When you search for a file name, and LW says there's multiple copies of it, how does LW know the files are identical. Is it based on the ascii text of the filename and maybe the file size ? Or is there some kind of long unique serial number secretly embedded into the file by LW at some stage ?

I'm asking because often I want to rename some files that I've downloaded but I don't want to mess up the effectiveness of other LW peers search results.

Go ahead and rename the file to something more useful. This won't affect the unique identifier number called the "hash." As long as you leave at least a part of a keyword in the name, others should be able to find the file.

Currently files are exactly verified by a SHA-1 hash. You can see this long identifier if you right-click on a file in the Library pane and choose "show magnet details." The hash isn't actually embedded in the file--it's created and saved in a separate table for lookups.

The hash is necessary so that people can request even many small portions of the file and be sure they will get the proper copy. This is essential when downloading from multiple sources, and to Verify that the bits of the file you got exactly match the bits of the original.

However, if I understand the process correctly, searches by hash are only used in odd situations.

File names are used for help find files in searches by keyword and also to give the green checkmark if a filename matches one you already have.

So, you aren't actually changing the file itself when you change the name. Just give it a more useful name and others should be grateful.

Thanks for your explaination it's very helpful.
I'd like to read more detailed technical info about how LW works, prefereably written by those who know the facts.
Any idea if there is an in-depth technical explaination, but written for the non-programmer general public ?

try http://gnufu.net/

http://www.limewire.org/techdocs.shtml is more technical

the developers discuss the network at http://groups.yahoo.com/group/the_gdf/

cheers