One of the things that environments that use the Second Life family of virtual world viewers are often criticised for is poor graphics, compared to the latest games (or even games from 5 or 6 years ago).  Usually this is defended in several ways - that the content is user generated, so isn't optimsed the way that game art is, that just dropping into a random space in a grid like Second Life or Reaction Grid isn't a fair test since again it's user generated, and there are beautiful and breath-taking places if you know where to look, that the graphics artefacts have to be streamed and aren't supplied preloaded on a DVD and so on.  All of this is true.   It is also true that the graphical environments rendered by games are typically richer and more detailed than even the best Second Life and OpenSim environments, if you have the necessary hardware in the form of a modern console or graphics card.  So it might seem natural to respond by striving to improve the graphical quality, to compete.  The futility of this approach was recently highlighted starkly during an 'office hours' discussion from a staffer at Linden Labs, the makers of Second Life.  They posed the question "what percentage of residents (users) would you all guess have 'class 0' (basic Intel motherboard) graphics hardware?". Various guesses were proffered before the answer gathered from Lindens own connection data was revealed - a massive 60%.  This is across the board as a percentage of all users.  If you are in a more constrained environment, such as a corporate office environment, my experience is the figure would be substantially higher. 

So aiming for graphical richness will have the effect of creating wonderful eye-candy that an ever decreasing set of people can access.  The Blue Mars virtual world is a case in point - based on the Crysis graphics engine, it requires high performance hardware to participate.  This approach also sets a precedent for constantly trying to keep up with the latest graphics developments which in turn keeps the potential audience low.  In an ideal world the graphics would degrade gracefully across all platforms, giving a cutting edge look for the best hardware but still run on the low end.  The challenge here is in the content generation, since to do that would require users producing content at a variety of levels of detail (this will be required  to some extent in the forthcoming 'mesh imports').

Now I am aware that I'm ignoring one important sector of virtual world users - and it's one of the few strong growth areas of Second Life - it's use as a platform for machinima. I think the approach here should be the one that is already being taken - most people serious about creating machinima in SL use a customised viewer with improved lighting and shadows that needs some serious computing horsepower to run effectively. In any case, if the ambition is to increase the number of users of virtual worlds significantly and make such environments a realistic, desirable choice for collaborating and social interaction, we shouldn't be catering specifically for a narrow minority (leaving aside the undeniably positive effect that great looking machinima clearly have). But all things considered, that 60% is timely reminder that to produce social or collaborative environments we should be targeting the mainstream and not the elite if we want to succeed.  Graphics alone do not make an environment engaging or immersive, something Nintendo understood very well with the Wii and DS.  Content and ease of interaction are the key.

(This article is based on one I wrote originally for Flying Island)