Dory is a player-paradigm interactive music system for freely improvised music. While indebted to research in decentralised, agent-based musical systems and those employing Markov processes, Dory sets itself apart by capitalising on the artefacts introduced by suboptimal machine listening affordances and by speculating on the role of episodic and short-term memory in improvisation, realised as a second order Markov chain. Dory exhibits both reactive and learning traits, but is unable to exclusively commit to one or the other. By employing a subsumption architecture, unsophisticated machine listening techniques and probabilistic treatment of musical information, Dory appears as an engaging musical partner that exhibits seeming novelty and creativity. In this video, the second short improvisation I played with Dory during the Improvisational Creativity Workshop, Monash University, Prato, Italy, on 20.07.2017 is presented.