Guest Editorial by Galen: Taking a Break from Sansar

It’s been a fun ride so far, but I’m ready for a break. I’m not the first and probably won’t be the last to do so. Early stages of growing ventures almost always experience changeovers in membership, owing largely to changes in the ventures themselves. In my case, I think I’m going to have to chalk it up to a lack of significant change in Sansar so far.

Sansar’s glacial pace of development is a strength in that it is being careful not to introduce too many bugs. And Linden Lab is acting strategically instead of tactically in introducing features and policies. But I would argue that that glacial pace is also Sansar’s greatest weakness at present.

Allow me to present a sampling of aspects of Sansar that I consider good and bad and also speculate a little on its future.

I’m not leaving Sansar entirely, but I have effectively halted new development of my products and services for now.

Some gratitude

Let me start with a little gratitude. My experience with Second Life since 2005 gave me a very clear impression that Linden Lab exists on a high mountain far away from the average user. I rarely reached out for help and thought it best to mostly avoid LL. Sansar gives the opposite impression. If you have ever heard the stories from SL’s alpha and beta era users about how fun and helpful LL was in those early days and been envious, I encourage you to join Sansar now. You’ll get regular chances to talk to some of LL’s most important people and influence the direction Sansar takes.

Linden Lab’s staffers have been professional, friendly, and helpful to me all along the way. And as far as I can tell, to pretty much everyone. It’s hard to overstate how rare and valuable this is in a technology platform provider. I hope LL will keep this spirit alive as Sansar grows.

Will Sansar grow?

Some might say that this is the most important question on the minds of everyone in Sansar. And among its competitors. Although I can’t know for sure, my current best guess is that it will not grow appreciably in the near future. This is a key reason that I’m taking a break.

What would it mean for Sansar to grow appreciably? I’ve been collecting gross usage data for many months now and have made some of it publicly visible. The rates of visitors make it clear that it hasn’t been growing this year. The peak number of visitors online at a time tops out each month at about 40 to 50 people. That means that there’s never been more than 50 people online at one time since April 2018, when I started collecting the data. And the individual days of each month have a peak concurrency typically from 20 to 50. While it should be fairly easy for LL to get that peak consistently up above that, I don’t think I would consider anything less than a tenfold increase (peaks of around 500 concurrent users) to signify real growth. I think we should consider targets of 1k, 10k, and 100k concurrent users per day as genuine milestones for the growth of Sansar or any of its competitors to reach.

Why hasn’t Sansar grown, then? Key people within LL will publicly and privately tell you that they haven’t tried hard to reach a mass audience yet. Their focus has understandably been on “going deep instead of going wide”, meaning adding the features content creators need to power their experiences before worrying about mass adoption by end consumers. While I agree with this strategy, I don’t think that’s the full answer. It’s not like LL is pushing back against hordes of people waiting to come in.

I’m going to argue in the following sections that Sansar isn’t growing because it is not yet ready to grow and won’t be anytime soon.

Why a new game engine?

Many people have asked why Linden Lab chose to create a brand new game engine from scratch. One simple answer is: Because that’s what worked with Second Life. But back when SL came out, there arguably wasn’t a solid off the shelf game engine available to build SL on top of. It made sense back then to roll your own.

You can’t really argue that anymore. Unity alone boasts over 100 engineers working daily to expand their game engine. New features seem to pour out of Unity constantly. How could Sansar hope to keep up with their pace?

I’ve started learning Unity using the many free tutorial videos online and by creating demonstration projects myself. Earlier this year you could have argued that Sansar’s rendering engine gave experiences a certain polish that was hard to come by with stock Unity, but with the beta release of their High Definition Rendering Pipeline, that slight edge has vanished. See Unity’s Book of the Dead technology demonstrator for an example of what’s possible now, even for a realtime game:

Arguably, all Sansar is doing now is trying to catch up to the basics of what is available via Unity, Unreal, Cryengine, and other mature and evolving platforms. So what is it that Sansarians are getting from this effort, which is probably consuming most of Sansar’s development budget and time?

The most significant benefit I’ve heard presented is consistency. LL argues that content we create for Sansar today will still be available tomorrow. However, I don’t think history bears that out. Most of the code I’ve written has become broken or badly performant as the platform has developed. I’ve kept ahead of this problem by frequently reinventing my product lines and encouraging my customers to keep upgrading. I’ve closed almost every experience I’ve ever created because they have broken down a little more with each passing update.

You might argue that that’s a problem limited to scripting, given how new and active this area of development is in Sansar. I don’t think so. We have repeatedly seen examples of changes to physics, rendering, and other aspects of the game engine that have broken old content.

You might argue that’s all because Sansar is in beta now. Eventually, LL will get to a point where they are happy with the platform as it is and never introduce breaking changes again. If only that were true; but I don’t believe it.

Moreover, I’m not even sure that’s a good thing. What good is a game engine that does not occasionally introduce huge improvements over time that make you want to abandon your old content? SL is replete with examples of this. Who wants to buy a car made of “prims” (cubes, spheres, and other basic shapes) now when you can get a mesh version that’s better in every way? And look at how much Bento has effectively outmoded almost all the older ways of creating and outfitting avatars. Sansar should actually plan to break things sometimes if they want to keep up with advances in technology.

The only other cogent argument I’ve heard about why it’s good for LL to create its own game engine is that doing so allows Sansar to have an easy to use interface for creating experiences. Well, it’s true that Sansar’s interface is simpler. But again, I’m not convinced that’s entirely a good thing. Moreover, if that’s the best argument, then Sansar may turn out to be in a race to the bottom, almost always favouring simplicity over features and performance. That strikes me as a losing strategy in the long term.

My own conclusion is that choosing to create a new game engine from scratch was probably a fundamental strategic error. At this point, I don’t think I could see LL backing away from this choice and starting over with an off-the-shelf alternative, which means that they have an enormous amount of work ahead to try to catch up and keep pace with the industry. Sansar’s experiences look beautiful, but that’s easy enough to achieve in other platforms already.

You are a precious snowflake

Following the successful model of Second Life, Sansar offers users the chance to create a customized avatar and identity. But I think it would be overselling Sansar to say that it has really achieved that.

One key to SL’s success is the mix and match approach to avatar construction. SL may drive new users crazy just trying to understand all of its terminology or even how to put on your pants like everyone else. But at least you can do it all your own way. In Sansar, you have two basic roads you can take. You can create and import a whole avatar or buy it from the store and look like everyone else with that avatar. Or you can use one of the basic male or female avatars, to which you can make minor tweaks. And then, you can buy clothes and accessories. It’s not awful, but it’s clear from so many complaints by visitors to Sansar from SL, and from many requests from Sansarians, that this isn’t a level of customization that is sufficient for users who view their appearance as a critical part of their identity. Sansarians just don’t feel like they can personalize their avatars enough yet.

Sansar will no doubt improve. Eventually we’ll be able to change “skins” on the default avatars. We’ll have many more adjustments we can make to the basic human avatars. We’ll be able to add custom animations and blend them with VR inputs. There are all kinds of great avatar things coming. Eventually. And slowly.

What is social VR, anyway?

One key assumption I’ve been making is that social VR is a separate animal from games. What I didn’t give much thought to is that Sansar is a game platform. I have been creating games and other interactive experiences in Sansar all along and treating it as one. And it is, technically. But as described above, it’s very limited.

Social VR is supposed to be about platforms for social experiences. Yes, interactivity is part of that, but the main thrust is supposed to be creating friendships and finding common reasons to meet and coordinate in a virtual world. But actually, at this point, I’m not even sure what the heck social VR really is, to be honest. There are already other games where people work together and make friends. There are tools for people to work together and socialize, including Discord, which Sansar’s community uses to great effect. I can go with friends and watch a custom VR concert by Imogen Heap in TheWaveVR. What remains that isn’t already covered by other offerings? Or is it just an amalgam of those things?

One thing a social VR platform can offer is a way for you to craft your own avatar and use it in all experiences. You don’t have meaningful portability of your identity across other multiplayer games right now, so that’s a benefit.

Another thing a social VR platform offers is a market for assets, including avatar fashions and objects like houses and trees for building experiences with.

It’s hard to overstate the value of having a well-crafted currency for microtransactions. That is what should power Sansar in the future as business-minded creators that make products, spaces, events, and other services get motivated to earn money for their work.

But what if none of this matters? What if people find most of what they want out of social VR in the fragmented alternatives to Sansar and other social VR platforms? Will users seeking games really come to Sansar to play HoverDerby if they can get more compelling games directly through Steam or other platforms? Will they bother coming for slick Sansar lighting when it’s easier to create a refined avatar on SL? Moreover, will they come to meet people here when SL and VRChat offer more compelling alternatives?

It’s all about multiplayer

Until recently, I bought into the idea that what separated Sansar from Unity and other game platforms was multiplayer. It’s one thing to make a game that you download and play by yourself at home. It’s another to be able to interact with other people in real-time to battle, cooperate, talk, flirt, and so on. My world was rocked when I learned that Unity, in fact, supports multiplayer games. Moreover, they are on the cusp of releasing a totally new version of multiplayer support, complete with a dedicated hosting option so you can focus just on creating and maintaining your game and customer base.

It turns out that Unreal offers some multiplayer support, too. So does Amazon’s Lumberyard. Which makes sense, given Amazon’s massive cloud service, which hosts Sansar and will soon host Second Life, too.

Why would this change anything for me? Because I thought this was something only social VR platforms like Sinespace were able to offer. To me, the server side was where the real magic happened. But at this point you have to ask yourself whether you would be better off creating your social experience from scratch using a game platform like Unity instead of Sansar.

Should I create my social experience in Unity?

The short answer for right now is: Probably not. I’ve done enough work so far in creating technology demonstrators of multiplayer VR games to know that it is eminently possible, but you’ll have your work cut out for you. If you don’t mind investing in solving some of the tough basic problems, such as synchronization of players and creating an avatar editor, the sky is the limit on what your game can do, compared to Sansar. But if all you want to do is buy some stuff in a store to play with and get your friends coming to visit, Sansar is the better bet for now.

But this still matters. I’m going to make a prediction now. It won’t be five years before there will be cheap or even free turnkey solutions available for Unity, Lumberyard, and other popular gaming engine/platforms that let you just start building an experience immediately without having to solve those basic problems. It will probably happen over the next year or two and in a gradual progression of features. But as that happens, people really will be asking themselves what the point is of having a social VR platform when they can use a DIY solution to create one from scratch and even host it practically for free.

I suspect some companies will also figure out a way to offer identities, including avatars, as a service to game makers. Imagine crafting your avatar in a separate program and being able to use it in 100 different games without having to recreate it or work with design programs to create compatible import files. That’s actually what Morph3D’s Ready Room service is launching into, starting with an integration to High Fidelity.

There’s no reason to imagine that someone won’t also figure out how to bring cross-game micropayments to many games to make creating internal economies easier and encourage greater asset portability.

Yes, some of this is speculative, but it’s worth considering a world where social VR is essentially miscellaneous to creating social experiences. A needless middleman.

Is High Fidelity the right way to go?

I’m convinced it is not. It may sound like the pipe dream I described above fits High Fidelity‘s model to a T. After all, you can host your own experiences (“domains”) on whatever server you wish. They have a blockchain-based currency that works across their distributed world. And you can modify their open source client and server software to suit your particular needs.

But nope, HiFi isn’t the same at all. It isn’t a popular gaming engine. Like Sansar, it is a proprietary technology geared primarily toward social VR. It largely replicates Second Life’s overall model, but without the centralization of servers and assets.

Is centralization good? There are some benefits to owning your own VR server, such as choosing how powerful a computer you need based on your expected usage. In October, HiFi managed to get up to 423 visitors packed into a single domain for a load test. This impressive feat required provisioning some really beefy servers that are more expensive to run than most people would care to pay for their own domains. Sansar’s experiences all run on identical servers.

But consider intellectual property (IP) rights, especially copyrights. You can go to Cubebrush or any number of other asset stores and buy models that you can then easily misuse beyond the terms of the sale. Asset creators face the very real prospect of copyright violations that are very difficult to prevent. This is the same problem High Fidelity faces. I would argue that their blockchain-based certificates of authenticity are at best a fig leaf that won’t really protect IP rights. Controversies have already arisen in HiFi and VRChat over copyright violations.

At least Sansar offers content creators the possibility of having their IP rights protected by having assets sold in the store kept behind the Great Wall of Sansar.

But ultimately, HiFi is going to face the same competition as Sansar from outside alternatives. We need to stop thinking that social VR is a truly distinct thing that won’t be affected by ongoing encroachment by popular game engines like Unity. We must factor them into our comparisons.

Is VR dead again?

No. Virtual reality (VR) and augmented reality (AR) usage are steadily growing, especially in Asian markets. Although the pace of innovation of the hardware seems slow for now, customers are already eagerly awaiting many new technologies in 2019, such as the Valve Knuckles controllers, the Oculus Quest stand-alone system, and Magic Leap One headset.

Content creators in VR platforms like Sansar should seriously consider focusing on creating VR-centric experiences, and not worry about making them desktop friendly. Why would I say that, given that most Sansarians don’t have VR equipment? Because there are already oodles of desktop (and even mobile) virtual worlds to choose from, including SL. If that’s all Sansar is, don’t expect it to take off. Recognize the “VR” part of “social VR” and create experiences that can’t be enjoyed in any way other than in an embodied first person point of view with hands and eventually more. I hope that the success of Beat Saber has hammered that point home by now.

What if Sansar fails?

It’s a bit sad that there are many vocal Second Life users who are hoping for this outcome in the belief that Linden Lab will use the money saved to improve SL faster. Personally, I’m not ready to predict Sansar’s imminent or future demise. I still think Sansar has the best shot of success among all the social VR platforms right now.

But let me just speculate for a moment what would happen if LL were to give up on Sansar development and essentially shut it down. I’m going to imagine it from the perspective of what I would do if I were at the helm of Linden Lab and not make an actual prediction, per se.

What would cause me to shut down Sansar? Most likely, this would result from a series of very visible signs that people are preferring some alternative to Sansar and that doom Sansar to have a small niche audience. If, for example, someone made a YouTube video showing how you could create your own multiplayer VR social experience from scratch in Unity in 15 minutes, that would be a solid sign. Or if HiFi’s rendering engine was as good and their typical daily concurrency peak was over 10k and growing, while Sansar’s remained flatly under 1k. It wouldn’t be one single thing. It would be several devastating signs like these that would do it.

Assuming I just shut down Sansar, what would I do with the remaining staff, budget, and experience gained from Sansar? The obvious answer is: Improve SL. I would probably take a big gamble that would still be bold but not as dramatic as Sansar. In particular, I would turn SL into a “hybrid grid”. Let me explain what I mean.

SL is a fossil. Yes, there’s plenty of room to improve it, but the gradual improvements to it are always supposed to be backwards compatible with content going back to 2002. That hinders SL’s potential immensely. That’s why LL took the big leap into the Sansar project as a totally new world to begin with: for a fresh start. I think they know that Second Life’s days are numbered and that something will eventually draw most of SL’s population away.

To breathe new life into SL, I would engineer a significant and only partially compatible version of the Second Life viewer and servers. Let’s call the current technology “SL classic” and the new part “SL next-gen”. The next-gen part of SL would take advantage of many of the lessons learned and technologies pioneered for Sansar. Picture having a new SL client that supports both classic and next-gen sims. Those sims could live alongside one another, as though two grids in one. Your account would be good for both. So would your money. But maybe you would have to create a new avatar in the new grid. Or maybe there would be some conversion utility. And some assets you own in the classic grid wouldn’t be fully compatible with the new one. And assets made specifically for the next-gen grid would be largely incompatible with the classic one. The overarching goal would be to gradually migrate everyone over to the newer platform and eventually retire the old.

Why do this? Because there’s a lot of good ideas in Sansar that can only be brought to SL if they are willing to break some things. For example, there really is no reason to run a sim 24/7 when nobody is using it. I estimate that LL has 5,000-10,000 beefy servers now running over 20,000 sims. With peak concurrency around 50,000 users, that amounts to about 5 people per expensive server, with actual people concentrated in larger numbers uncomfortably on only a few of them. If there are around 22,000 sims presently active now and only 10% of them have at least one person on them at a given time, that’s about 2,000 sims active and averages out to 25 people per active sim at peak concurrency. You could trim those 5,000-10,000 sim servers to more like 1,000-2,000. And in the process, you could potentially cut total sim server costs to 1/5 what they are today. Pass those savings along to SL residents and renting an entire private sim could average out to US$50 per month instead of $250. Imagine paying $5 a month to rent a 1/16 parcel with over 2,000 prims budgeted instead of for $22 and up.

But let’s be realistic. There are going to be some sims that need to be open 24/7. So maybe it would make more sense to charge sim owners based on uptime. Practically speaking, it would be more like people with popular sims continue paying about $300 a month, while those who have unpopular parcels on relatively inactive sims pay zero or just some small maintenance fee; maybe a dollar. I think most SL residents would agree that a move to demand-based uptime fees would totally change the equation. People who just want to create something for fun or are just getting started with their venture would love the idea of having effectively free land. Moreover, this would shake up the land market, because sims would have genuinely differing value based on how popular they are and thus how much they cost. That was the case in the early days of SL and it drove the development of SL’s most lucrative private market early on: land sales and rentals.

It is entirely possible that this drastic change to the way land fees and uptime work would result in many residents choosing to rent whole sims instead of small parcels. Rarely used sims could effectively cost nothing, so why bother choosing a smaller parcel?

Moving to demand-driven uptime will require a change to the scripting model. You wouldn’t be able to put a “server” type object on a piece of land and expect it to be available 24/7 unless you were willing to pay for that uptime. For this and other reasons, I might choose to go with a C# based scripting language for the next-gen grid. Sansar’s script API features a number of approaches that encourage scripters to budget server resources carefully that is very different from the approach taken with LSL. And the new land pricing model may encourage people to pay small rental fees for tiny parcels that are online 24/7 just to house their server objects.

One key reason we can’t use VR equipment with SL is that SL has a relatively low framerate for sims, maxing out at 45 frames per second (FPS). That’s true even if your own video card purrs along in SL at 300 FPS. LL chose to standardize Sansar’s servers to 90 FPS and targets that minimum for VR clients. So that would be something worth changing in the next-gen sims in this hybrid grid. This would need to be true for scripts, too. But this could bring the real-time dynamic systems I got used to creating in Sansar to SL. Right now, using scripts to animate objects in SL is woefully limited, making many interactions clunky at best. Running a next-gen sim and scripts on it at 90 FPS would be a genuine game-changer and make SL a relevant player in the social VR sphere.

PBR is here, even if SL doesn’t truly support it yet. This would be a great candidate for a next-gen SL client. Just getting designers to stop manually baking shadows and faking things that PBR materials handle easily would be a massive change. Introducing something like Unity’s HD rendering pipeline would give content creators a chance to start over and years of new capacity to chew on. And PBR-centric content would be readier for the advent of mixed raytracing and PBR rendering that is on its way.

A next-gen grid would give LL the chance to realign its pricing model with reality. The gradual introduction of land impact (LI) to replace the older prim counts was a good move. But SL still does not let content creators and users feel the real cost of large textures. And certainly does not properly make end users bear the cost of resource-heavy avatars. It is not unusual for a single avatar visiting a sim to have more polygons and texture memory usage than the entire sim. If anything, this creates perverse incentives that keep SL from growing to allow more than around 80 people to comfortably be together on a sim. Metering the resource usage of avatars and allowing parcel owners to limit access or charge varying fees based on that would alone encourage a significant growth of venues that can be popular. But more generally, a more comprehensive computation of storage, network, and rendering costs and incorporating them into usage constraints and upcharge fees would be a smart move for a next-gen grid.

Possibly offering an instancing model, wherein thousands of players can exist on parallel instances of the same sim, may be just the thing for attracting mainstream musicians and other content providers back to SL. This may be a step too far away from SL’s model, but it’s worth considering. Another possibility would be offering upgraded server hardware for those that wish to provision for larger on-sim populations or heavily interactive games.

There are many possibilities that would open up if I were in damage control mode after Sansar had died and I wanted to know what to do next. But I would likely favour doing some sort of hybrid grid as described above and seeking to gradually migrate SL’s residents and ventures into the newer technology platform. That’s what would make SL’s population grow again and give SL many more years of life ahead.

Not dead yet!

But Sansar is not dead. It’s still going and growing. Nevertheless, it seems to be in my best interest to take a break from it.

My main reason for me leaving, for now, is the glacial pace of development. If Sansar does die, it probably will simply be because one or more competitors outpaced it. But in the meantime, Sansar’s pace of innovation is too slow for me. I’ve spent the past 15 or so months creating value for Sansar’s residents, but it isn’t paying off yet. I suspect I have had more financial success than most in Sansar, but until money starts flowing from end consumers for goods and services, content creators like me will have to keep waiting. Sansar’s feature deficit is arguably the main thing standing in the way of that. I predict that Sansar is at least another year out from being a ready enough platform with which to create compelling content and avatars; enough to start drawing a mass consumer audience. And even when those features are there, it will take time for the back and forth process between early adopter creators and early adopter consumers to create the feedback loop that inspires masses of creators to start investing and thus drawing masses of consumers.

In the meantime, I’ve taken on two separate projects outside Sansar. One is in Second Life and will start paying right away. The the other involves Unity and is speculative.

I plan to keep an eye on Sansar and continue to support my customers there. I’m not completely leaving. But I am chastened by my own career needs and by the realization that Sansar isn’t going to be ready for “prime time” for those who want to make a career of working here within the next few months.

Ryan: I want to take this opportunity to publicly thank Galen for all the hard work he has done to date to help build Sansar. He has been a key player in making Sansar what it is today. I consider him a scholar, a gentleman and a friend, and I wish him the greatest success in whatever work he chooses to undertake in the months and years ahead, on whatever platform (SL, Sansar, or something else).

Liked it? Then please consider supporting Ryan Schultz on Patreon! Even as little as US$1 a month unlocks exclusive patron benefits. Thank you!