why does dig +trace change outcome?

[Hal Brooks]

Some name servers, but not all, are unable to resolve 
krensavage.com

But even those which are ordinarily unable to resolve 
krensavage.com seem to be able to get answers with dig if 
the +trace option is used.

For example, this web site:

http://www.kloth.net/services/dig.php

offers a browser interface to dig.  If you use their 
localhost as the server and query for an A record, but 
don't specify "trace", then the unhappy result is 
"connection timed out; no servers could be reached", but if 
you do specify "trace" then you get 6 answers.

Similarly from the command line:

 	dig @dns1.uga.edu krensavage.com        # times out
 	dig @dns1.uga.edu krensavage.com +trace # OK

But other name servers handle it fine (and dig works 
even without +trace):

 	dig @ns1.usg.edu krensavage.com         # OK
         dig @ns1.usg.edu krensavage.com         # also OK

The only shortcoming I see with the krensavage.com setup is 
that the whois record only has 1 name server 
(yns1.yahoo.com) listed.

For comparison, olivebarn.com, has similar characteristics 
(though the whois record lists both yns1.yahoo.com and 
yns2.yahoo.com), but it doesn't result in the same failures 
I'm seeing for krensavage.com.

Any and all help appreciated.

-hal