1: \begin{abstract}  

2: We start from recently published numerical data by Hatano and

3: Gubernatis~\cite{HATGUB} to discuss properties of convergence to

4: equilibrium of optimized Monte Carlo methods (bivariate multi

5: canonical and parallel tempering).  We show that these data are not

6: thermalized, and they lead to an erroneous physical picture.  We shed

7: some light on why the bivariate multi canonical Monte Carlo method can

8: fail.

9: \end{abstract}

10: