is this expected or am i on crack?

hello all,

first off, some information:
=====
=====
the table:
             Table "facts"
  Attribute  |     Type     | Modifier
-------------+--------------+----------
 keyword     | varchar(80)  |
 description | varchar(255) |
 url         | varchar(255) |


the sql:
select keyword from facts as f1
where 1 <> (select count(*) from facts as f2 where f1.keyword = f2.keyword)
order by keyword;

my system:
Linux kernel 2.2.x
p166
32M ram
200M+ swap
=====
=====

it took about 65 minutes to complete. i know that it is doing alot of work,
but it there a way that it could be sped up, like something i could configure
or something that i could do to make it faster?

just looking to see if the time it took to take this is to be expected?
postmaster was using from 78-98% of the cpu for the whole time.

in hind site i wish i would have made keyword not null primary key...
*sigh*, learn from mistakes i guess. now have to fix the duplicates and
move all the rows into a new table with keyword as primary key...

thanks for any help,
and developers: thanks for developing a real non-toy database for the masses,

-chris humphries