I was motivated by the Centering Theory (Grosz et al. 95).   I experimented with the Recency Constraint (Hobbes 78) and the S-List algorithms (Strube 98) by hand on a small corpus, taking into consideration anaphoric noun phrases as well as pronouns.  Due to the size of the corpus, statistical results were not telling.  But hand-tagging the corpus allowed me to have greater insights into the interaction between the algorithms and the corpus.  Most of the results are therefore by observation from the actual application of the algorithms on the corpus – which though small, contains some of the most contemporary writing styles in the English Language.  Careful measures were taken ensure the amount of world knowledge is under control.  Observational results included (a) the necessity of scaling up the algorithms to a global level, enabling it to handle anaphoric noun phrases; (b) a possible discrepancy in the performance of algorithms when applied to text with different writing styles; and (c) an improved performance when combining the Recency Constraint and the Centering Model, suggesting greater improvements if more algorithms are combined.  The proposals that I presented imitates how human perform the same task.