Search for “Fast Easy Fun Second Life” in Google, and you get two million results. By now, it’s quite hard to believe that anyone vaguely familiar with Second Life hasn’t heard the new motto, launched by Philip Linden during his Town Hall meeting last Friday (July 30, 2010).
If you missed all the fun, you can still watch the archived video stream from Treet.TV. If, like me, you’re too lazy to spend an hour watching videos while you should be doing something more productive, you might be more comfortable reading the full transcript. If you just want the highlights, you can get them from SLOG, Daniel Voyager, or Adric Antfarm. For thorough analysis, well… you pretty much can get them from any blog on the SLogosphere I guess; there’s hardly anybody who did not write on the subject.
Overall, what I can say is that this was, strangely enough, the most informative Town Hall meeting ever held by Philip (co-starring with BK Linden, and moderated by Wallace Linden). Usually, Philip is vague and avoids the tough questions. This Friday, however, he was way more specific, at least on most points. If that’s just trying to deal with the insanely high expectations that so many have put on his shoulders, or really a change in the Lab’s attitude towards the SL residents, I don’t know. But it was rather refreshing to listen to a different Philip!
Let me be honest and state, from the beginning, that I’m not part of the hyper-enthusiastic, mega-optimistic crowd that believes that all problems in SL will be over by the end of the year. I’m carefully optimistic. My major issue is that I really do not understand what really happened at Linden Lab. There have been so many conjectures and wild theories floating around that it’s pretty much impossible to figure out which one is right. Most of them explain a few of the issues but quickly lead to contradictions.
Recent history and the contradictions
For instance, let’s assume that M Linden, two years ago, defined a strategy for two years, and saw what the major problems were: users register like crazy (we still get 10,000 new ones every day, and this number has been consistent since 2006 when LL introduced free basic accounts; we’ve reached 20 million registered accounts by the time Philip was on stage. No, they’re not all alts) but they don’t remain in-world — the retention rate is so low that the number of simultaneous logins is diminishing over time. Then the economy was not growing exponentially, and that meant that LL needed to figure out a new source of revenue. A lot of projects were semi-abandoned or incomplete. Most users continue to come from outside the US. Education and business were growing in number, but finding a lot of problems to install SL behind their firewalls.
M Linden successfully assembled a strategy to deal with all the above. We don’t know what he has promised the board, but it was clear that his focus was on SL Enterprise (a new market for LL, with a completely new source of revenues), getting a bigger slice from the content creation business via SL Marketplace (since the SL economy is worth 0.6 US$ billions but LL doesn’t get much from that), and making sure newbies wouldn’t go away so quickly, by giving them an “easier” browser and revamping the first hour experience.
All excellent goals! (from LL’s perspective, of course)
Now, we don’t really know what happened. LL hired like crazy in those two years, mostly to create its Enterprise division, and expanding and enhancing Marketing & PR (the new Amsterdam offices, for instance, allegedly never had any developers — just marketing, PR, and sales). They hired Big Spaceship to redevelop the SL websites’ look & feel, and to provide input on the upcoming SL 2 Viewer. Since LL was and remains a rather profitable company, with plenty of available cash, they could afford all of that, and I’m sure that LL’s Board was fine with the strategy.
Then two things happened. First, I have no idea how many SL Enterprise boxes were sold, but I’m quite sure that the answer is “not many”. Education, for instance, is more eager than ever to enter Second Life, but as soon as they hear about the prices, they move quickly out to OpenSim (or, sadly, Unity3D, just because it has meshes, a Web-based viewer, and runs on the iPhone and Android too. I say “sadly” because OpenSim, for me, is just the younger brother of Second Life, but not really a different platform). Corporations have a mixed attitude. The ones more willing to use virtual worlds are the ones with the less money to invest. Over these two years, being in the business myself, I have seen multi-billion corporations just willing to shell out a handful of US$ to do some work on virtual worlds — and soon figuring out that there was no way they could do it in SL. Universities, more than ever, are engaged on very-long-term projects — after the “experimenting” phase in the early days (when Pathfinder did the hand-holding for them), they’re now entering the mature research stage, which means that research using SL will now unfold in the next decade or so, as projects start to take 2, 3, 5 or more years, like all other long-term academic projects. But… they might, at some stage, expend all their little funding and just switch to ReactionGrid or any other cheap OpenSim grid operator.
And, of course, nobody could predict how badly the overall reaction to the SL 2.0 Viewer was. I have to say I’ve been quite disappointed with the reaction. It was perhaps the first time in my six years in SL that I heard so many praises of the 1.23 user interface, when for the rest of my (virtual) life, I just heard the worst possible comments about it. It’s true that people resist change. But I never felt such a strong resistance before! I imagine that it was at the level that LL experienced when moving from a “prim tax” economy to a “land tier” economy, with the difference that we are now millions of residents and not hundreds.
Nevertheless, SL 2.0 also failed to capture new residents (who wouldn’t be biased by an ugly, old interface). Clearly there is absolutely no difference in terms of the retention rate. It’s not just the UI. Something else is wrong.
At this point, something changed at LL. Some speculate that the Board had given M a free hand (after all, Philip went to work on his new company, continuing to develop his beloved Love Machine) and just saw at the figures earlier this year and entered into a panic — the huge investment in everything was not paying off, and, while LL remained lucrative, it certainly was becoming far more dependent on the quirky fluctuations of the income from residential users, while the company was supposed to be 100% engaged on the corporate and academic markets. M Linden might have panicked and just fired a third of the employees. Even this is incredibly strange — the firings seem to be arbitrary, random. Team leaders of the most crucial elements in SL’s technology have been kicked out, while corporate marketeers remain on board. People engaged in addressing the community to hope to make the first hour experience better were the first to leave. And of course, some had already left LL, like Robin and Pathfinder; neither left to immediately get a new position on a competing business, but remained unemployed for a while, so we have to imagine that they were invited to leave.
And then, the final stroke: M leaves, too.
Hmm. Looking at the scenario, a lot doesn’t make sense. Can we seriously believe that the Board didn’t look at M’s numbers for two years? No. Is there an explanation for why certain key Linden employees were kicked out, crippling LL’s development and community management abilities very seriously? No. These things simply don’t make sense to us outsiders.
The Return of the Prodigal
So now we get Philip back as “interim CEO”. One of my first predictions completely failed to materialise: that Philip would introduce LL’s CFO & COO, Bob Komin, better known as BK Linden (now with an avatar picture on the page with his job description!), as the future CEO. Corporate-savvy analysts tell me that the major downsizing had to pass through Bob, so he was part of the process.
Probably Philip (correctly) feared a mass panic if he dropped the news of having selected a new CEO on top of us. It’s well known that Bob, after six months at LL, never created an avatar and never felt the need to do so. He’s rumoured to be ruthless and efficient as CFO & COO — but also completely clueless about Second Life, the virtual world (while he might be a financial wizard about Second Life, the product that Linden Lab sells. Both are usually confused, but they are hardly the same thing). Philip described him with similar words that he used to describe M Linden two years ago — which did not escape our attention. But the most interesting bit was that apparently Bob called Philip back from the freezer and put him in the limelight again; so there is a strategy here, one that probably started to make sense when Philip did his surprising appearance at SL7B. Somehow, the thinking heads at Linden Lab (correctly) assumed that they needed a figurehead to capture some optimism regarding the future of LL, and, after so many people having been kicked out, there was not much of a choice: Philip is, and always was, the resident visionary, and pretty much irreplaceable in that role. He admitted on the Town Hall meeting that “[Linden Lab has] not begun the process of selecting the long term CEO.” So maybe Bob’s not our uncle… I mean, not our CEO.
Here is a summary of the new strategic orientation for Linden Lab:
Fast — Easy — Fun
Well, this replaces the old “Your World, Your Imagination” motto, and is probably a bit more simpler to understand. Philip promises that it will be the goal, short-term and possibly long-term, for Second Life. “Fast” will mean getting rid of lag (more on that later) but also faster development/deployment cycles of bug fixes; “Easy” will probably mean rethinking the user interface on the next iteration of the SL Viewer; “Fun”, well, interpret it as you wish — I might see in this the notion of SL as a residential product with a focus on entertainment (and all forms of content creation are entertainment). But it might just be Philip spreading a meme without clear notion on what he means 🙂
Yay, now we have this lovely meme! What does LL do with it? What are its priorities? Philip explains:
- back to basics
- winning back the lead
- all about growing virtual content economy
This is consistent with his previous blog posts and messages.
“Back to basics” is how Philip describes the technological challenge of drastically reducing crash rates and lag. This is way more complex than it sounds. It’s not merely changing the internal focus from “features” to “bug fixing”, because LL has been doing that since 2006 — also thanks to Philip (and probably against the will of Cory). The two years before M Linden came on board were all focused on stability and performance; we got almost no features during that period, except Windlight, which was not even internally developed, but just a company they bought. We got very lucky that we managed to get flexies and later sculpties at all. During the M Linden days, we started to see new features creeping back again: from working media-on-a-prim to multiple clothing layers, to new alpha textures on skins, all introduced on SL 2.0. Ironically, Philip’s work at LL during that time was to launch Snowglobe as LL’s open source, user-contributed viewer, and to create a mechanism where textures would finally be downloaded with regular HTTP calls instead of streaming them (which burdens the poor sim servers). This has been a long development, which was originally only available for the map textures, and allegedly, this August, with the release of server 1.42, will finally be expanded for all textures. Philip himself says that 60% of their traffic is just texture streaming. If all textures are pushed into Amazon’s cloud (like the map textures), it has to have a huge impact on the overall performance. If Philip also allows the developers to tweak the viewer so that textures retrieved via HTTP get cached better locally (right now, you’re limited to merely 1 GByte of local cache — which is way too little!), this would even be better. In any case, the good news is that textures might also be cached by a local Web proxy server, which should be awesome for university campuses and corporate networks — the more users connect from the same address, the more textures will be cached locally at the proxy server, and the lower the burden on LL’s own grid network — improving performance more and more.
Ironically, all this works today on OpenSim already — using LL’s standard viewers. It’s just LL’s server software that has been delayed (deployment of 1.42 is scheduled for August 18, but Philip mentions “this week”). And that’s probably one of the things that Philip means with the next point: winning back the lead.
Now this again opens up a lot of possibilities. Clearly Philip wants that the future SL 2.X viewer becomes the standard, and not Emerald or any other 1.23 derivative; and I’m also sure that the nifty advanced improvements of OpenSim (like regions of any size; the ability to do on-demand load performance by splitting avatars on a region between several CPUs and keeping all in sync; native megaprim support; pushing the IM/Group Chat outside the sim server; and much more, like the insane amount of caching mechanisms that OpenSim supports these days) have made Philip rethink LL’s priorities. If that’s all true, it means that at least someone at LL is looking beyond their navels, which is a very good thing!
Fighting crashes, fighting lag
Well, Philip at least was honest saying that both menaces — crashing and lagging — have an insane amount of sources, and they’re at least tackling a lot of them. Getting texture streaming and IM/Group Chat out of the sims should improve things dramatically — sims will still have to worry about object geometry, tracking down avatars, and parcel lists, but it’s definitely good to see textures & IM out of them. I found a sentence intriguing: “lag is also problems with many people in same area / you experience that in simulator and in client / we are attacking that”. Now I’d love to know what that means 🙂 Avatars are one of the major sources of lag, but it’s really hard to deal with that, since SL avatars are insanely detailed (think of all the attachments they have and the huge amount of memory all their scripts take to run!). So it’s anyone’s wild guess on how they plan to diminish avatar lag. I have absolutely no clue! Babbage was working on the scripting side of things, limiting the amount of memory avatars could use for their scripts, but since he left, the status of this project is unknown. And multi-prim attachments might be replaced in the future by far more manageable meshes, but Philip, later on, says that meshes are not going to happen soon. So… what’s LL’s magic to deal with avatar lag?
Shorter release cycles…?
For those of you who enjoyed the days before the Golden Age of 2006/7, SL was being developed at an insane pace. Every other week the grid would go down on Wednesday, and we got a new sim software loaded on all the servers, and a brand new (mandatory) viewer to download. Usually that meant a lot of new features (although I believe that up to 1.4 this was even more drastic) and bug fixes — and a lot of new bugs, too! Currently, LL seems to launch a new server version every other month (on average), and viewers are… unpredictably released. By contrast, third-party viewers get frequent releases, usually weekly at least, sometimes even more often. Philip wants to catch up and do that for LL’s own software as well.
There might be a problem here. Philip also says that he wants all code to be more thoroughly tested. So these two goals work against each other: shorter release cycles, but with more code testing? Currently, a feature or a bug fix takes around 6-18 months to get implemented (except for emergency measures to deal with security issues). A lot is on JIRA waiting for far longer than that — often with submitted code to fix a viewer issue. Let’s face it: LL is a very, very slow developer — because the code is insanely complex — and I personally don’t see how Philip is going to change this to keep up with his promise. Just look at the number of changes from one sim server version to the next. 1-2 months elapse, and there are at most a handful of fixes or features. And Philip assumes that these have not even been properly tested…
It’s about the economy… or not?
The next item on Philip’s agenda left me a bit confused. He started by saying that the in-world economy was indeed a priority, and what makes SL so special compared with other virtual worlds — none come even close to the US$ 0.6 billion of annual transactions! But then he also says:
The concept of the biggest economy.
The virtual economy, not that SL is only about people making money.
It’s not all about selling.
Fundamentally, SL is a system where we create virtual content and experience it together.
Success of world builders is the key driver.
it’s beyond the software and tech
We need you to build world and create content.
it’s hard to measure the success
I’m not sure how to interpret this, except perhaps that Philip just means that it’s less important to become rich in SL than to actually create compelling content. But he goes on saying that they will look closely at some metrics: “We will set up many needles around that. / To watch the success of virtual content creators.” Ok so far. A concrete measure for that: fixing Search. He goes on explaining that he’s aware that Search doesn’t work, he has increased the size of the team working on the fixes, and he’s aware that a broken Search means less sales. All this is a very good sign: he’s aligned with content creators’ needs and has been listening to the thousands of complains about the broken Search.
He also points out a lot of things related to the economy. It’s not just the Search that has to be simpler, it’s the whole shopping experience. If you find something in-world, you should be able to try it before you buy it (the mind boggles at how that is going to be implemented!). Delivery of boxes with content, which have been the only way to sell content grouped together for years (well, that and selling folders with inventory), makes things complicated for casual buyers and new users, as you have to rez the boxes first to extract the content. Somehow this is going to be fixed too. Somehow it will be related to the Marketplace (which is not going away, but it won’t replace in-world shops either). Somehow it’s all related to this economy that Philip calls “bigger than shops and XStreetSL”.
The first hour experience
Philip has not forgotten M Linden’s goal to make the first hour experience much better. However, he adds a twist: “it’s not about learning the interface / it’s about getting them to people and experiences beyond welcome / a link to Destination Guide or search on the web should be the starting point for new users”. Now we’re talking! Somehow, new users have to come in-world and immediately get immersed into places they like and talk to people with whom they will feel an immediate affinity. How exactly this will be done is anyone’s guess; I already made my own suggestions 🙂
Bye-bye, SL Enterprise?
Well, not exactly. The difference is that there will not be a special team just developing things for the enterprise market, but all efforts will be to improve the experience to all users, not just a few of them, with “special corporate projects”. I sensed here some criticism aimed to the previous CEO: resources were pooled to develop SL Enterprise, which could have been used otherwise in ways that would benefit us all.
An example that speaks to a broad point, SL Enterprise, we don’t know who is using SL.
We think it is used by educators, those casual users, by people at work.
What we do with SLE is we’re working toward focus on core user experience that affects all users.
Educators and business are affected by lag, we have to hit those first
We’re stopping work on things that don’t impact all users.
In SLE, we’re not trying to move away from use at work, but we aren’t going to work on deploy behind fire wall.
We will work to support them on the main grid.
So this is a small twist; on his latest blog, Philip promised that he would not be abandoning business and educators, like many (yours truly included!) incorrectly assumed. What they are going to do is not to differentiate between users: all projects will either benefit all types of users (residential, corporate, educational), or they won’t even be started. In practice this means that the focus is back on Second Life the virtual world, and not necessarily Second Life the nice VW platform that can be used by others for their special requirements.
Well, this was a bit unexpected and perhaps the most frustrating announcement for me personally:
Applying filter of what we are thinking about now, we want to implement it so it has impact, but not negative on frame rate and rez time.
We are working with beta users to be sure it doesn’t slow down experience.
So what does this mean? Usually, this kind of wording is employed when LL basically drops a project. It could be a result of thousands of amateur content creators pleading with Philip not to implement meshes because it would ruin their business — and those thousands definitely represent a considerable amount of the in-world economy. Nevertheless, LL really cannot continue to delay meshes further. The risk of losing all educators is very real — academic institutions have access to zillions of free and open source meshes already, and want to deploy them in SL, because they cannot afford either to learn how to build with prims, or to hire people to do the buildings for them. A friend of mine told me that meshes would be implemented in less than a year on OpenSim. If that happens, SL, as a platform for research and education, will be dead by 2011.
Perhaps Philip doesn’t worry. Perhaps he knows that educators, by themselves, contribute little to the virtual world economy. According to him, anything that is not aligned with the core vision of Fast/Easy/Fun combined with Back To Basics/Keep the Lead/Grow the VW Economy is to be dropped.
Some final comments and questions from the residents
After the comic relief moment regarding the upcoming makeover for his avatar (with a nice immersionist statement — “This is my alter ego”), Philip pretty much put the focus on “his” virtual world as “the” social environment of the 21st century. It was quite interesting to see how he pretty much dismissed all the speculation of a “dumbed down” SL that would be embedded in Facebook, or similar integrations with other social networking tools. I read in his words that he finds all those social networking tools seriously lacking in the rich experience that, right now, only SL provides — so he wants that the focus remains on making SL a better social networking tool, not emulate — or integrate! — others. This is fascinating to a certain degree and runs contrary to what is happening all over the Web, where developers are eagerly developing more and more integration between all the major social networking tools. Perhaps Philip is ambitious enough to state the claim that all the rest of the world is simply wrong — and stuck on 2D metaphors — while Linden Lab is pushing Second Life far ahead in the future.
Remember that “social networking” throughout the 1980s and until the mid-1990s meant IRC and Usenet — and MUDs. We switched from that to the Web. Philip seems to be saying that the Web is “old news” and that Second Life is a much better social networking environment. Well, if that’s what he means, I really have to agree, and I’m sure that the million regular users of SL will agree with Philip as well — even if 2 billion Internet users will not.
There was also a final emphasis on more two-way communication with the residents, either using the Town Hall (with this new “lottery” method) or other models. This will include meetings and messages from project managers and team managers, which is quite a welcome addition, since at this moment, nobody really knows (outside LL, that is) who is in charge of what projects, or what their status is. Needless to say, this is the kind of thing that everybody will be pleased to hear — and we will all sincerely hope that Philip implements it as part of LL’s corporate culture. It’s very welcome news!
Since there have been so many rumours of “dumbed-down versions” of SL (like a Web-enabled SL viewer, or a mobile viewer), Philip adamantly refused to accept that SL would be “limited” in any way. In his words, “We won’t change nature of product or depth of experience in SL.” We heard that there is a team (Philip honestly admitted they were just two developers) working on a Web-based version of SL, but that this will only be released when it allows the full immersion in SL like the main viewers do. And he repeats: “we won’t subtract capabilities but add speed and elegance to what is there”. But Philip recognises that the SL 2.X viewer neither works for builders and other power users, nor does it work for brand new users. So, he says, “we are thinking about how to iterate week to week to add capabilities for power builders and separate them from capabilities needed by first time user who needs to get clothes on and walk around”. How exactly that is going to be implemented so that it doesn’t violate the notion of “not subtracting capabilities” is unknown.
Apparently the open source effort will start to be directed (another novelty!). This seems to indicate that LL will set a list of key features/bug fixes they expect their viewer to get, and ask the community to contribute code for those areas. This is an intriguing change of approach to what LL has been doing so far (which is, well, mostly ignoring contributions).
Linden Lab has certainly turned yet another chapter in its fascinating history. The projects nearing completion — for instance, the launch of the SL Marketplace which replaces XStreetSL — are still the touch of M Linden, but some (like textures via HTTP) surprisingly come back from the days Philip was still active as an engineer at the Lab. Except for meshes (release unknown) and fixing script lag (Babbage’s project, status unknown), pretty much everything that M Linden wanted to do has been implemented; with a reduced employee count (they’re still 250, though), everything new that happens from now on will be Philip’s work, once again, and it’ll be fast, easy, fun. Unlike on previous Town Hall meetings, at least on this one we a got much clearer vision. We have a new motto, which is actually the way everything at the Lab will be evaluated from now on. We have new channels of communication. We have a proto-roadmap and a focus on concrete issues, and while the details elude us at the moment, we got the promise that the project and team managers would explain what they’re working on. The open source initiative on Snowglobe will get a revamp and a new direction. And Philip continues to underline the importance of this virtual world, its environment, its economy (which goes beyond money and sales but is all about content creation), its inhabitants — residential users which make it work. So even if naturally a lot of details are vague, I think that Philip presented a very good picture that he still knows what SL is about and has both feet planted on good solid ground — while his dreamy head is already looking into a future that we can’t yet imagine what it will be. Of course I shall curb my optimism and adopt a “wait and see” attitude; but at least I think I never saw so clearly what Philip’s intentions are, and it seems quite reasonable to assume that LL will once more become the tool that turns Second Life into Philip’s vision — hopefully one that is aligned with what residents actually think that SL is for.
We really can’t ask for more.
As is often the case, this is one of my articles that actually took more time in preparation than in writing. I had a picture from the Town Hall meeting being streamed via Treet.TV to a small audience at Colonia Nova (CDS), but at some point it was clear that the motto “Fast, Easy, Fun” had to be somewhere in a picture with this article. Well, obviously this called for yet another vanity picture of myself with the motto on a T-shirt, just like Hamlet did with Callie Cline’s photo on NWN. I went to the SL Marketplace to look for a shirt, but didn’t like any of them. Besides, doing T-shirts is easy. So easy, in fact, that I thought I should do something a bit more fun than a plain T-shirt. Thus I googled for some clothing templates and found one I liked for an asymmetrical halter top on Avatars3D.com. Back to Photoshop to do some simple changes, and an upload… just to find out that the template produced rather poorly stitched tops. Hrmpf. Well, my first reaction was that nobody would know, if I just got the right camera angle… but my second reaction was that if I did something a bit more polished I could perhaps offer it for sale on the SL Marketplace. I opted for wasting a few more hours clone-stamping, blurring, blotting, and skewing blotches of pixels until the preview feature of the import texture box showed me everything was nice. Then I uploaded the texture and… eeek! I forgot, this preview feature is still buggy, and the real avatar mesh is somehow different from what the preview shows… Since this meant uploading a lot of textures until I got it right, I gave up and logged in to ReactionGrid, where texture uploading is for free, and after several attempts, finally got a passable texture. Back to SL with the 2.1 Viewer, armed with Ana Lutetia‘s poses and Windlight settings, it was time to take a few pictures, go back to Photoshop (with some free FilterForge plugins) and create an ad for SL Marketplace. Whew! Fast? Nope! Easy? Not at all! Fun? Oh yes, lots!!
It’s really when doing all this work that one really starts to appreciate the incredible amount of time all these fashion designers take to create their fabulous content — and how cheap it is, compared to the time they have spent producing it!