I know that SandForce based ssd's compress files as much as possible in order to increase performance. Typically it is not possible to compress files 100%. A portion or some bits and pieces of a file will remain uncompressed. So far so good. At the other extreme the new Intel 510 series ssd's handle large incompressable files very efficiently?
Do files exist which are 100% incompressable? For example, are "raw photos" taken with a digital camera considered to be incompressable files? Do games like Crysis2 have large incompressable files?
Photos and videos are highly compressible because they use lossy compression. That's something that a storage device CANNOT do - it must use lossless compression because every bit in the data being stored may be important. You wouldn't want your tax files or spreadsheets to be giving you "approximate" results.
Back in the days when tape drives started compressing data at the hardware level they'd always claim a 2:1 compression ratio. But that figure varied a lot based on the kind of files you wrote to the drive. For compressed formats like JPG and MPG you'd basically get no compression at all.
The same will be true of the compression algorithms used in SSD controllers. In general, the more the file is compressed already, the less additional compression you'll get when it's stored on the SSD.
RAW files do have at least some compression done on them - if that wasn't true then every RAW file from the same camera would be exactly the same size. The difference is that a lossless compression scheme is used so that all of the pixels can be exactly reproduced.
I'd imagine that the bulk of data associated with a video game would be visual and audio information. The visual stuff would definitely be compressed, and I'd think that the audio is likely to be as well.
Since lossless compression relies on patterns in the data, completely random data would be impossible or near impossible to compress. I'm not sure if pseudorandom or "true" random data would produce any significant differences in compression levels though.
Basically compression relies on replacing a pattern of recurring information with a representation that uses less space. Just imagine an essay with supercalifragilisticexpialidocious that appears 100 times. Now replace each instance of that with # and mark in the beginning that's what # means. You've just compressed that essay ^_^. Of course, if # is used elsewhere then you may need to use another representation for #. It all depends on the compression algorithm used. You can actually increase file size slightly if the amount of information added to the top by the compression algorithm is greater than the amount saved.