The application of D-brane methods to large black holes whose Schwarzs
child radius is larger than the compactification scale is problematic.
Callan and Maldacena have suggested that despite apparent problems of
strong interactions when the number of branes becomes large, the open
string degrees of freedom may remain very dilute due to the growth of
the horizon area which they claim grows more rapidly than the average
number of open strings. Such a picture of a dilute weakly coupled str
ing system conflicts with the picture of a dense string soup that satu
rates the bound of one string per Planck area. A more careful analysis
shows that Callan and Maldacena were not fully consistent in their es
timates. Tn the form that their model was studied it can not be used t
o extrapolate to large mass without being in conflict with the Hawking
-Bekenstein entropy formula. A somewhat modified model can reproduce t
he correct entropy formula. In this ''improved model'' the number of s
tring bits on the horizon scales like the entropy in agreement with ea
rlier speculations of Susskind.