Towards a Simple Stochastic Model of Transcription