I am looking for an example of a simple hashing algorithm that can be shown to be vulnerable to multiple preimage attacks for a specified input length.
For example, if I know that the input data is exactly 160 bytes and generates a 16 byte hash, then there exists some method of finding other inputs that are 160 bytes in length and generate an identical 16 byte hash without resorting to pure brute forcing.
Please note that I do not mean using something like MD5 or SHA1. I understand that these are designed to make this impractical. This algorithm would be a textbook example of the flawed design of a hashing algorithm and how that flaw can be practically exploited. I have tried Google, but much of what turns up is theoretical research on known, secure algorithms.