If you needed reminding that I have become a grumpy old man, here it is.
On Friday we had a nice talk from Chris Dowden, currently a postdoc at Royal Holloway. He talked about “Agreement protocols in the presence of a mobile adversary”, and you can read the abstract here.
Briefly, a number of networked nodes have to agree on a value (in Chris’s example, the time to meet for lunch). But some nodes have been taken over by an adversary, who can make the nodes pass on incorrect information, possibly different information to different people, and can store incorrect information in the node’s memory. The value is suggested by one particular node, and the aim of the protocol is to ensure that all the nodes not controlled by the adversary agree on a value, and that if the node suggesting the original value is not controlled by the adversary, then the agreed value is the one suggested by that node.
A well-defined mathematical problem. What made me uneasy was the idea that the aim is to agree on a value, even if we have no way of knowing whether or not it is the correct value.
Does this remind you of anything?