MPC 2015 - Programme

That simple, compelling model provided not only a refinement order for deterministic security (inverse refinement of set-partitions) but, since it is a lattice, allowed the construction of the 'least-secure deterministic program more secure than these other deterministic programs', and its obvious dual. But Landauer treated neither demonic nor probabilistic choice.

Later work of our own, and independently of others, suggested a probabilistic generalisation of Landauer's lattice, although it turned out that the generalisation is only a partial order, not a lattice.

This talk looks between the two structures above: I will combine earlier qualitatitve ideas with very recent quantitative results in order to explore

• What an appropriate purely demonic lattice of information might be, the 'meat in the sandwich' that lies between Landauer's deterministic, qualitative lattics and our probabilistic partial order.
• The importance of compositionality in determining its structure.
• That it is indeed a lattice, that it generalises Landauer's lattice and that it is generalised by the probabilistic partial order.
• Its operational significance and, of course,
• Suggestions on how it might help with constructing (secure) programs.