What the heck is CRC32 ?

CRC stands for Cyclic Redundancy Check. It's a way of calculating a single number from a file of any size, so that you can be sure that you downloaded it correctly, witthout errors.

The person putting up the file also gives its CRC number, and after downloading it you're supposed to calculate the CRC on your local copy. If they're the same, there's a good chance that you got it without errors.

CRC-32 is the 32-bit variant of this, so there are around 4 billion possible codes. That means you've got a 1 in 4 billion chance that despite the file having errors, it still produces the same CRC.

There are other, much better systems, such as MD5, which produce much longer codes to minimise errors even more.

[Edit]

For your "Duplicate Files" program, it probably looks at all the pairs of files that are the same size. If their CRC is the same, they're probably the same file, and you can delete one of them.

Because CRC is not perfect, the program also probably does a character-for-character comparison on files that produce the same CRC, just to be doubly-sure. I know if I were to write such a program, that's what I'd do!

It's be quicker to calculate the CRC for a file, and then compare the CRC against all the other files' CRCs, than it is to do a byte-for-byte comparison of every file against every other file - at least until a potential match is found!

It's normally a way to check that the data you received hasn't changed. If 2 files have the same CRC32, there's a chance that they're identical. (Using MD5 gives better odds of not getting a false positive in this case.)