trial from the list that eventually will be run by the Confederation of Demcoratic Simulators :)

Linden Lab’s winking at professional 3D modellers and artists

If you’re a professional 3D modeller/designer/artist, good news for you. Also on May 23 (hopefully!), Linden Lab will give you a new primitive type: sculpted prims, a fancy name to introduce NURBSesque shapes into SL. After years of demanding “meshes in SL” (remember, the number of 3D designers world-wide is much smaller than these people would like to claim; popular products like Poser, for instance, sell about 150 thousand copies, and I always wondered how many of those are actually SL users…), this is the closest we’ll get (for now).

Linden Lab was very, very clever. It’s obvious that SL has to “look better” — less primmy, more organic, more professional content. A hundred metaverse development companies and their thousand RL companies demand it. But — they cannot destroy the prim economy by introducing meshes and a fully-blown “polygon count economy” — it’s too drastic a change. This system they’ve developed, although pretty unstandard (meshes get converted into a “colour map” which is a regular texture — not much unlike the RAW maps used for instant island terraforming, although with less detail), keeps the prim economy in place. Meshes are not part of the asset server; the system uses standard, 64×64 textures, which can be tied in into the streaming engine of LL, and stored in the inventory like a standard texture. It looks exactly like one, too, and it can be modified like a standard texture as well, and can be accessed (or changed) through LSL. The developers have even been talking about allowing streams of Flash video tobe used to dynamically generate the textures, and thus reshaping prims on-the-fly — a boon for artists, but something that all content creators will enjoy for things like furry attachments (say, muzzles opening, ears twitching, etc.).

It will also mean, to an extent, much richer content. Shapes that take a lot of prims — complex chairs, statues, etc. — will now only require a single one. Vehicles limited to the annoying 31 prim limit will now be able to have fantastic shapes with incredible realism. But a side effect is reducing dramatically the texture count on a scene — one of the worst causes of lag. A cube has 7 textures, which have to be stored somewhere, even if they’re not visible. A “sculptie” will only require two — one for the shape (64×64), and one to apply on top of it. Thus, on average, the texture count could, in theory, go down, although the realism of the scene will improve dramatically. And less textures to display means a scene rendered faster — thus, less lag. Imagine that 100-prim-hair can now go down to a handful of sculped prims :) … not to mention getting rid of the annoying invisiprims for shoes, which can now be perfectly sculpted and modelled to it your feet!

The first generation of sculpties will not be flexible (yet), but once we have the whole set of tools — sculpties, texture-change through Flash videos, an in-world tool to create them without requiring external tools, and flexible sculpties — the whole shape of SL will change (pun intended!) and will definitely go beyond our wildest dreams. No more “blocky” SL — it’ll have the same quality of other popular metaverses once more. LL has played catch-up and will probably win the bet with a very efficient solution.

What is the negative side of it? We were used for a long time to consider texturisers — specially the ones doing clothes and skins — as well as animators a special breed of content creators in SL, since they require deep knowledge of external tools to make their designs. Sure, you can do your own clothes through Apperance Mode, but it’ll never be the same. Sure, there are lots of “texture packs” in SL, many of those for free, and even more available through download from the Internet, but it’s clear that a special building/item/device requires specific and detailed texturing, which is only available to a professional designer.

Sculpties are almost impossible to do on a “trial-and-error” base. You’ll not only need a professional 3D modelling tool — and there are hundreds available on the market; although not all will be able to export to SL’s native “texture” format — but deep knowledge on how to use it. If you have never used a 3D modelling tool in your life, you won’t be able to learn it over a weekend, no matter how hard you try — 3D content creators in RL study for years and years until they are able to do what they want with those tools. So, the amateur is out of the picture. They’ll still be able to play with the regular tools, of course — but until LL introduces their own integrated sculpting editor, the amateur will have no choice but to buy other’s content. In a year or so, nobody will buy a 13-prim chair anymore, when you can get the same thing done with a single prim.

It’s a tough trade-off. Pretty much like animations — for two years now we have been “promised” an in-world animation editor, and “physical avatars”, ready for almost a year, but not yet integrated into the main code, are a way to make that simpler — you’ll have very high quality items from a handful of professional 3D modellers, and lots of junk (and copied material) from the rest of the residents. “Sculpty packs” will probably become as popular as texture packs are these days. But if you require a low-prim scene with an organic look, you’ll need a pro to create it, not a “talented amateur”. Thus, the rift between talented amateurs (6 millions) and professional artists (a few thousands) will continue to widen. Still, as the ratio of content producers to consumers reflects more and more RL (about 1:10) this might not be noticeable: not everyone wants to design their own hair, just pay L$250 for something that looks good, has a very low prim count, and low lag so that you’re not pestered by the club owners…

The amazing qArl Linden and Cube Linde

This is the hot dynamic duo of LL’s developer team. While the rest of the team plods along banging their heads against the wall with Havok 4.5, physical avatars, HTML-in-world (SL seems now to be the only platform that does not support such a simple thing — even kids on Sourceforge playing around with primitive 3D virtual world software put that at the top of their list, and it works), or even Mono, these guys are the Golden Angels of salvation. After sculpties, there is more to come — shadows on the ground, and shadows over other objects. Or changing the default sky texture. Or more detailed prim movement without breaking the shaky physical engine underneath. Also, with every new release, dynamic reflections get better and better (mirrors!), even if it’s so hidden under the Debug settings that nobody uses them. The current version is too good — people use far more “shiny” than is good for their health, and this gets mapped all over the place with incredible detail, but unfortunately you need a pretty decent computer to render it without lag. It’ll need some tweaking to adjust those algorithms to low-end computers, but I’m sure that SL’s new developers are eager to start seriously push the limits. Yes, they have a goal now: SL has to outdo Sony Home, which will be the reference in content quality from June onwards, and they’re struggling very hard to come to the same results. Alas, at some point, this will also mean updating the avatar’s mesh — and one wonders how they will make that transition without breaking the clothes market.

But I personally have very high hopes from this team. They’ve worked on the Hollywood CGI special effects market. They’re good. :)

Scalability

Cory and his team will very shortly present a white paper on what will be “SL 2.0″ on the server side. We, from this side of the screen, will not notice any real difference on how SL operates — the client will not be affected —except that suddenly IMs will work, teleports won’t fail, inventory won’t be lost, and our L$ account will always display the correct amount.

What’s the “big change”? It’s all happening under the hood. For some months now they have been dealing with the issue of having two co-location facilities — in Texas as well as California — and tracked down the many problems of a de-centralised grid. And the veredict is not good: SL does not scale well outside a single grid. A new, radical model needs to be developed from scratch, and that’s what Cory’s been doing.

The server simulators are self-sufficient — this was the part that Philip and Cory always got right. What they seemed to have ‘forgotten’ was that some things need to be centralised, and their approach so far doesn’t scale well beyond a single grid. So, they’re introducing a new model which will be presented to the public soon, and closely thereafter implemented. It’s not just “throwing hardware at the problem”, like many people think (“you’re rich, why don’t you buy faster machines or migrate to Oracle?”). A little-known secret told by any teacher of a computer science university degree is that “a good algorithm beats a faster machine every day”. This was why Google, with “old PCs running Linux”, was able to index the Web faster than any other competitor at their time. It’s not computing power, just designing far better algorithms.

And this is what they have been rethinking. It takes time to do so. Contrast it with the run-of-the-mill variety of current metaverse platforms. They don’t even care about running a multi-million-user platform across different hardware and de-centralised grids. Nobody has ever attempted it. World of Warcraft uses their own servers, there is no “need” to think about a way to “de-centralise” their server base to allow “others” to run WoW servers — only Blizzard will run them. OpenCroquet is used inside academic labs. Multiverse targets corporations and game companies that will control their user base — they’ll know where they are and how they’ll connect. They don’t care about a de-centralised grid, because nobody really needs one! Even IBM, Sun, Xerox, and possibly Microsoft, have designed “metaverses” for internal consumption only.

Second Life, however, has to literally grow beyond its (electronic) borders. They cannot “afford” to think on a centralised model, but imagine how the grid of 2010 will look like: a huge mainland probably run by Linden Lab in a dozen co-location facilities across the world; several “licensed” grids run by corporations and universities, all connected together; and individuals using their own sims running from home and allowing a handful of friends to connect. All this is “part of the Metaverse plan”. Only Linden Lab (and, well, the user community around SL) are thinking about this model. The Electric Sheep Company’s OpenMetaverse project or OpenGrid are examples of things that need to address world-wide interconnection across several grids, using the same SL protocol (the Sheep were very clever, they used the 30-year-old DNS system to distribute assets — way to go!).

Nobody else is thinking about this. Except for Linden Lab. And the time to announce this is quite right, because…

The “Open Letter”

… once more, residents are congregating to push their Luddite views upon Linden Lab. Called “Project Open Letter“, this has about 4000 signataries, and their purpose is quite clear: prevent, as quickly as possible, that Linden Lab introduces new features without fixing all outstanding bugs.

This is hardly a “new” iniciative. In my (almost) three years in Second Life, I have been very unfortunate to enter it on a stage where the major developments (SL 1.4, with the introduction of XML-RPC and animations) had just been introduced. Until June 2004, SL’s development cycle was quite clearly defined: features, features, more features, and another set of features. All were major changes when launched and had bigger impact than say, flexiprims a few months ago or sculpties right now. It was the only way to make sure the interest in SL raised and attracted the media’s attention; there were only 12,000 users in SL in June 2004.

Since that time, things have been quite more difficult for Linden Lab. They tended to focus on fixing a few bugs here and there until the resident population quieted down and started demanding for more features. Then, encouraged by the repeated pleas of “new shiny things”, LL entered a development spree and would launch a brand-new version with cool new things — only to find out that, deployed outside the lab, things would suddenly exhibit several anomalies (more on that later). Residents would rally at the Government Mansion and spit venom and claim Cory’s or Philip’s head. LL would try to launch a few more features on subsequent releases until the “wrath of the residents” would be unbearable. And then, it would mean about six months of painful debugging, dealing with scalability, introducing new servers, and so on. LL would “lose” half a year in their state-of-the-art technology, which looks more and more outdated every day a new game is released on the PlayStation 3, while it was top-notch in 2002 and still reasonable in 2004.

After this cycle, they would gather some encouragement, and try to introduce a whole bunch of features again. This would go on for a few months; starting in 2005, they even planned the technological releases for half a year or so, and quite often managed to keep their promises! But — alas — the Luddites with their pitchforks would raise again, and force LL to stop all development until all bugs got fixed. And LL would comply; another 6 months of waiting; another half a year wasted; another period where the competition in the 3D market would launch incredible games with beautifully rendered pictures, and the LL developers would just sigh and delete the code on their superfast new renderer, and get back to make sure flexihair still looks good when viewed from 20 m away.

This cycle has been repeated quite often. In late 2006/early 2007, LL had at least advanced towards a different model: assynchronous releases, ie. making the client separate from the server, without requiring people to log in to the (mostly empty) Beta Grid. This introduced the incredible advancements watched on the First Look series. I was personally overwhelmed with the elegant solution: keep the Luddites happy (there wasn’t a major release for almost half a year!), and let the rest of the world log in with the First Look viewer and get access to those fantastic new tools and features. It worked very well — until they had to merge the “new code” with the old — with unpredictable results. Alas, we have seen the pitchforks again — this time, however, they have the RL media backing them up. And this means that LL had to answer them.

Linden Lab will still give us voice (for those that really enjoy it) and sculpted prims, but I guess that will be all that is “noticeable” until the year ends. Most of the development efforts will now be targeted to scalability issues — a very worthwhile endeavour, of course, since reducing things like lag, instability, inventory loss, and login/teleport difficulties will go a long way with appeasing the Luddites — and minor tweaks here and there. Shadows, for instace, will probably come with another series of First Look viewers, as well as things like an in-world editor for sculpties, or client-side inventory storage. It might even possible preview the physical avatars on First Look, as well as HTML-on-a-prim — at least in a limited way that doesn’t require a new release. More deep changes, like Havok 4.5 or Mono, or a new mesh for avatars, will need to be postponed for 2008.

It’s very hard to explain to anyone without formal training in complex distributed network architectures why the impact of a slight change is almost impossible to foresee, when something that was thoroughly tested in a “lab” environment suddenly breaks apart when deployed on the “real” grid. The best analogy I can come with is to try to predict the weather for a century by observing just local changes — the weather is a chaotic system, and we don’t know the initial conditions, so it’s impossible to predict accurately anything beyond a week or so (the time the air masses take to cross our planet) using statistical methods. Still, this never prevented the environmentalists to predict a “Global Cooling/New Ice Age” in the 1970s and a “Global Warming” in the 1990s, and this also means that engineers and network specialists will still try to predict how a certain feature will impact the grid overall. It’s a task as hard as anything else. I usually tell my friends how a single line of code on a router on an insignificant country suddenly shut down by mistake one sixth of the traffic between Europe and the US for about fifteen minutes; it was impossible to predict the impact of that change (nowadays this would be harder to do!).

However, for people used to predictable environments (“if I urn up the heat, it gets warmer in the room”) and believing that science is all about predictability (and have not heard about quantum mechanics ;) ), they are quite unpleased with LL’s usual answer when something gets seriously wrong: “we did not know how strong the impact was going to be; and yes, this was thoroughly tested”. On complex environments like SL, some of the deployment is done instinctively because there is no scientific model that can be used to make an accurate prediction. People will work on “gut feeling” based on their past experience. LL has just a few years of experimenting with something quite new and unlike anything else in the world. In a decade or so, we’ll have Metaverse Engineers who will know all about it by reading textbook cases and knowing what to do (or what to avoid). But — this is all too new yet! Even the oldest network engineers at LL aren’t psychic to predict the behaviour of such a chaotic system. The good news, of course, is that they get better and better at it; they also get better at simulating the grid’s behaviour to give them more accurate tests; and, more important than that, they’re also working on creating a truly decentralised grid (Cory’s Pet Project) instead of a scalable, centralised one — which will definitely make things so much easier to deal with.

Imagine a day where LL will be able to selectively give you, on your sim, the features you wish :) If it crashes your sim, it’s your fault; but it won’t affect anyone else.

Ah well. Linden Lab could ignore the Luddites in 2004, they can’t do it in 2008 anymore, but they’re definitely working to please them. According to Cory, 72% of their developers are just doing bug fixing right now. Leaving 28% for innovation, research and development. For such a tiny company as LL, this is actually not bad.

Hopefully LL re-introduces the series of the First Look viewers again — letting us enjoy a better Second Life while the Luddites remain happy within Sony Home and its vastly superior static content :)