Note: As I promised in this update to my most recent blogpost, here is a very timely guest editorial by Galen, someone who was a very active content creator and programmer in Sansar. We agreed that, under the circumstances, it would be better to publish this guest editorial sooner rather than later.
Given my frankly codependent three-year relationship history with Sansar (and yes, codependent is the most apt word I would use to describe it), I think it best that I step back completely from writing about Sansar, or rejoining the official Sansar Discord. However, I will make my blog available to other writers like Galen who wish to write guest blogposts—editorial or otherwise—about Sansar. (I do reserve my rights as blog owner to veto any blogpost submissions I consider unsuitable.)
All the images used to illustrate this guest editorial were taken and submitted by Galen.
The future of Sansar
A Guest Editorial by Galen
Sansar is dead. Everyone else was afraid to say it. So I’ll say it. And there it is.
No. I’m not an insider. Yes. I’m speculating. Take this editorial as the opinion of someone who has been on the outside for a while now.
“Congratulations! You’ve been selected to be among the first to create social VR experiences with Sansar.” That’s how the July 6th 2017 email to me began. I spent a few thousand dollars ordering a fast gaming PC and an HTC Vive. I set up my account and logged in as soon as I got those delivered and configured. Within a couple days I had my first door opener script in the store as a freebie. I quickly built a little scripting empire with loads of freebies and eventually for-sale products. Not to mention doing countless hours of free mentoring and paid consulting. I met some of the most talented 3D artists ever. I couldn’t believe how much talent was already there before I even arrived for the closed beta.
I knew about the other social VR worlds emerging. I dabbled a little. I specifically chose Sansar. Why? Because Linden Lab. They got virtual worlding right with Second Life. Many people mistakenly believe SL was first and so nobody else could compete later. This could not be further from the truth. There were quite a few successful virtual worlds available and even popular before SL wiped them all away. Let’s not forget Active Worlds. They were 10 years ahead of SL. Yet people migrated from AW to SL in droves in the early years. Why? Simple answer is that SL was better. I had lots of reasons to believe Linden Lab would do social VR better because they had the experience and knew the formula.
But did they? Was I the only person who was bothered that the majority of the Sansar team seemingly had little experience creating or maintaining Second Life? More than a few I talked to had barely visited SL. They were starting fresh. They might as well have been a new company competing with SL without the benefit of all that experience. Which it seems is effectively what they were.
And now it seems they’ve fired most of the Sansar team. Few lessons learned on the way in. Few lessons learned on the way out. This is how it looks to me.
I spent a lot of time defending LL in their decision making regarding Sansar. By the time I left in late 2018 I was done defending them. I still sympathize with everyone though. I like Ebbe Altberg, LL’s CEO. I like all the LL staff I met. Many of whom helped me in my own projects. I certainly like and respect the many friendly and talented residents of Sansar. Most of the early adopters seem to have fled like I did. I respect everyone who poured tons of passion and money into Sansar like I did. We did what we could to make Sansar a success in hopes that Sansar would propel us on to something amazing. And enduring.
So what went wrong? Lots of people have expressed differing opinions. I can’t address all of them. I’ll examine a few of them broadly. But I want to focus on my own. I have a solid idea of what I think went wrong. And a solid idea of what I think LL should do going forward. They can profit from Sansar yet.
So why do I think Sansar is dead? As far as I can tell LL has eliminated its Sansar development team. As with most business ventures, if Sansar is not moving forward then Sansar is moving backward. This is the same move High Fidelity made shortly before shutting down their project. LL most likely will not shut down their servers anytime soon. It probably doesn’t cost them much to store all the experiences we created. As I write this there are only 16 publicly visible instances active with visitors. That shouldn’t cost very much in AWS fees. So there’s no real reason for LL to shut down its functional system. Nor thus to announce an actual end to Sansar.
But now Sansar is a zombie. It seems to be on autopilot. Maybe a few people left to maintain it. And probably a few other people to continue preparing for some planned official events. Again, this is my speculation.
If Sansar is not really dead then why stop development? Because it has failed to thrive. It’s that simple. LL put a lot of money and effort into creating and promoting Sansar. But it did not take off. Not like Second Life did. Not even close.
So why did the one succeed and not the other? Lots of explanations have been floated. Most of the ones I’ve heard revolve around technical deficiencies. The avatar isn’t very sophisticated or customizable. You can’t work together on building a scene. You need a beefy computer to run Sansar. And it has to be a PC. As a software engineer I can sympathize with how frustrating these sorts of complaints can be. But I don’t think they were ever the fundamental problem. Why not? Because SL would never have taken off by this same reasoning. The technical platform wasn’t really better than some of its competitors who had many years’ head start on SL. And it was very buggy in the early years, even after its explosion in popularity.
One thing SL had in those early days was a bold and innovative development team. Philip Rosedale led a freewheeling process that churned out big new features every week, it seemed. They were never finished. They were buggy. And they were cool. That had changed by late 2007. A new management team and process traded limber speed coding for cumbersome quality engineering. This wasn’t all bad. They managed to mostly end the grey goo attacks. And many other forms of griefing. They tightened up a lot of loose nuts and bolts. But they also brought the rapid pace of feature development to a near halt.
I think LL brought that same dreary spirit of sluggish development to its bold new experiment in Sansar. They had such a good starting point. But can anyone really say that they thought the slow drip of minor feature updates was anything like SL’s early days? Were we really better off with timid releases that had fewer bugs than we were with a gusher of crazy experiments that regularly crashed sims and clients in SL’s heyday? I know a lot of creators and users of Sansar complained about bugs when we found them. But I think this is a little shortsighted of us. They didn’t hinder SL’s growth at all. They shouldn’t have hindered Sansar’s either.
I’ve argued many times that commerce was the real driver of Second Life’s success. It’s one thing to offer someone a product they like. It’s another to offer them a product they can profit from. Whether with money, prestige, or any other thing. Second Life introduced the ability for creators to govern how their creations are used through permissioning. And they created a frictionless currency that eventually enabled creators to exchange their earned lindens for US dollars and some other real-world currencies. In my opinion nothing was more important than this set of innovations.
Yes Sansar eventually had these features. Kinda. Sorta. They even introduced an innovative mechanism for creators to earn ongoing royalties as downstream creators sold their incorporated components. But in my opinion they simply failed in this critical area. They were slow to introduce the sansar dollar currency. They took way too long making it so you could directly pay people in sansars. And as far as I can tell they still have not made it possible for users to pay scripted in-world machines for services. Like paying for an hour in an amusement park. Or paying a tip jar at a concert that splits revenue with the house. And they haven’t enabled scripts to pay those machines or pay into users’ accounts directly. Like auto-payment of recurring fees like rent. Or wages for employees.
I don’t think I can really blame Linden Lab for this. I don’t think LL would be allowed to create the linden currency and its market in today’s regulatory environment. They shut down the alternative currency exchanges in part to comply with stricter banking regulations that emerged ostensibly to combat money laundering and other ills. This is probably the main reason LL was so slow in introducing the Sansar dollar and in making it easy to use. This is almost certainly a significant factor influencing other virtual worlds. It may well be why High Fidelity opted for a quasi-independent cryptocurrency. Not strictly owning the transaction ledger probably exempted them from SEC reporting requirements. This could be what’s stalling VRChat’s virtual economy too. In this sense Second Life is grandfathered into something that can’t be easily built from scratch today.
It doesn’t help that the Sansar dollar is not at all frictionless as a currency. They charge a lot to buy Sansar dollars. They charge a lot to sell Sansar dollars. They charge a lot to buy things with Sansar dollars. They charged a lot to give the gift of Sansar dollars. (It seems they eliminated this fee eventually.) I spent a lot of time defending LL’s need to profit from their platform. And I understand why it would be hard to introduce or increase fees later in time. But I think it is impossible to overstate how important the nearly frictionless (and fee-less) use of lindens is to SL’s ongoing success. Hundreds of millions of US dollars in perhaps billions of annual transactions attests to it.
CORRECTION: It seems my information is old. Apparently LL no longer takes a cut from from gifts. It seems they only now charge for store sales, cashing out, and for transferring from your USD balance to PayPal.
The inability of people to easily use their lindens to buy things in Sansar is arguably one of the other opportunities LL missed. I’m sure plenty of the creators in SL who dabbled in Sansar would have gladly spent some of their SL-earned capital in Sansar if they could. But let’s expand the scope of this. It makes way more sense when you realize that many of those same creators really wanted an easy way to bring their creations into Sansar. And many regular users wanted to port their inventories. I’m not going to argue that this would have been easy to implement. In fact I argued early on that this was a bad idea for many reasons. Sansar was its own new thing. It deserved a clean break from the downsides of SL’s old technology. And intellectual property owners in SL deserve a say in whether their goods can be ported anywhere else.
But I think it’s time to reconsider this idea. If Sansar is dead then Linden Lab needs to decide what to do with its development budget going forward. Exactly who didn’t come to Sansar? Why, Second Lifers. Who were looking for Second Life 2.0. Which they were told Sansar was not. So it’s obvious what they want. They want SL 2.0. They almost got it in High Fidelity. But LL is uniquely positioned to do this the right way. It’s a compromise way. Something between the clean break of Sansar and the tepid development path SL has been on for over a decade now.
I gave the following proposal for what to do if Sansar failed in another guest editorial in 2018. Let’s explore that proposal a little more here.
The heart of my proposal is to create a new technology platform and brand it as Second Life 2. The critical thing to do differently from the Sansar project is to make it so it is at least somewhat compatible with SL. The most crucial thing to share is users’ identities. Followed by their bank accounts. No separate accounts. All the same as now. Everything else is negotiable. This should not be.
Next up would be the grid. SL2 would exist within the same space as SL1, the current grid. I don’t necessarily mean that an SL2 sim would have to be exactly 256 meters squared like SL1’s sims. One option would be for them to be some multiple of that size. For example a 1024 x 1024 SL2 sim would occupy 4 SL1 sim slots. The SL1 grid would need some upgrades to be compatible and to make it relatively easy for users to cross from one grid to the other. And maybe a better option is just to punt by creating some sort of grid-to-grid teleport system. The SL2 grid can be like a parallel world where you simply cannot “see” across the divide. Or maybe only through specially designed portal windows/doors. That sort of wizardry can be created down the road and wouldn’t be required on day one. The only critical requirement is that a user can travel fairly easily between SL1 and SL2 sims.
What about avatars? Do they need to be the same? I would argue that they shouldn’t be. The avatar that Sansar had introduced wasn’t all that bad. It had quite a few solid innovations. And it was arguably easier to dress up than SL’s arcane mess is today. Maybe this would be worth starting over with lessons learned from SL and Sansar too. Maybe some sort of hybrid that would allow skins and some other avatar assets from SL1 to be ported to SL2. Or not. I do believe that SLers will tolerate the fact that they have to create and outfit new avatars in the new grid. I think this initial irritation will be far from a deal-breaker for them. In fact it will likely spur a whole new fashion race to cater to SL2 without killing the SL1 fashion industry. Spend some time developing the basis for this. Don’t hack this part.
One dubious design choice in SL is that there is effectively no limit to how computationally expensive an avatar can get. I proposed in Sansar to introduce a mesh complexity budget to allow users to have as many mesh clothing attachments as they wish by balancing how rich each attachment is against what else they wish to wear. I recommend something similar for SL2 avatars. If they go over that budget they start paying fees for the extra weight. And SL2 sim owners should be able to limit entry of avatars based on their complexity.
What about VR and all the visual glitz of Sansar? Yes! Definitely. Do it. I think almost everyone agreed that Sansar looked great. Just don’t do it at the expense of the live editing experience. I know there are lighting and other optimizations that come from compiling a scene in Sansar. There is an easy compromise though. Selectively bypass or even disable those optimizations during building. Do background compilation as the scene gets updated. If SL users can understand progressive loading of sims then they can understand progressive baking of lighting and sound optimizations. And that’s another thing. Let’s accept that users want to start interacting with scenes as soon as possible. Progressive loading may have downsides for some use cases. But whole-scene loading has way more downsides for many casual uses. This is something you can have both ways though. It should not be hard to develop a progressive loading scheme that’s based on distance to the viewer. Things nearby load first. Then things slightly farther away. And so on. They’ll likely feel more like it’s loading faster. And then you could also allow SL2 sim owners to decide which of the two modes they prefer to require visitors to enter via. SL does have some prioritization to its loading order but it’s not strictly distance-from-you oriented.
One of Sansar’s most elegant features is its on-demand loading of scenes on the server side. I recommend that SL2 sims follow this model. Allow sim owners to decide whether to pay a premium for always-on service if they wish. But otherwise allow empty sims to auto-unload after a while of disuse. Make it owner-configurable how long that timeout period is. And have those sim owners pay only for active time.
One interesting possibility for SL2 sims would involve a radically different notion of what a parcel is. Let’s say you have land leased in large square units like in SL1. But let’s say when you parcel that off into smaller chunks you are really creating separate sims. The equivalent of whole scenes in Sansar. Each parcel-sim would run on its own processor just like a scene in Sansar. But you’d still be able to see that parcel as part of a larger property. The owner of that larger property could charge the parcel owner rent for the privilege of being included in their valuable neighborhood.
One problem with sims that don’t stay online 24/7 comes when you are in one sim and the neighboring one is offline. What do you see in that case? Although it’s not perfect here is one proposal. Every sim gets stored as a model already. In SL1 sims they are dynamic. In SL2 they could be static models. When you are in one SL2 sim you could look out far into the distance and see potentially hundreds of scenes on sims (parcels) that are currently offline. How? By having your client access the static models of those sims instead of trying to talk to active neighboring sims the way SL works now. Those sims’ models can be stored in different level of detail (LOD) versions to suit their apparent size to the viewer.
This is all heady stuff. Lots of coding work to do some of the above. One option is to just go with separate scenes like in Sansar for the first release of the SL2 grid and evolve more integrated approaches over time. Again, SLers will tolerate this just fine.
Practically speaking, the SL2 model is going to require a hybrid SL client that contains both SL1 and SL2 codebases. The Firestorm client has managed to keep pace with and largely outcompete the main client from LL. So I imagine that team would do just fine in collaboration as the SL core team develops the early prototype client.
Should SL2 require a high end gaming machine? Or be dumbed down to work well with older machines? I don’t think it has to be dumbed down. But probably the best answer is to let users decide how far they want to go with their computing hardware. The budget-conscious user who just wants to hang out with friends could ramp down the graphics settings to suit their old machines. And power users could ramp the settings all the way up for their photo shoots and VR applications. Over time most users will gradually adopt more performant machines to enjoy all of SL at its best.
So why bother doing this SL2 stuff in the first place? The answer is simple. The goal should be to phase out SL1 over time. SL1 and SL2 would likely coexist for quite a few years. The SL2 grid would start out as a curiosity to many. And a promising place to try new artistic and business endeavors. Especially if the SL2 grid is truly VR capable. Think of these parallel grids as analogous to how SL has both voice and text-chat modes. Many users exclusively favor one or the other. Some use both interchangeably. No doubt the same will be true for the parallel SL1 and SL2 grids. But if most of the new development is focused in SL2 then most people will gradually spend most of their time there. Most won’t even notice the gradual change in their behaviors.
But again, why bother doing this? Why not just keep upgrading SL as it is today? This editorial is already too long so I won’t go into detail. I’ll just say that SL is held back by some of its early design decisions. Most of them made lots of sense in the early days. And now they weigh SL down. The bottom line of those choices is that they keep SL somewhat expensive for many users. They require LL to maintain an overly large hardware investment. They limit designers’ choices. They keep SL looking a bit cartoonish. They prevent many realtime gaming dynamics. They prevent VR adoption. They make it difficult for larger corporations to form to bring ever more grand creations into existence. In short the technical limits are holding SL back.
I and some others have argued that LL made a mistake by not allowing Sansar to have adult content and activities. This would be another benefit of building a parallel SL2 grid. LL would not have to introduce different rules for both grids. The existing culture of SL should be allowed to flourish in the same way in SL2 as it comes online. I know that stuff may scare away some media companies with deep pockets and an aversion to anything more risque than Toy Story. But it’s also apparent that if LL actually banned adult content from SL then SL would immediately vaporize. It’s an important part of SL’s success.
And more generally it is Second Life’s residents who have made SL such a success. Most of them have been unhappy with the overall feeling that Sansar was a waste at best and a betrayal at worst. I think they are shortsighted in this. But there it is. Their opinions matter. I’m convinced that developing a Second Life 2 grid as a parallel to the current grid and allowing users to be themselves in both is a recipe for success. And not just in keeping SL afloat in the stagnant growth pattern it seems to have held for over a decade. This would very likely make SL flourish anew. This would be a real success for social VR. This would be what brings many of the SL refugees that populate VRChat back home.
I know I’m an outsider at this point. But I haven’t completely abandoned my wish to see Sansar succeed. I was actually hoping to create a virtual presence for my science fiction stories in Sansar. I was hoping there might be more opportunity for me to return someday to do so much more. But that’s probably not going to happen now.
There’s so much more I’d like to say. But this is a start for me. And hopefully it encourages a bigger discussion. I think it’s time to admit that Sansar is dead. But Second Life is alive and well. And ready for an SL2 project. One that respects the current investment hundreds of thousands of people still have in SL today. A couple years ago I didn’t think I’d be the one to say this. But I’ve changed my mind.