Shareaza

I'm trying to get a few files that are apparently fairly rare. They're part of a larger set other members of which are more common. There is a search query, call it "X", that matches them all and narrower queries, call them "X Y" and "X Z" for example, that match specific individual ones.

Here is the strange thing. The broad "X" query finds everything, including occasionally the rarer files in the set. All of the less rare get found in numerous copies and downloaded easily; all of the rare ones get found occasionally with sources in ones and twos and a couple of these have also eventually downloaded.

But the narrower "X Y" searches, though each matches the name of one of the rarer files found by the "X" query, never seem to generate any search results (other than spam)!

What would cause this phenomenon? It seems that the narrower search should if anything be more effective, and certainly not less -- a rare file whose name matches the "X Y" search should be found by both searches whenever both are run and a source for it is online. There's no reason for it to ever be found by the "X' search while never being found by the "X Y" search, given the assumption that all its sources don't intentionally and systematically withhold results that match narrow queries they receive, while responding honestly to broad ones. (And that seems to be a pretty safe assumption -- who in their right mind would program a p2p client to do so, and how would every single source for a particular file end up happening to use such a client when such clients are expected to be super-rare if they exist at all? Assuming Shareaza 2.5.3.0 is not such a client, as seems extremely probable, it's no longer true for the two of the rarer files that I have thus far succeeded in downloading, in particular. Surely others have downloaded one of the other files that don't have a client that behaves in such a manner, then.)

Sure you don't put your terms in "" (quote mark) right?

I want to know why this happens. maybe one of clients does not work correctly.

That seems silly. All-numeric query terms are quite normal and legitimate -- most commonly, year numbers like 2010.

Query terms, yes. Words, no. So they are handled differently from actual words. Which might or might not be connected to what you were seeing. Thought I'd mention it since your question was about Shareaza handling a search term oddly.

Incidentally, if X is a legit word, you can try making a query like "X X_without_the_last_letter Z". That would tell if Shareaza counts the number as a search term and fails the search because it only finds hits for 50% of the query. Assuming of course that this is related to a number, which you haven't actually said that I noticed.

Actually, in the case at issue the query terms were both fully alphabetic. I just thought it odd what you said that implied that you couldn't narrow searches (other than client-side) by numbers such as 2010.