241a242,244
> The concept of {\em competitiveness\/} has several closely related
> definitions.
> 
250a254,280
> 
> An {\em algorithm\/} can be $C$-compeititive, but a {\em problem\/} can also
> be $C$-competitive.
> We say that the online problem
> $\calP$ is $C$-competitive if there exists a $C$-competitive
> online algorithm for $\calP$.
> 
> Note that if $\calA$ is $C$-competitive, then $\calA$ is also $C'$-competitive
> for every $C'>C$.  (Here we must use the assumption that $\cost$ is
> always non-negative.)  We define the {\em competitiveness\/} of an
> online algorithm $\calA$ to be the minimum value of $C$ for which
> $\calA$ is $C$-competitive.
> Similarly, we define the competitiveness of an online problem $\calP$
> to be the minimum value of $C$ for which $\calP$ is $C$-competitive.
> 
> \subsection{Competitiveness of Randomized Algorithms}
> 
> Suppose $\calP$ is an online problem.  We can define a {\em randomized\/}
> online algorithm for $\calP$ is one of three ways.
> 
> \begin[enumerate}
> \item
> As a distribution on the set of all deterministic online algorithms
> for $\calP$.
> \item
> 
> \end[enumerate}
