[EM] Responses to some of Forest's ideas

[Richard Moore]

Bart Ingles wrote:

> Suppose you have absolutely no polling data of any kind, but know enough
> about the candidates to place them on a policy continuum.  Should this
> still be considered a zero-info situation?


No, it shouldn't. However, in order to affect strategy, 
there has to be a way to determine voting probabilities to 
some degree of precision. If a candidate is on the 
borderline (zero-info) then it wouldn't take much evidence 
to sway a voter one way or the other, so the precision 
doesn't have to be great in such cases. But if my utilities 
are 100, 75, and 0 then I'd better be real sure of the 
probabilities before I exclude the second candidate from my 
ballot on that basis.

Also, putting the candidates on a policy continuum may work 
in theory, but in practice there are other factors that this 
ignores, such as political organization, financial 
resources, and charisma.

Richard