Love the Byzantine Generals problem :) - ThreadSky | a Reddit-style client for Bluesky

etheryth.bsky.social • 15 days ago

Love the Byzantine Generals problem :)

The email game is based on / strongly related to the coordinated attack problem of two generals standing on either side of a valley to be attacked... described beautifully here by my colleague at
Cornell CS, Joseph Halpern:

Comments

alxfed.bsky.social•15 days ago

The incomplete information games rarely have any resemblance of winning strategy. The uncertainty, especially the uncertain amount of uncertainty like in this case prevents the system from assuming any semblance of equilibrium too.

etheryth.bsky.social•15 days ago

I think i think that equilibrium is found over time in an ecology of multiple games and singular agents with intent on purely winning may not be helpful to that. Then again ... zd strategies share a lot with executive function in society no?

alxfed.bsky.social•15 days ago

Also, this type of problems demonstrate that there are actually _two_ types of 'Information Theory' as Wiener rightfully pointed out, - a 'Differential Information Theory' of Shannon and the 'Integral Information Theory' started by Wiener in his book 'Cybernetics'.

etheryth.bsky.social•14 days ago

On 1991 i was in art school and, influenced by Stelarc, i found myself reading Wiener's Cybernetics. I didn't have much context in which to understand it and i didn't really go back there while doing my degree. I should, and i will reply to this when i have refreshed on it, but for now this is 1/2

etheryth.bsky.social•14 days ago

... also making me think about the many definitions of entropy, and moreso a recent-ish paper that enumerated over a hundred i think it was, theories of intelligence ftom IIT to pan-psychism and beyond!

Wrt to Shannon and the quotient H, hey now there's an apparently immovable object 🙃

alxfed.bsky.social•15 days ago

It's a problem with diverging infinitely branching solutions.

etheryth.bsky.social•14 days ago

Bifurcations galore :) I have tried to think about how reaction-diffusion might be useful as a process model for complexity in ai reasoning but admit I haven't gone very far with it.

Comments

Posting Rules

Reply