The notion of the negative slope algorithm was introduced
by S. Ferenczi, C. Holton, and L. Zamboni as an induction
process of three interval exchange transformations. Then S.
Ferenczi and L.F.C. da Rocha gave the explicit form of its
absolutely continuous invariant measure and showed that it
is ergodic. In this paper we prove that the negative slope
algorithm with the absolutely continuous invariant measure
is weak Bernoulli. We also show that this measure is derived
as a marginal distribution of an invariant measure for a 4-dimensional
(natural) extension of the negative slope algorithm. We also
calculate its entropy by Rohlin's formula.