Why are businesses moving out to other platforms?

2010, the second year of the financial crisis1, has brought a sudden increase in projects using virtual worlds. Speaking from the perspective of my company — and from conversations with other similar companies developing content for Second Life® — this seems to be gathering new momentum. It’s true that a huge part of development in virtual worlds now come from universities and research institutes, but that doesn’t mean that companies are “asleep” or ignoring virtual worlds. They’re not. They just have completely different expectations than they had in 2006/7.

During the “hype years” of 2006/7, and as has been extensively written and explained elsewhere, almost all companies came to Second Life for the “media splash”. They met with serious criticism by the SLogosphere, as residents pointed out that SL was “all about community” and the notion of “build it and people will come” was a complete fallacy. This proved to be correct, as almost all experiments in “media splash” had little impact or result, with some minor exceptions (which have, since then, become case studies) that were lead by teams with a lot of expertise in SL.

A few experiments lasted several years after the “hype era”, and many were targeted to the community (Orange Island comes to mind). But slowly they also disappeared. Serious community management takes a lot of time and a lot of money (if you’re a company employee working full-time on managing communities), and the returns are questionable. Marketers love numbers — especially big numbers — and even a long-term community in SL will reach “thousands” of people, not “millions”. While the use of ads on Facebook is also questionable, marketeers have this cosy feeling that for a few US$ they’re able to reach “millions” (even if I’m prepared to claim that 99.99% of them never see any ads — this number is based on three times the effectiveness of email spam, so it might even be lower than that). Thus, I claim that an ad displayed “to a million Facebook users” is actually only seen by around 100 users; while a series of events in SL reaching thousands will, effectively, get those thousands to acknowledge your brand (or product). And you will know who they are (well, at least their avatar names). Of course, an ad on Facebook is much, much cheaper…

The SL residents have this feeling that businesses have completely gone out of SL, and that even universities are “not doing much”. This feeling is based on simple observation. Few businesses announce their presence. Universities don’t seem to have open islands any longer. The media hardly reports anything being done in SL. Somehow, the residents just think that the hype was gone.

Nothing could be further from the reality. There are now ten times more companies in SL than at the peak of the hype era, and probably a hundred times more universities and research institutes.

So… where are they?

There was a huge shift in focus: these projects are now all internal. Instead of a media splash, community management, events-for-a-general-audience, or any kind of public approach to a virtual presence, all the organisations are now turning to the usage of virtual worlds to address very specific tasks that are hard (or expensive) to do elsewhere — be it on the physical world or on other online technologies (like the Web). While e-learning comes to mind as one of the major uses, some other areas are simulation training (it’s easier to train employees to deal with crisis and emergencies in, say, a virtually recreated hospital in SL, than to shut down a physical hospital to do drills — and way more effective than putting employees in a room with some paper-based maps of the hospital and have them move chits around and tell others what they’re doing) or prototyping. A lot of pedagogical material that is intended for a limited, specialised audience is also placed on privately accessed spaces; thus, the military (to quote a typical example) might not be so interested to describe to a general audience how they train their soldiers in SL.

What this means is that all these projects are “invisible”. Nobody talks about them. They might pop up on internal newsletters or on the corporate/campus intranet, but not beyond that. They might get a two-line mention on a general-audience newspaper saying “Company X is using SL for Y”, but since residents cannot even look up X’s island, this news is quickly forgotten. Also, the scale of the projects has increased. During the “media splash” era, companies would set up their virtual presences in a month, sometimes less; universities quickly opened their campuses in about the same time. Any project much longer than 1-3 months was very rare (CEOs would fear they’d be too late if they took so long in the fast-paced environment of SL!). It was not uncommon to announce the intention to come to SL, followed by a press release a few weeks later when the virtual presence was formally opened to the public.

These days, however, projects in SL are medium-to-long term. From my own experience, they take, on average, 6-18 months to decide, and often 3-5 years to complete. Most of the readers of this blog will not even be around in SL to see any of those projects (the average user loses interest in new technology after 3 years). There are, of course, exceptions. Research projects, for instance, might go ahead indefinitely — they might have started in the 1990s when virtual worlds were brand new, and are now migrating to Second Life and OpenSimulator, and continue to be around for another decade or so (eventually adopting new technologies in the process). Government, military, or huge megacorps might have similar time scales. In the fast-and-furious environment of SL, these projects are so vast in the time they take that it is as if they didn’t exist. They might just publish some “progress reports” every other year or so, which will remain mostly unnoticed by the media, as no public space is ever presented.

Nevertheless, in the past months, I have encountered an intriguing trend. It is generally assumed (and one can just point at the SL Educators mailing list for a reference) that Second Life has only one valid competitor: OpenSimulator, which is basically the same thing, but in full control of the organisation which deploys it internally to create their own grid, since it’s free and open source. Full control in this context does not only mean access (because for that Linden Lab provides the SL Enterprise boxes as an alternative to the main grid) but the ability to change the simulation software to tailor it to specific uses, which Linden Lab obviously doesn’t allow with their own code. OpenSimulator is a natural choice for academics researching virtual worlds since it allows them to change the code at will as part of their research.

OpenSimulator’s main advantage is, thus, the ability to run your own grid, in no matter what hardware you have available. A humble, retired PC, connected to one’s internal Ethernet will be more than adequate to do a reasonable project — assuming, of course, that there will be not many simultaneous users in it. That’s a reasonable assumption for a ‘small’ project. For a larger-scale project — say, a campus grid for thousands of users — the hardware and network requirements increase by several orders of magnitude; and if the organisation wants to additionally give widespread access to “their” grid on the Internet, well, then that means designing a completely new systems architecture to handle the load. In fact, except for some very specific situations (such as for certain military projects, which have to remain confidential for obvious reasons), the actual costs of managing an OpenSimulator infrastructure can go up — even if you can “crowdsource” volunteer work from the students of your university campus — to a point where Second Life becomes an attractive alternative again! There is some middle-ground, too — as Hypergrid Business constantly reports, there are plenty of commercial OpenSimulator grid operators to choose from. Their prices vary — always cheaper than Second Life — but so does the quality of their infrastructure. Also, those commercial grid operators fight for financial stability: a small setback for a tiny start-up may be too much to keep afloat financially, and when that happens, either they are lucky enough to sell their service to another (hopefully more financially stable) grid operator or they will be forced to shut down their grid.

This is not so dramatic as it sounds, because (unlike SL) you can very easily do full backups of all your uploaded content — aye, that includes prims, sculpties, meshes, textures, sounds, animations, scripts and avatars — and move it easily to a different grid (or even temporarily store it on your personal computer — from where you can even run the simulator(s)!), just uploading the content again. There is a pretty standard content archiving format, OAR (OpenSimulator Archives), as well as its companion file format for archiving inventory items, IAR. Both have been around for a long, long time and are very reasonably stable; that means that they will work across all OpenSimulator grids, no matter what hardware it runs on, or whatever database backend it uses.

While OpenSimulator strives for full compatibility with Second Life (until Linden Lab started to include commercial libraries directly on its viewer — such as the Havoc physics engine — you could even use Linden Lab’s own viewer to easily connect to an OpenSimulator grid; these days, however, third-party viewer developers, unable to sublicense those commercial libraries for use in OpenSimulator, have to replace them with FOSS alternatives), it lags behind Second Life in a few aspects — but not that many. After all, most changes in the last years appear on the viewer side; while the Linden Lab system administrators might be tinkering on the servers as well, such changes are mostly “invisible”, in the sense that the communications protocol between server and viewers doesn’t change dramatically. Sometimes it does (such as the change from UDP as the underlying protocol for many assets, which became optionally HTTP over TCP/IP, and later almost mandatory, since LL wants to leverage on the ability to keep HTTP objects inside reverse-proxy caches, to ease the load on their “core” simulator servers — this was something that OpenSimulator needed to change as well, and change it did!), but, more than often, the “changes” in the communications protocol are small and insignificant. OpenSimulator developers are paying close attention to whatever goes on Linden Lab’s Beta Grid, which gives them plenty of time to support any future changes, at about the same time that Linden Lab pushes their efforts from the Beta Grid to the “main” grid. This has been true for most of the things in the recent past. Additionally, of course, because OpenSimulator is free and open-source software, the core developers have added lots of nifty cool features that Second Life does not have — such as extra scripting functions allowing a much closer interaction with the underlying simulator, or, well, NPCs (non-playing characters), behaving just like any other regular avatar, that can be very easily automated with a handful of lines of code. Oh, and OpenSimulator grids are not “isolated” from each other: thanks to Hypergrid Teleport, avatars from a grid can jump to a completely different one, often bringing their inventory with them (content developers can optionally flag their own content as being allowed to jump between grids — which is the default setting). This has been around for a long while as well and is also quite stable these days; originally, it was even possible to “hyperjump” from the Second Life Beta Grid to a few OpenSimulator grids, albeit with lots of limitations; unfortunately, Linden Lab dropped that amazing feature æons ago and has not the least intention of bringing it back.

But OpenSimulator is not “perfect”. Besides the hardware and the cost of networking infrastructure, there is the need to keep a maintenance team with long experience of systems and network administration just to have a reasonably-sized grid to be operational at all times, with enough performance to make users happy. You can get the same performance on OpenSimulator as you get on Second Life — sometimes even far better! — but it requires a decent amount of knowledge about tweaking operating systems, database management, distributed systems, and network optimisation. That’s not for the faint of heart. OpenSimulator has some documentation on its wiki, and there are a few mailing lists as well as IRC channels to go for help, but, in general, it’s not that easy to figure out how everything is put together. Old-time veterans from OpenSimulator will of course easily adapt to changes over time — the basics are still similar — but newcomers will be astonished at the vast array of parameters that need to be tinkered with in order to get OpenSimulator to do what you need it to do (thus the reasonable success of many commercial grids, which have that kind of knowledge, and, for a reasonable fee — much more reasonable to what Linden Lab would charge you — will provide you with a simulator connected to “their” grid, possibly offering you even far more capabilities that you’d get as Estate Owner in Linden Lab’s Second Life grid).

No, the biggest issue with OpenSimulator is… lack of people.

This is naturally common to all so-called social virtual worlds, and the reason why every time I predict that a new company entering this interesting market will fail, my predictions usually are right. In fact, I predicted that even Linden Lab’s second virtual world — the much more sophisticated Samsar — would, ultimately, fail; and so it did. Not even Philip Rosedale, co-inventor and co-founder of Second Life, was able to keep his own “new” world afloat, High Fidelity (his company turned to a different market). The reason for the very-long-time “survival” of Second Life is simply because it has a critical mass of content developers, which, in turn, means a critical mass of content. No other platform has ever come close to the sheer amount of content that exists in Second Life — and while “having content” is not necessarily the only reason for SL’s ongoing success (because the company running the virtual world also needs to have a valid business model — one that guarantees them a long-lasting cashflow; other business models, so far, have all failed, no matter how interesting and promising they sounded, such as Philip Rosedale’s own High Fidelity virtual world, which was “free” in many senses of the word, but the company would get a tiny fee on the monetary transactions made through the system — which they hoped to be enough to pay for their ongoing development costs; it wasn’t), as it turns out, is definitely a big issue.

Perhaps not surprisingly, the second-largest social virtual world out there is… well, OpenSimulator as a whole, even if it is not technically a virtual world, but rather “islands on the net”, many of which can be bridged via hyperjump (but not all!). The main reasons are that OpenSimulator attracts the same kind of people that are keen loyalists of Second Life — including content creators — and the main mode of interacting with the virtual world is exactly the same (using the same viewer). Thus, there is no need for “adapting” to a completely different way to interact with the virtual world, using new viewing software, learning a new interface, etc. and so forth. Whatever skills one has in Second Life will be used precisely in the same way in OpenSimulator. But another main reason is that the business model of those commercial OpenSimulator grid operators is exactly the same as Linden Lab’s — at a much smaller scale, of course. The business model of land ownership is a battle-tested model; it has proven to survive for almost two decades, and we know all the pitfalls (but naturally also the advantages!) of such a model.

Most importantly, from the perspective of the user, moving from Second Life to OpenSimulator is not really an effort — with the sole exception that you will have to abandon all content you owned in Second Life. The rest is basically “the same”. Even the argument that there is more or less lag in OpenSimulator, or that the regions look horrible… well, that’s true for Second Life as well, no two regions are at the same level of content quality and simulator performance. There are many reasons for that, some that are more or less under the control of the region owners, some that are more or less under the control of Linden Lab (or, for OpenSimulator, the commercial grid operator), but there are also many reasons where you can basically do nothing about it. This is true for both platforms: the difference is that in Second Life, being much bigger (by at least one order of magnitude — even if some statistics show that the total land area in OpenSimulator grids if all were connected together, rivals Second Life, the truth is that there are just a small fraction of residents…), and therefore hosting way more content, has far more choices. Because OpenSimulator grids are seen by professional 3D content creators as “less secure” (this was true in 2007… as it was true on Second Life around 2003, before we even had a permissions system!), many are very reluctant to face the Brave New World(s) out there, simply because they are not confident that their content cannot be freely copied without payment. Strictly speaking, this is not true — although, naturally, grids are prone to hacking, especially on those where the grid moderators/maintainers might not be so knowledgeable as Linden Lab’s own system/network administrators, and thus leave their grids less protected from server-level hacking attacks — but, as far as I know, this is a rare event. In fact, it’s much more “myth” than “fact”. But it is true that early OpenSimulator grids accrued some bad reputation precisely because they weren’t well managed and prone to attacks — a few of which did, indeed, succeed (bringing grids down, stealing in-world money, getting access to user data, and so forth). Alas — it’s tough to live at the frontier!

But the years have passed, the OpenSimulator software is more robust to overall attacks, it is far more stable, and grid operators are more serious about their business ventures. This did, in fact, attract some content creators — perhaps not the “top tier” creators of Second Life, but a few from the tier below the top, who found a community of users eager to buy high-quality content, just like in Second Life, but having a much harder time to get it…

Now, it’s worth pointing out that all the above is of little relevance to some companies and especially to most university research teams. For them, “nice content” is a bonus, not a requirement. They have special purposes in mind, some of which require a much deeper level of access that Linden Lab does not permit on their Second Life grid, but which is fully accessible under OpenSimulator — you can simply change whatever code you wish at the simulator level, recompile it, and enjoy the results. The amount of flexibility — the mere ability to develop at the core of the simulator, and not merely on an abstract layer provided by scripts written in Linden Scripting Language (LSL) and running on virtual machines, is an advantage that OpenSimulator provides “out of the box”, so to speak, since it was designed to work that way. OpenSimulator is also quite modular — you can add or remove components as well as create your own modules, and stay away from the core code. And, of course, having full access to the simulator, it’s far easier to integrate it with whatever technology the company/university/institution has in mind. LSL gives quite a lot of possibilities (and people are creating amazing things with what it can do), but it only goes that far; to get the extra edge, you need to dig deeper in the simulator code. And possibly even on the viewer, too…

What these companies and institutions are presently doing is, of course, unknown. We can speculate based what they are willing to share, once in a while, sometimes giving interviews to the above-mentioned Hypergrid Business — if they happen to know about its existence and have any interest in sharing their projects with the community. This might be the case with some universities — and these sometimes even have “visitor areas” on their grids which can be accessed via a Hypergrid jump — but it is safe to assume that most are “closed”, they’re “hidden projects”, with a specific timeline and a restricted audience, and possibly we will only learn about those once they have finished and reports are published. Projects, in the institutional and corporative world, have a beginning and an end.

Naturally enough, those kinds of projects may not use OpenSimulator at all. There are plenty of alternatives out there, from more obscure solutions to the ubiquitous Unreal Engine, or the commercially very successful Unity development platform. These require far more programming and high-end content creators, but, naturally enough, the results can be exactly what you’ve got in mind. OpenSimulator is, compared to those, the most successful low-end “tinkerer” platform for developing virtual worlds: true, it requires some knowledge to set up, but it will work “out of the box”, ready to start working on it, even if all you have is a handful of items in your inventory and a flat, green, 256x256m bit of a world to walk around in solitude.

But compare it to the alternatives: you start with a ready-to-build environment, where you only need to add content. You have a viewer — which you did not need to develop on your own — ready to display anything you create. You have several tools available, all already included in the viewer, which allows you to immediately place content in-world. That means that a new project can start in the morning, struggling through configuration parameters, while in the afternoon you can already walk around your freshly installed virtual world and start your project by adding interactive content — even if the first attempts might be crude, and visually/æstethically unsatisfactory, the truth is that they will work.

Contrast that with weeks, nay, months of development by a team of professional programmers working on a popular engine/development platform. Months until you can finally launch an application and “enter” your virtual world. And if it’s not what you had in mind… there will be more months of programming ahead. Sure, the results may be visually far more impressive than anything built on Second Life or OpenSimulator. But the trade-off is the very lengthy and complex development time, which takes ages to complete until you can finally interact with the most basic items.

Second Life and OpenSimulator give you that sense of instant gratification, allowing you to experiment with the virtual world environment immediately — and do it on your own. You don’t need a whole staff of professional programmers, proficient in developing virtual worlds for gaming engines; you don’t need a group of high-end, professional 3D content creators, that have the kind of know-how required by the AAA league of console/computerhttps://www.thoughtco.com/what-is-aaa-game-1393920 games. No, you only need yourself, a bit of curiosity, a bit of imagination, a bit of time to learn a new tool — and that’s it, your project has started and is instantly available for your first experiments. And while the environment will not look like a AAA console game or a flight simulator for training commercial airline pilots, the basic concepts will all be there.

No other existing platform can provide that kind of results from the very first day you install it. And that’s why Second Life or OpenSimulator are still being used for these kinds of projects.

Oh, and as a bonus, you immediately get access to a ton of content — if you need it — as well as a vast community of hundreds of thousands of users, all in the same virtual environment. Again, no other 3D development platform can offer the same. And while it’s true that most of these institutional projects are never meant to be seen or visited by the public at large, it’s still nice to know that you can show them off to a considerable audience by merely tipping a checkbox (or possibly adding a configuration line). All your potential audience will use the same tools, the same viewer, see the same environment, and all of that comes straight out of the box. Contrast that with the way most virtual worlds are delivered today — as full-blown applications, to be downloaded and installed, each with its own peculiar interface.

There is still a lot that Second Life and/or OpenSimulator have to offer that nothing else can.

  1. Note from the author: This article was originally written in 2010 — but I left it unfinished back then. There was really not much to add, but some references needed to be edited, and there were a few paragraphs missing, but I knew what I wanted to write back then. It feels weird to finish it over a decade later, in the middle of a pandemic; but several of the “truths” stated above still hold today (if not all!), so I finished it and decided to publish it. Take everything with a pinch of salt: some things might not be 100% true any more or are simply oversimplified and overstated in the text. But Second Life is most definitely still around and thriving; OpenSimulator grids may come and go, but a few have been around for far more than a decade, with a steady stream of users (and revenue for their grid operators!), and OpenSimulator development is ongoing — it hasn’t stopped, and it has continued to follow its goal to keep a decent amount of compatibility with Second Life (adding new things as Linden Lab develops them) while providing some extra functionality that Linden Lab simply cannot allow on their own platform. So, in a sense, the article essentially continues to apply today, even though I suspect that there are far fewer institutional projects being done in 3D environments these days — but they are not zero. ↩︎