Poster 'Magnuz' is totally correct. This is in fact a very specific, narrow scope *implementational* patent.
The assigned inventor (who actually was one of the first people working in VR - totally legit) had come up with a cool way to implement a VR such that it didn't need to be restarted when new objects were added, which apparently was a problem in 1997. The idea of using machine code blocks as corresponding to individual 'virtually physical' objects is an interesting idea in and of itself. It's actually a pretty good patent.
However, this is all academic, as modern systems have not been implemented anything like this invention. The "doctrine of equivalents" will not remotely extend the claims to cover current systems. In current systems 'virtually physical' objects are represented as data (with perhaps a smidgen of uncompiled scripting code), and there *is no problem to begin with* of having to restart simulation instances to add new items.
In short, this is a like patent for an assembly of cogs which the portfolio owner is trying to pass off as a patent for an assembly of Lego bricks.
Not good ammunition for trolling with.