How 3D Should Second Life Be?

Last week I wrote that a graphically upgraded Second Life-- say one with dynamic shadow-- might cause a "balkanization" of the metaverse experience.  That caused some outcry from several Linden staffers in Comments, so I should explain further.  I can summarize my concern with this fairly defensible statement:

There's little evidence of mass demand for an intensely immersive 3D virtual world;  instead, indications suggest the market shrinks in inverse proportion to increasing immersiveness.

Why do I say that?  Look at the hard numbers:

Top virtual worlds/MMOs by use, as of 2008:
- World of Warcraft, 10 million subscribers
- Habbo Hotel, 8 million monthly active users
- RuneScape, 5 million monthly active users
- Club Penguin, 4 million monthly active users
- Webkinz, 4 million monthly active users

(Source, with recent updates)

There's several worthwhile observations you can make from this list.  First, none of them feature next gen, top-of-the-line 3D graphics.  (WoW is 3D, but developed with graphics that run fairly well on older computers; also, the visuals are not realistic.)  Besides Warcraft, however, none of these top MMOs are 3D at all; rather, they're 2.5D.  And while one hopes that 2.5D-based MMOs will whet the market's interest in a more immersive, graphically rich virtual world, the exact opposite seems to be the case.  (The still-popular Habbo Hotel was launched in 2000, and the cartoonish graphics are basically the same.)

Only after you drop down several million users do you start to see MMOs and virtual worlds incorporating next gen graphics that require high-end 3D cards for optimal viewing-- Lord of the Rings Online at about one million subscribers, Age of Conan at about 750,000 subscribers... and Second Life at some 550,000 monthly active users.

Why is this happening?  Here we enter the realm of speculation, but it seems that most people experience sensory overload with too much immersion; instead of being drawn into the intensity of the simulation, they're repelled by it.  Gender may be a factor, and that's particularly worrying.  A Georgia Institute of Technology study suggests that the overwhelming majority of girls far prefer 2D games over 3D games.  (Forget about technical considerations for a more important concern: If Second Life gets more and more 3D, will women begin to leave it?)

Whatever the reasons, the numbers are hard to dispute.  The reigning virtual worlds are simply not photorealistic, and in terms of maximum popularity, there's no indication photorealism is a selling point.  Sales of the Xbox 360 and PS3, next generation videogame consoles which boast high-end 3D graphics, are waning, far eclipsed by the Wii.  Torley Linden notes that it's entirely possible to run WindLight-driven Second Life on a relatively cheap PC, but far as the market is concerned, technical specs seem beside the point.  "[S]imilar concerns were raised about BioShock," he writes, "which has higher min. requirements than SL."  That's precisely the problem.  BioShock is a masterpiece, one of the greatest games ever made, one that everyone should play-- but the painful fact is only around 400,000 people bought the PC version.

Linden Lab's mission statement is, “To connect everyone to an online world that improves the human condition.”  I suspect all who read this blog supports that sentiment.  But if the Lindens truly believe their own words, the question becomes: what technical improvements actually make it more difficult for most of everyone to connect?

Image credit: Katharine Berry, from her Flickr stream.

Comments

[Argent Bury]

For me, SL is not appealing because of its verisimilitude. It's appealing because it's just real enough to grab me, but it lets my imagination fill in the blanks.

Seems like this debate is just another symptom of differences in what people want out of the platform. Some want to push the hardware to its limits because they can, some want a better model of reality, some want things to stay pretty much as they are.

For now we all seem to be able to get what we want at the same time, but if some of the optional things we are seeing start turning mandatory there will be a divide, as you say.

[anomouse]

I LOVE the shadows but wow do you need a GREAT video card to make it work. I will upgrade my card but I hope they dont make shadows manditory this year. If LL let people know that by next year they would have shadows it will give time to save some green to get the hardware updates.

[qarl]

hey Hamlet - would you be up for an informal (very unofficial) in-world talk? my office hours this friday?

[Tateru Nino]

There's already an enormous separation in the experience of users, based on their graphical hardware and PC capabilities. Try wandering around for a year with your draw-distance set to 64 (I spent my first 2.5 years in SL like that). Everything *feels* very different.

Likewise lighting isn't available to some, or windlight skies, or anything other than untextured water. I didn't feel particularly disadvantaged by the lack, but it *does* mean that you perceive and you *feel* the virtual world in very different ways.

[dandellion Kimban]

Immersion is not dependable on photorealistic graphics. It happens by the imagination, not the visual illusion. Most of us know that immersion can be very deep on the sims that are deliberately cartoonish. And we all felt that something is missing in perfectly textured photoreal sims.

It is not just virtual worlds and games. If we take a brief look at the history of art, we'll see that realism is just a short episode. most of the times, visual was communicated not by copying but by abstracting the idea behind it.

[Sophrosyne Stenvaag]

Something that far too many engineers and business people fail to understand is that virtual worlds draw people for two key reasons: they appeal to the imagination, and they provide an alternative to the atomic world.

Efforts to make virtual worlds more like the atomic world logically make those worlds less appealing: if they don't offer a real difference, but have ever-higher barriers to entry, their market share will decline.

SL's success was built on creative empowerment; increased hardware demands are disempowering. My year-old top of the line non-gaming laptop can't run the release candidate viewers, barely runs the main client, and is sort of adequate for running the OnRez viewer.

If you tracked my time inworld, it's directly inversely proportional to "improvements" in the viewer. I'm spending far more time in WoW these days, because WoW still runs on my hardware.

If my life weren't in SL - my family, my home, my business - I'd have given up and left sometime back in February. The technical problems of the world itself, and the excessive demands on my hardware, just wouldn't be worth it.

But, it seems LL continues to be dominated by just those groups that understand the casual user least - engineers and corporate-focused businesspeople. With every increase in unwanted features and every ham-handed customer relations failure, they reduce their own user base significantly.

[dandellion Kimban]

... and to connect my comment with Tateru's (since they were typed at the same time)....
A very dear friend of mine has her settings on the low end (no reflections, short drawing distance, etc...). At the same time mine goes with all the reflections and sometimes draw all 512 m (if I am not moving too much and the sim looks good enough and coherent for that). But, no matter that our snapshots are very different, if we are standing together, we do share the same space.

[Marianne McCann]

For me, the issue of how immersive the shadows will be is, at the time, irrelevant. I run a 2 month old macintosh, and therefore am not up to date enough to run it.

I personally would love to have shadows. But it seems clear that this is going to be aimed at a very small number of "power users," not even SL regulars like me -- and certainly not the bulk of SLs residents.

[Veeyawn Spoonhammer]

For me it goes back to content creation. The 2.5D worlds are great but I can't create as freely as I can in SL. I can't whip the camera around instantly to check out the other side of my creation. I know that a mass audience may not be interested in creating anything but the most immersive 3D world is the one I want.

[Ignatius Onomatopoeia]

I'm doing a lot of in-world racing lately. Finally found a "game" in this "world" that captivates the aging RL hot-rodder in me.

I just hope that LL will let those of us running older hardware continue to be able to "Drop back" when fancy graphics are less important than, say, being able to render a pack of cars fast enough so I don't run into them.

At times I'll want to ooh and ahhh over the scenery. At other times I'll just want to "stand on it" and win the race.

[Ann Otoole]

More than 6 local lights, shadows, windlight, glow, voice facial animations... all great for machinima. less than 30 FPS in the captured video? not so good. So imagine for a moment what all is needed to boost the FPS in world to somewhere around 70. Once we get to all these features plus enough FPS to have 30 FPS in the captured video then look out because SL will have become a very serious filming medium.

So yea LL needs to keep pushing technology. There is a lot more work to be done. Sadly tis tru that even with "windlight turned off" the computer is working harder than before. Thuse why it is nice for viewers like the old school Nicholaz viewer are there for optional use. I would prefer LL maintain a light client for people with less powerful systems to use while enabling a serious cinematic experience for those of us that have an interest in serious filming and cinematic game play.

[Ordinal Malaprop]

I vaccilate continually between the 2D/3D - actually, Second Life has a number of different independent dimensions, sound and so on as well, it is a little tricky to produce an exhaustive technical specification but let us say "3D" for the nonce - and the 1D, in other words, a purely linear textual feed, created via Inform. I have to say that if a MUSH version of Inform was ever developed and became at all used, I would be in SL far less often.

Visual effects are wonderful things to have, but let us not imagine that they have all that much influence. An alternative service that looks a bit ugly but offers more compelling content will always win. Consider "myspace". And that is not even considering the minimum requirements necessary to enter a world and receive reasonable performance (64m draw + 5fps is not "reasonable").

[Nicholaz Beresford]

With those viewers I build I'm constantly going forward and backward between the versions.

The interesting thing is that while my computer can run WL decently, I'm mostly starting up older versions.

It is very hard to explain (those above have mentioned imagination and that may play a part), but the older viewers have a consistent visual experience.

Mental compartmentalization may come to play also. As long as the 3D worlds are unrealistic, they live in their own mental compartment, being judged by their own category.

The more realistic they get, the more they will be compared to real life experience ... where they will ultimately fail, because you can never mimick the subtle delicacy of a real experience.

It's just a theory of mine, but I think the better the graphics get, the more people will be dissatisfied with them.

[Nicholaz Beresford]

Speaking of the consistency, one example are the trees or ground. In the old viewers they don't stick out as bad, because they sort of match the visuals of everything else (like he cartoonish sky).

In WL however, they simply look ugly and out of place.

[Two Worlds]

Hamlet, quit being ignorant. It's not "immersion overload", we're not in the Matrix and you're not Morpheus. Try a few more common sense theories like "more popular and well-known", "free vs. subscription", or "ease of learning the game". I'm sure pretty much every British teenager who plays Habbo Hotel has figured this out long ago.

Seriously, you get paid to do this?

[Olivier]

I think there is no problem if there are 2 viewers:
-1 with full graphical fonctionalities
-1 light viewers for slower computers

Let's get the best of those 2!

I think that the more realistic SL will be the more real SL experience will be, but reading all those comments it's seems obvious that SL need to maintain a light viewer in the same time.

[Nexii Malthus]

The problem is that we need a truly proper client setup built on a flexible modular open source foundation. Basically we need a viewer based on libsecondlife for networking and the graphics that any old computer could run. I just wish there were a group as dedicated as to what the libsecondlife effort was on making a client foundation.

[Patchouli Woollahra]

Two Worlds:
=truckload scathing barbs=
Seriously, you get paid to do this?

Yes, he does.

Seriously, YOU get paid to do THIS?

[Spatha Spatula]

I'd be interested to learn what definition was used for "virtual worlds/MMO". The worlds listed are quite different from each other, and in the type of user they attract. Because of this, it’s possible to draw conclusions that may not apply to the genre that includes Second Life. You can also draw wrong conclusions from such a mixed group. For example, with this list you might also think that "virtual worlds/MMO" primarily appeal to people younger than 19.

As for the contention that “the overwhelming majority of girls far prefer 2D games over 3D games”, might that reflect a lack of interest in shooters, which are often 3D? On the other hand The Sims, which is 3D, is one of the most popular and a favorite of women and girls.

[Markus Breuer (Pham Neutra)]

Funny, how many comments a post like this generates. :) You seem to have hit a sweet (or sore?) spot, Hamlet.

I can't follow your line of reasoning though. While I won't argue your numbers (showing that a lot of 2.5D worlds are vastly more popular than next gen 3D ones), it is tough to draw conclusions from such a fact. When two factors are correlated this just means that they are ... correlated, NOT that one is the result of the other in a cause-and-effect relationship. :) Assuming cause-and-effect when there is correlation is one of the most popular mistakes in logic.

There are many reasons for the popularity of 2.5D worlds. And the most important one is probably their easy availability and (often in the browser) and their low hardware demands (nearly anything goes). I very much doubt that people will run away from Habbo or Penguin as soon as these games improve the visual fidelity. :)

They will run away (or not come anymore), though, if the makers of these platforms DEMAND certain hardware standards (or compulsory training sessions to learn a complicated user interface). Second Life's hardware/software demands and steep learning curve are already the most important barrier between the platform and a more widespread adoption. It would be a dangerous decision to raise this barrier - even if the engineers in the development department would love to have better graphics so they won't have to be embarrassed anymore when having a drink with their pals from the gaming companies.

Visual fidelity (not necessarily realism) is great - but certainly not necessary for most applications of virtual worlds.

I am not afraid of balcanisation, though. The user experience of Second Life already IS balcanized (as Tateru points out). In the long run we will a huge Metaverse, which is rendered in different qualities on different devices and clients. And it probably won't hurt people too much, if they have to live without shadows and reflective water on their iPhones, even when their friends working from a high end PC at home see a photorealistic world.

[Connie Sec]

Hmm, well, I'm all for improvements in visual representation, but perhaps we can look on this another way. If, it seems, that constant graphical improvements do not increase users, and the "ease of use" is the main point that holds people, perhaps this allows Linden Labs "breathing space" to concentrate on real problems, like its database, (making it perhaps more distributed), scalability,stability and ease of use.

[Laetizia Coronet]

I'm getting sooooo tired of this thing. Windlight, shadows, shiny water, reflections... to hell with all that, sorry. Get the damn thing to work and do *not* create two-tier viewers. How the hell is any serious (as in non-flogging-crap like so many) content creator going to create content if he hasn't got the slightest idea about how others may see his product?
See, I can make a website that looks reasonably good on any computer you visit it with, even with limited use of fancy stuff. Heck I can make it on a 386 if need be. But I can no longer make a simple shirt because of all the effects some may or may not have running. It could look like crap when my artificial shadows clash with the real ones.
If you want to be the next gen Internet, you want to be a broad platform, not a narrow high tech one. Stuff yer shadows.

[T_S_Kimball]

I find it ironic that my two most engaging - perhaps even addicting - multiplayer online games are so far apart in terms of technical quality.

1) Uru Live (aka Myst Online) was arguably one of the most visually engaging MMOs, yet it could not fully survive in a normal MMO format given two chances. Niche market? Perhaps. But it also had large barriers to entry and an interface that was sorely lacking (sound familiar?).

2) Travian. A browser-run real-time MMO in an RTS style (i.e. resources, cities, armies, etc). As long as you had a browser that had Javascript enabled, you could play. It had much of the intrigue and subtleties of Eve, for example, without needing anything fancy (though having tools for massive collaborations is important later in the game). Games of this style probably have a world-wide player base that can rival WoW.

If LL intends to continue with their vision, they must consider who their market is. Unfortunately for them, its not the enthusiast or midrange markets.

Please, think about it, since I don't at all understand why an nVidia 7800 GT works perfectly fine on 1.19 and totally sucks on 1.20. It took an 8800 Ultra (WTF??!?) to finally get the same standard of viewing on 1.20 that I get today on 1.19 (both with and without Nich's EC-f version).

As a parting note, I also think that shadows should be dropped; It was a major cause of client-side issues on Uru from day 1.

I'm tempted to visit those office hours on Friday, if just to see what will happen. That is, of course, if I can actually login in the first place. :p

--TSK

[Pavig Lok]

I'm all for shadows... one day... but I think the client needs to scale both ways. The thing many of these other worlds have going for them is they're light and engaging, rather than all encompasing. Different types of immersion work for different activities.

If you want to run for your life there's halflife and bioshock. If you wish to live in an endless mindless pachinko dimension there's peggle or zuma or popcap games. If you want to live in a totally conversational reality, there's irc (or Mis Ordinal, perhaps investigate the mushes muses and moo's around these days, which have inform like languages, should you wish to build in text.) All of these are perfectly suited to the immersion they are designed to provide.

SL sits in the middle of all of these, as a general purpose vw - it will one day need shadows and super-realistic rendering in some contexts. It'll also one day need to acknowledge that for some, it is the worlds fattest IRC client. The experience needs to scale down as well as up.

At the moment I'm optimizing builds for atmospheric shaders OFF, as for the foreseeable future that's going to be the most likely way business, academic, and laptop toting folk are going to view them. I love looking at the eye candy on my quad core 64 bit high end gpu beast but I started building in sl on a machine i picked up on the curb - and that was a year ago. Half the non gamers I know see sl about the same quality ON MODERN MACHINES as I did on a box some gamer left on the street.

I don't think there's an inverse relationship between photorealism and immersion as Hamlet hints, but deffinitely an inverse relationship between photorealism and accessability. We must remember that most people's pc's are for browsing the web and doing office stuff, and the gpu monster they buy is a console for the living room. Not all pc's are designed to be gaming boxes. Increasingly they're viewed as data appliances and scaling down in spec rather than up. Given that - it's unlikely photorealism will be a standard in non-gamer virtual worlds for some time, even if the capabilitiy exists for some.

Just my two cents...

[Two Worlds]

I agree with Laetizia. First of all, consider how much downtime Second Life has--whether accidental or planned. I would guess that ANY platform with that much downtime--whether it be an online application, website, or online game--would by now be experiencing a mass exodus of users. I don't play WoW so I'm not sure of this, but I'm guessing World of Warcraft's downtime isn't near as bad as what we're used to--and consider the fact that people actually run businesses in Second Life (not the multi-millions as you'd like to believe, true, but maybe just enough for beer money every now and then).

And as people here have said previously, Second Life should have capabilities for a BROAD amount of users, not a NARROW one, with only the most cutting-edge graphics cards.