🚀 go-pugleaf

In article <BwrLMH.B6y@usenet.ucs.indiana.edu> chalmers@bronze.ucs.indiana.edu (David Chalmers) writes:
>In article <1992Oct24.135558.15100@oracorp.com> daryl@oracorp.com (Daryl McCullough) writes:

>>In terms of falsification, you could just as well have said "All
>>non-black things are non-crows". In this form, to test the theory, you
>>would look for a non-black thing which is not a non-crow. Getting rid
>>of the double-negatives, once again we find that we are looking for
>>a non-black crow.

>>So, while the process of establishing a law by induction seems to
>>allow a green tie to be evidence for "all crows are black", the
>>process of attempting to falsify the law doesn't allow such
>>irrelevancies to count as evidence.

>Hang on.  To falsify this, we're looking for crowish non-black things,
>right?  I happened across a non-black thing (that green thing in my
>closet), and it turned out not to be a crow (in fact, it was a tie).
>Doesn't that mean my attempt to falsify failed, and that this should
>be counted as evidence.

The apparent paradox can be clarified by looking at it as induction
should be treated, namely, by the use of probability.  To really get
at the guts, one needs a Bayesian interpretation, but his does not
add much to the analysis.

For simplicity, let us say that there are 10^6 crows and 10^50 objects
which we can recognize, most of which will be non-black.  Now if a
crow is selected "at random" and found to be black, we have ascertained
that one of the 10^6 objects named "crow" satisfies what we are trying
to show.  However, if a non-black object is selected at random and found
not to be a crow, we have only established that onw of the 10^50 objects
satisfies it.  In addition, we have reason to believe that crows are far
more similar than non-black objects; we consider them to have a common
biological origin, not going too far back, while many of the non-black
objects are not even in the avian classification.

-- 
Herman Rubin, Dept. of Statistics, Purdue Univ., West Lafayette IN47907-1399
Phone: (317)494-6054
hrubin@pop.stat.purdue.edu (Internet, bitnet)  
{purdue,pur-ee}!pop.stat!hrubin(UUCP)