Project Sansar is just around the corner, but I doubt that Second Life is going to disappear

As I patiently sit at home while waiting for one of my servers to complete its tribulations (long story!), I was briefly scanning some of the ‘old’ places in the SLogosphere (do people still call it these days?) where I used to hang around for inspiration about what would be the Next Best Thing coming to Second Life and/or OpenSimulator. And, not surprisingly at this time, the issue about Project Sansar (which is actively inviting content developers to participate in a slightly-more-open-almost-beta-phase) robbing Linden Lab of all resources so that Second Life will crumble to dust comes up — like it does every time the Lab announces a further step towards completing Project Sansar and opening it to the public.

Disclaimer: I did apply for becoming an early-adopter tester of Sansar, but LL (as usual!) ignored me 🙂 Oh, how I remember some talks with Lindens iRL about the many times in the past when the Lab choose to ‘ignore’ me…

Anyway, as far as I know, LL simply has no manpower (not even by kickstarting a huge community of generous developers willing to share their work for free) to establish a shiny, new virtual world which has approximately the same amount of content as SL — and that simply will never happen. Rather, what LL is hoping is that top creators in SL, who all work exclusively on meshes and have done so for years now, will simply upload the same content they are selling in SL to the Sansar-created ‘experiences’, or ‘migrate’ their own fully-meshed ‘experiences’ to the new platform. There might be a problem when those platforms require extensive scripting, however; I have no clue how objects in Sansar are scripted (and that was actually the main reason for asking for an ‘invite’), but AFAIK, it will be almost surely be something totally different from Second Life or OpenSimulator. Nalates Urriah claims that

Sansar will use industry standard tools, animation, and programming language.

which lead to interesting discussions about what the ‘industry standard’ programming language would be, a bit all over the place, and, of course, nobody really knows — because those who know are NDA’d to secrecy. The rest is allowed to speculate, although Sansar’s registration form for potential parties interested in previewing their technology asks if the person has knowledge of programming in ‘C#, Java, LSL’, which certainly allows one to conjecture that these languages — or one closely related to them — are the best likely candidates. The truth is that everybody who experimented Sansar so far talk about its builds, not about its programming… but, anyway, we can safely assume that whatever LL picked for Sansar as a programming language, it will not be LSL (Linden Scripting Language), and whatever it will be, it will be utterly incompatible with what is being currently scripted for Second Life.

In other words… LL has been quite clear in the past that, although they have considered the possibility to allow certain content to be migrated from SL to some of the Sansar worlds, in practice that would tie their new software too much to ‘legacy’ support, and that, of course, has a huge cost — of development, of maintenance, of having much more limited options to allow Sansar to do the amazing things it is supposed to be able to do. So, no, this will hardly happen — ever. It’s not just about prim-based objects. You could theoretically export all prim-based objects as meshes (this is what happens inside the SL viewer anyway, and it’s been that way pretty much forever) in an automatic way, but except for unscripted objects, this will just solve half the problem. Even a simple pose ball dropped on a chair would not work in Sansar.

Therefore, like all attempts at ‘new’ virtual world platforms, Sansar will have to compete for developers to be willing to do the hard work, and while many are generous with their work, and are certainly willing to work for free, most (for good reasons!) are not. This will mean a very slow initial growth rate, at least in terms of new users. Nevertheless, I remember that I was astonished at the amount of content that had been produced by as little as 5,000 residents when I joined SL — 5,000 people working for free for a year can mean a lot of content! But while it’s conceivable that LL might be able to attract small game developers to use their platform to create novel ‘VR experience’, at the end of the day, if those companies are really small, they will wonder how much return they might get from such games — which means allocating human resources to Sansar which could have been more profitably employed in designing mobile games or web-based so-called ‘casual games’, where a small developer has a much higher chance of becoming marginally lucrative. And, of course, the huge software houses developing Triple-A games have their own engines and their own platforms, they won’t be willing to trust their multibillion dollar incomes to a small-ish company who has just released the ‘next best thing in VR’.

In other words: independent professional 3D content designers are far better off using existing marketplaces to get their income — and this means things like the Daz3D Shop, of course, but also Second Life, which has enough critical mass in terms of users willing to pay for quality 3D content (and the same is true for, say, IMVU as well) — instead of jumping into the Sansar bandwagon hoping to establish themselves as the ‘first’ developers and thus being also the first to earn cash — and be able to grow faster. But that wasn’t even true for SL at the beginning. Professionals laughed at the in-world tools and how crude they were, even for 2003, and ran away. They came back only when things started to become interesting in terms of having enough people willing to pay for content — and then the pros came in and displaced all amateurs, even those who had years of experience in SL.

But is it conceivable that such developers would do the same for Sansar…? Hardly. They have bills to pay. That means focusing their talent and energy into things that they can turn into money in a reliable fashion. This is the main reason why the pros are all at Second Life, but are incredibly reluctant to offer their content on any of the OpenSimulator grids, even if it is not much of an effort to upload their content there. Unlike what will happen with Sansar, content developed for SL will work in OpenSimulator as well — at least in 99% of the cases, and with a little planning ahead, it can be guaranteed to work everywhere. So it’s not really about ‘wasting’ time to ‘learn something new, from scratch’ — it’s ‘build once, distribute to several grids’. But few do that. Why? Because the pains of dealing with content that disappears on shaky OpenSimulator grids, or the risk of content theft, and so forth — real or perceived, that is irrelevant — makes the pros give up on ‘experimenting’ on ‘new and unproven’ grids. Given enough time, they will slowly change their minds, as the Kitely Market so well shows. Nevertheless, contrast the thousands of products on Kitely with the millions on Second Life!

Of course I might be all wrong about it, but for those of you who have known me for a long time, I always ‘go after the money trail’ to judge what projects will have success, and which ones will not. LL has obviously several choices with Sansar: licensing it to the US armed forces, for instance, where they have already an open door and can successfully compete with the many other VR products which work exclusively with defense contractors — and which, under Trump’s administration, are very likely going to get much more money to invest; or — who knows? — try, once more, to push the technology to Facebook, since their own VR project has clearly failed (in spite of having hired so many ex-Lindens!) and will probably never be launched…

To be honest, the only real and serious alternative to Second Life continues to be… OpenSimulator, in the sense that it is the only virtual world platform which is totally compatible (well, 99% compatible…) with SL, and it’s really the only place where content ‘made in SL’ can be reused exactly as it has been originally designed. But — and this is a HUGE ‘but’ (pun intended 😉 ) — OpenSimulator grids are notoriously unstable compared to LL’s, and the main reason for that is the sheer amount of hardware and system administrators and engineers that LL is able to hire to keep their own grids running. OpenSimulator grids, by contrast, may have wonderful software, but grid operators are tiny companies (sometimes one-person-operations), with few access to resources, and hardly able to manage a grid with a fraction of the complexity of LL’s. And that’s why they only have a fraction of the users, and a fraction of the content (well, there are many other reasons, too). To stretch the analogy: aye, they also charge just a fraction of the price that LL charges, but you literally get what you pay for.

It’s boring to spend all the time programming complex AI stuff on OpenSimulator, so I’m happily adding a few tree meshes just to liven up the place!

So… I don’t know, but SL, even with a very slow decline in population (and consequently income for LL), is still their ‘cash cow’ and will continue to be so for many, many years. It’s possible that some ‘technological advances’ made for Sansar eventually trickle down to SL, and this will be especially the case if LL does not secure a reasonable user base for Sansar — and I’m talking about institutional customers, not end-users — which allows further development on Sansar to continue solely based on the income they get from it. This is unlikely to happen for several years — and during that time, LL will still need to extract as many L$ as it can out of SL to keep themselves financially afloat. The problem here is the very-long-term — at some point, SL will not generate enough money for LL to keep it going. The only possible (but not likely) way for LL to generate an income at that time is not by cutting prices and expect people to rush in; this will not happen, since LL knows very well, from past experiences, that the price of tier is not very elastic, i.e. by cutting it by half you do not get twice the users, but just get existing users to convert to slightly more expensive tier (which will be cheaper with the price cuts) and essentially get more land for the same amount of money they are willing to spend; also, of course, by slashing the sim prices in half, LL would have to explain to their board why suddenly their income was divided by two (even if the overall profits might remain the same…). So… this will not happen.

Instead, LL may reduce their costs slightly by improving the SL viewers and the sim server software, so that they are more stable, give less problems, offer more tools for residents to fix their own issues instead of having to ask LL employees to fix it for them… so, improving their software will eventually lead to less tech support calls, and that means needing less people, and thus being able to maintain the same grid for lower costs. Still, that works up to a level — the cost of tech support will never go down to zero — because there will always be the need to call someone from the Lab, even with the best software in the world.

What will the next step be? Mind you, at the current rate, this will only happen around, say, 2025 or so. I think that LL will have no choice but to get rid of their own proprietary server software — and let go of their server developers — and start using OpenSimulator themselves. The reason is simple: OpenSimulator has its own pool of developers (with some hiccups after 2015, but things are still progressing, I get updates almost every week, so I’m sure there are people working in it), some of which funded by the Overte Foundation (assuming it’s still around, they were created to fund OpenSimulator development and allegedly did so from 2011 to 2015, but I haven’t heard from them since then), with contributions by the many programmers and developers that use the software. Linden Lab could very well use it on their own hardware and nobody would really notice the difference; after all, many OpenSimulator grids are closed too, just like LL’s. Sure, they would have to migrate a lot of content, but 99% of it would work straight out of the box — and the more time elapses, the more OpenSim remains ‘compatible’ with LL’s latest and greatest developments. They have actually considered this around 2008/9 but abandoned the idea — at that time, OpenSimulator was still lagging behind SL’s own proprietary software by a huge margin — mostly because of coding differences, different programming languages, and so on. OpenSimulator, for instance, is modular, while SL’s own server software — at least back in 2009 — was monolithic, but compact and (allegedly) efficient. OpenSimulator had been refactored to address a lot of redundant code (you would not believe what kind of crazy things there is hidden under the hood — once, I found that they had a fully-fledged Prolog compiler somewhere in the code! Prolog is a language developed in 1972 or so for programming artificial intelligence, and possibly someone thought that it would be a nice thing to have on OpenSimulator as well…) in 2011, if my memory serves me well, mostly to start addressing a lot of innovation that LL was bringing out at the time, and it still keeps trying to match LL’s software feature for feature. And, of course, LL might also emulate Kitely‘s intelligent business model to deal with servers-on-demand. That would certainly cut on hardware and hardware maintenance costs, while still being able to deal with a huge grid which is mostly empty all the time (and will always be). Actually, I have been wondering for years why LL doesn’t simply use all their hardware to create their own cloud services using an open-source cloud management software — it would fit perfectly for a ‘sparse grid’ model, where only a limited amount of people are online simultaneously and clustered on specific grids, while most regions are empty and therefore do not require wasting CPU and memory cycles. So, again, there is room for cutting the costs even more that way — after all, LL has invested in a lot of hardware, and it makes sense to use it more efficiently (and they already use cloud services from Amazon, mostly for caching; but their own core servers, running the sim software themselves, could also be deployed on an internal, private cloud; with the advantage that free and open-source software as the one I linked to actually uses the same API as Amazon, so deploying on either the internal, private cloud or the external one provided by Amazon could use the very same tools that LL has already developed).

Anyway… to be completely honest, I used to be a much greater fan of OpenSimulator before I looked at its code 🙂 — although I certainly commend those legions of programmers who did an amazing job at reverse-engineering LL’s software — and so I can understand LL’s reluctance in using something they have not developed themselves. But they might not have a choice there. They will still and always be the largest grid by far, since no one else can claim the same amount of content; it’s impossible to replicate, in a few years, what literally millions of people have done over a decade and a half. The sheer manpower used to develop SL’s content is simply unthinkable; it would be close to claiming to be able to launch a social network competing with Facebook, saying that they would have pretty much the same amount of users and the same amount of content!

So… honestly… I think we’re good up to, say, 2030. After that I can imagine that LL will really start to get into problems, because their business model will start to crumble at the edges. There will eventually be a point of no return: this will happen when the amount of paying users simply drops below a certain threshold which no amount of cutting costs will allow Linden Lab to make a profit. And after that… either Sansar is a huge success and brings in a cartful of gold into LL’s coffers, so that Sansar can do for Second Life what currently Second Life is doing for Sansar, or, well, they will have to drop SL.

But that will not come so quickly. As I’ve said over and over again, LL has actually hit on the nail by getting the correct business model for a virtual world. Granted, this actually wasn’t how LL initially planned SL (it was supposed to be a subscription-based system), but it was how it evolved, how it was fine-tuned, and how it brought in millions to LL’s shareholders. Not billions, but millions. We should not underestimate LL’s ability to have been around with a very innovative product, survived the threats of ‘competition’ over and over again, and still comes ahead as the leader of the pack (granted, at a time that the pack has dwindled to a handful of wolves…), paying millions to their shareholders, year after year, and having done so for at least a decade. Linden Lab is much more than a technological success; it’s a business success, and that is much more important.

Now, will Project Sansar become another success, or will it just be a ‘distraction’ from LL’s core business? Again, we can only speculate. I believe that a licensing model, especially one that can be licensed to Really Big Institutions (and of course the US armed forces is the best choice by far, since they already are consumers of VR technology), and that does not need many of those to keep LL’s afloat as a business — that will work out in the end. Linden Lab just needs to market their technology to a handful of Big Corps and convince them that they have a much better platform than everyone else’s (and that includes, of course, Philip Rosedale’s own High Fidelity). They have, of course, a good (and long) track record of profitability — and that matters when you’re signing contracts.

On the other hand, they’re also (in)famous for failing at everything else but their core business. This is no bad per se; after all, we all know how Google catastrophically fails in pretty much everything they put their hands on, and even when they have a good idea once in a while (like Google Docs — which they have acquired, not developed in-house), they easily get outbeaten by their direct competition. The list of Google failures is miles long, but, alas, Google can be forgiven for all of them — since they have so much money from their core business that they can afford to fail. Linden Lab is not Google by about a factor of thousand, but they have the same stance: they make enough money from SL to be able to afford to explore alternative venues of getting an income, and, even if they fail, they will have learned something from the lesson, and eventually it will trickle down back into SL. Or eventually not. It matters little when you’re able to make investments without hurting the company — R&D is good, acquiring knowledge (even if it’s only about how to avoid failure in the future) is good too, failure is ok if you learn with your mistakes.

So, at a personal level, of course I wish LL a big success with Sansar — one that allows them, as a company, to step into a bright and stable future, keeping stakeholders and customers happy. While I only reluctantly admit that Sansar may turn a few heads here and there, catching the attention of a handful of developers, I still believe that the only people that will really be interested in Sansar, at the low end of content development, will be SL users. The rest of the mainstream world will most certainly continue to play Triple-A games from the top companies, not from a non-descript unknown company out of San Francisco. But it’s highly likely that LL gets a ‘big name’ to give Sansar a push in the right direction, allows them to pay off their investment into the new technology, and guarantees its further development in the near future. They are better positioned than anyone else at this moment to do exactly that (because they have a track record of success with a virtual world that nobody else has); and their closed-source, proprietary technology will appeal to those companies who frown upon the unknown dangers and pitfalls of open-source solutions. While on the other hand academics and freelancers and small companies wishing to test something ‘bright, new and shiny’ will very likely pick High Fidelity first, since it’s open source, reasonably well documented, and has a transparent development model — exactly the opposite of Sansar. Also, High Fidelity’s business model is not easily understandable at this point (Philip is still busy burning venture capital at this stage, it’s still too early to worry about making money), while Sansar, even if it has not been fully disclosed yet, points at a much more familiar model to what we have in Second Life, at least for the end-user — but it might be much better sold as licensed platform, in effect becoming what ‘Second Life Enterprise’ was supposed to be a decade ago: a virtual world that LL can shrink-wrap and place a price tag on it before shipping it to corporate customers.

In the mean time… it’s the end-user, we SL residents, which make LL a happy, profitable company, and LL is not forgetting us so easily. SL will always be the ‘fallback’ option for LL if Sansar also fails, like everything else that LL attempted to do. There is most certainly a lot of concepts, ideas, and even code that might be possible to adapt to SL, so that the past years of developing Sansar have never been a waste of time and money — sometimes you need a bit of detachment to think of novel ideas to deal with old problems. And certainly LL will rely on a happy crowd of SL residents to pay their tier, and therefore keep the company afloat, while they are busy pushing Sansar to new kinds of customers — or even eventually round up a few more investors to help to pay for further developments of Sansar (and potentially Second Life as well), although LL has been reluctant to do so in the past and has always preferred to do their R&D out of their pocket and avoid being ‘dependent’ on shifty venture capitalists eager to ‘make money fast’, often steering companies in a totally wrong direction just to get more bang for the buck. In a way, I feel more confident in LL when they spend their own money in whatever projects they are interested in investing: it’s risky, yes, but at least it’s their risk, and they can afford to lose money that way, since they will not have to pay it back to anyone else. That’s still the ‘old way’ of making careful business (an interesting anachronism for what is supposed to be a hippy, cool Californian tech company!) and one which has served well to quite a lot of huge megacorps that have been around for a hundred years; there is absolutely no shame in emulating those who have succeeded in the long term.

Even Coca-Cola survived some of their product fiascos (anyone remembers the ‘New Coke’?) and are still around to tell the tale; my confidence in Linden Lab is actually reasonably high at this moment, since, in spite of all the doomsayers for the past decade, they have long outlasted them, and are still around, innovating, and investing in new technologies, not fearing to fail; after all, they certainly have dealt with lots of tension and many risks and all sorts of issues and have survived all that. So far! 🙂

About Gwyneth Llewelyn

I'm just a virtual girl in a virtual world...