We consider the task of compression of information when the source of the information and the destination do not agree on the prior, i.e., the distribution from which the information is being generated. This setting was considered previously by Kalai et al. (ICS 2011) who suggested that this was a natural model for human communication, and efficient schemes for compression here could give insights into the behavior of natural languages. Kalai et al. gave a compression scheme with nearly optimal performance, assuming the source and destination share some uniform randomness. In this work we explore the need for this randomness, and give some non-trivial upper bounds on the deterministic communication complexity for this problem. In the process we introduce a new family of structured graphs of constant fractional chromatic number whose (integral) chromatic number turns out to be a key component in the analysis of the communication complexity. We provide some non-trivial upper bounds on the chromatic number of these graphs to get our upper bound, while using lower bounds on variants of these graphs to prove lower bounds for some natural approaches to solve the communication complexity question. Tight analysis of communication complexity of our problems and the chromatic number of the underlying graphs remains open.