If we watch this from tcp, I think we can figure out what is happening, and I am almost certain, that in my experience, with redirects, GoDaddy is clearly at fault here.
Here is the tcp data from the first linked site:
(stripping non important ack/syn start of transaction)
ACK PUSH
GET / HTTP/1.1
Accept:
/
Accept-Language: en
Accept-Encoding: gzip, deflate
User-Agent: Mozilla/5.0 (Macintosh; U; PPC Mac OS X; en) AppleWebKit/416.12
(KHTML, like Gecko) Safari/416.13
Connection: keep-alive
Host: www.catalogueofships.com
Packet received at 1134084628.096297
TCP packet from 64.202.167.129:http(80) to 192.168.1.101:49928 (73 bytes)
FIN ACK
HTTP/1.1 302 Moved Temporarily
Content-Length: 0
Location: /?ABCDEFGH
This is standard http 302 temporary redirect. Now, we could argue that they should probably use 301 permanent redirect, so the search engines actually index the redirected to page, but that would not solve this. FYI, in most all cases, a 302 redirect will be ignored by search engines, as it is a temporary status code, one to be removed soon, ie: server too busy, look elsewhere.
Anyway, the location field is the key /?ABCDEFGH is a valid URI, but that is about it, and in all my development, I have always learned you need to use a FQDN in the location field of an http header.
RFC on the matter, is a little vague and I can not fully interpret it:
http://libraries.ucsd.edu/about/tools/rfc2616-10.html
I think this is the important part
If the 302 status code is received in response to a request other
than GET or HEAD, the user agent MUST NOT automatically redirect the
request unless it can be confirmed by the user, since this might
change the conditions under which the request was issued.
Now, they are sending a valid GET, so it should not apply, but for some reason, I have a feeling this is what is happening.
If I make my own 302 redirect, and watch that, for example;
TCP packet from ip.add.re.ss:http(80) to 192.168.1.101:50246 (113 bytes)
ACK PUSH
HTTP/1.0 302 Moved Temporarily
Server: myserver
MIME-Version: 1.0
Location:
http://www.amazon.com/foo
Look at the location field, it is a FQDN
and URI.
I looked in the webkit bugzilla and see no one has reported this, this would be the best way to get an answer for real, so I posted it here:
http://bugzilla.opendarwin.org/show_bug.cgi?id=6016
Using the built in feedback mechanisms in OS X probably wont do much good at getting an answer in a timely fashion, the bug tracker certainly has for me in the past.