Cybernetics
02 · Ashby 1956 — regulation → today's AI
Foundations
- Claude Shannon — information entropy, H = −Σ pᵢ log pᵢ
- W. Ross Ashby — requisite variety, H(R) ≥ H(D) − H(E)
Related
- Ashby Says — fifty of Ashby's sayings, by topic
- Unsolved Problems in Cybernetics, and Subjects for Exploration — Ashby's list of open problems
Topics
Research
The Status of Requisite Variety, Revisited
Ashby's Law as a principle-theory constraint
The LLM as variety transducer: Ashby's intelligence-amplifier architecture and Beer's viable system model as a diagnostic and predictive framework
Rebuilding Ashby's Amplifier
Prompt, context, harness, and loop engineering as the components of a 1956 design
The Hollow Human: A Cybernetic Anatomy of AI and Ethics
70 years ago Ross Ashby designed an 'intelligence-amplifier'. Have we finally built it?
Intelligence as appropriate selection — and the category error at the heart of AI
Intelligence as Selection: Ashby, Beer, and the Design of Viable Autonomous Vehicles
The Machine That Outplayed Its Designer
Ashby's 1952 question, and the sixty-five-year journey to its answer
The mystery of economic growth
A note to the IIEA's European Recovery Project
Writing
Ashby's Variety & Shannon's Entropy - A History
Shannon, Bourbaki and his journal: where Ashby's variety came from
Ashby's Variety & Shannon's Entropy
Variety, entropy, information, and the regulator as a correction channel
Call it entropy...no one knows what entropy really is.
A limit accepted, and the law of requisite variety
The Life and Opinions of Tristram Shandy, Gentleman
Palimpsestic syndromes, adumbrationism, and laws of eponymy
70th anniversary of Ashby's presentation at Dartmouth's summer AI workshop
Ashby set out to amplify intelligence, not simulate it. Seventy years on, I think that's what we've built.
The mystery of economic growth