$n$-fold complete Segal spaces are one model for $(\infty,n)$-categories; there are other models. More precisely, they are supposed to be a model for weak $(\infty,n)$-categories.
The distinction that I think you are asking about is between weak and strict. Strict $n$-categories can be easily defined by a recursive definition: a strict $n$-category is just a category enriched over strict $(n-1)$-categories. A strict 1-category is just a plain-old category. Though easy to define, strict $n$-categories don't seem to capture the things people want an $n$-category to capture.
One such feature is that strict $n$-categories don't satisfy the "homotopy hypothesis", which says that an $n$-groupoid ($=n$-category in which all morphisms are in some sense invertible) should model homotopy $n$-types ($=$ spaces whose homotopy groups vanish above dimension $n$).
In fact, this failure only occurs for $n \geq 3$; I believe this is the type of bad behavior Lurie refers to. Another failure of strict $n$-categories happens when you try to talk about higher monoidal structures.
If you haven't already, take a look at the papers by Baez-Dolan on arxiv, which discuss a lot of these issues.
