News for hackers. Stuff that matters.
Re:The failure of Second Life by ChuckMcM on Monday May 23, 2011 @01:23AM (#2575050)
As others have pointed out, the article's title is linkbait. The actual article says:

1) Second life lets people design levels.

2) This gets very expensive in bandwidth a high draw levels.

3) This makes it 'unplayable.' (aka a failure)

Clearly its not the case that its unplayable since a lot of people play it. Also the geometry / texture issue is huge when moving from place to place, but manageable when you're in one area. Thirdly the big draw for second life seems to be weird sex acts which are probably mostly mental anyway.

It is one of the few attempts at creating a Metaverse. As the author points out there is a tremendous asset management problem if you try to stream everything from the server. It could be improved.

Given modern disk drives I'm sure you could cache locally pretty much everywhere you went with some lookaside cache code to invalidate local copies when meta-data indicated an update. The author alludes to something like that by asking for a better architecture.

Now it would have been a bit more interesting if it was written from the perspective of "this is what it does well, this is what doesn't work for me" and then go on to speculate or do the math on how one might improve it.

I was looking that the NoSQL dB space as a metaverse management plane. It could work although everyones view of the world would only converge, not instantly be the same. You could use something like MongoDB, it would be read-mostly for geometry. Back when I designed a castle building MMO (never shipped sadly) the textures were fixed (all clients had them) but you could paint them on different geometries. It reduced the bandwidth requirements to sending around vertices and a texture vector (id, origin, u and v vectors). It would be fun to code something up like that with the Unreal engine.

There are no comments on this comment yet.