I receive this question a lot when I talk about the Ogoglio project, so here's my linkslappable answer.
As detailed on the project site, the primary goal of the Ogoglio project is to build Ogoglio City, a web based urban space for creative collaboration. When I started working towards this goal I surveyed the field with the idea of extending an existing platform. Second Life looked promising, so I read the Second Life fora, learned to build and script, attended events, subscribed to the blogs, and talked to people about their experiences using the service.
Second Life is interesting and I log in regularly, but it is not a good host for Ogoglio City for the following reasons:
Second Life is a company town.* It deals in company issued currency, it controls all of the natural resources, it defines what is structurally possible, and it defines what is socially acceptable.
The Ogoglio platform must be open and it must be a mesh of third party servers. As Flickr is different than fotolog, districts of Ogoglio City will be different from each other. There are many choices to be made in terms of identity, economy, security, copyright, and support (to name just a few). It seems unrealistic to force the entire world into one set of decisions. This design brings as many warts as beauty marks, but it is broken in the same ways as the web, which makes sense for a web based city.
Second Life cannot be open sourced. Core pieces of the infrastructure (e.g. Havok) are not owned by Linden Lab, and it seems unlikely that they can rewrite them without a major change in company culture. If they do open source the servers, it will be like the swiss cheese code that Netscape released as the first Mozilla, missing core functionaly. In the end, the Mozilla project rewrote the whole browser from scratch and kept almost nothing but the name.
Can you imagine a web without the Apache server? I hate to bring up VRML, but it is a great example of how a technology was crippled by business models which depended on proprietary code but ignored user experience. Having an open source license is obviously not sufficient to ensure a viable technology or business model, but for some projects it is necessary.
Second Life is not "of the web". Technology problems like this could disappear given enough time and a healthy development process at Linden Lab, but right now applications in Second Life are not built and managed using the same toolchains or patterns as those in use by millions of web developers.
The world is learning how to deploy and manage large scale web applications, and I see no reason to reinvent the wheel (e.g. with a custom scripting language like LSL) when so much work is done to make free and great tools for the web.
Second Life is not "on the web" Linden Lab is focused on building and deploying a rich stand-alone desktop client, but this means that to get into Second Life people must be able to install applications, upgrade them every week or so, and they must be willing to devote a large portion of their screen, RAM and cycles to being in Second Life.
I've written a bit about the ideal technologies for the Ogoglio client, but the basics are that it sits inside people's browsers, it usually doesn't require an installation or manual upgrade, and it is possible to shift between active participation and passive awareness applications.
That said, I readily acknowledge that Second Life has one major characteristic, existence, which makes it better than the platform I've laid out for the Ogoglio project. It is much easier to pick something to pieces than to create, and though I'm busy creating and recruiting creators I respect Linden Lab for creating the closest thing to a metaverse this world has seen.
