Erik Biever informed me that the previous RPI formula was
.2R0+.4R1+.4R2.  This means that if the R2 term is ignored that the
multiplier of R1 is twice that of R0 -- the same as it is now.  Thus my
thought that the current formula would increase the desirability of
playing games against good teams was flawed.  It sounded good when I
was thinking about it yesterday. :-)   It is still the case that
increasing the weight of the R1 term with respect to the R0 term should
have that effect, and I suspect that since a more mathematically derivd
RPI would have the R1 multiplier smaller than the R0 multiplier that
either formaulation has this effect.
 
On a related topic, Dick Tuthill stated earlier today that the the RPI
is not a truncated Taylor series.  That is correct.  I was earlier
alluding to the post that I made last March (which can be obtained from
the H-L archives :-) ) where I showed that a consistant formualtion
could lead to  .5 R0 + .25 R1 + .125 R2 + .0625 R3 + ...  I just didn't
want to post the math again.  It is available for anyone interested.
 
Also Dick Tuthill's original argument that one should multiply R0, R1 and
R2 makes sense.  I can add that like changing the multiplicative
factors in additive formula one can change the importance of each term
in his formulation by using powers for each term.  e.g. perhaps
R0**.25 * R1**.5 * R2**.25 .  Just a thought.
 
Ralph Baer