Weirdly, I answered almost exactly the same “invert the RNG” question on StackOverflow (https://stackoverflow.com/a/15237585/1204143) just a few months before you posted your question to Crypto.SE, except that I attacked the Java RNG instead of the MSVC RNG. They both use a simple LCG design where outputs are truncations of the internal state, so the attacks are very similar to each other.