[Epistemic status: Written in 2010. Possibly of historic interest only.] Goodhart's Law states that "Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes." "Control" here means "policy", such as conducted by a government or central bank. To give just one classic example,...
(Originally posted at https://plus.google.com/+ThomasColthurst/posts/5fcw5wgVrpj in 2013) I love the premise of Peter de Blanc's "Ontological Crises in Artificial Agents’ Value Systems": you are robot who used to think the universe was described by some set of states O1, only to wake up and find out that the universe is better...
I answer the "why are trees so tall" question at http://whirledofideas.blogspot.com/2015/05/why-are-trees-tall.html . The rough answer is that height limiting treaties need to be enforced, and enforcement is costly, and for the most obvious enforcement mechanism that trees have available (negative allelopathy), enforcement is not observable by other trees. So if any trees did enforce a height limiting treaty, the trees that freeloaded on their enforcement would out-compete them, and that's why we see tall trees and not height-treaty-limited trees.
There is a large existing literature on pruning neural networks, starting with the 1990 paper "Optimal Brain Damage" by Le Cun, Denker and Solla. A recent paper with more references is https://arxiv.org/pdf/1803.03635.pdf
Especially for the study of consciousness and mental states, I associate this useful tactic with Daniel Dennett's term "heterophenomenology".
[Epistemic status: Written in 2010. Possibly of historic interest only.]
Goodhart's Law states that "Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes."
"Control" here means "policy", such as conducted by a government or central bank. To give just one classic example, the Federal Reserve might observe a strong correlation between inflation and unemployment rates which breaks down once the Fed tries to manipulate the inflation rate to reduce unemployment. [1]
The "any" in Goodhart's Law overstates the case -- I regularly use the observed statistical regularity between eating and feeling less hungry for control purposes, and as far as I can tell, this... (read 2576 more words →)
(Originally posted at https://plus.google.com/+ThomasColthurst/posts/5fcw5wgVrpj in 2013)
I love the premise of Peter de Blanc's "Ontological Crises in Artificial Agents’ Value Systems": you are robot who used to think the universe was described by some set of states O1, only to wake up and find out that the universe is better described by some set of states O2. What should you do?
Specifically, let's say that the robot used to act so as to maximize a utility function U: O1 -> R. The problem is to construct a new utility function U': O2 -> R.
Briefly, here's the solution Peter de Blanc gives in the paper: assume you have some side information about how probability distributions... (read 905 more words →)
FYI, there are published counterexamples to Cox's theorem. See for example Joseph Halpern's at http://arxiv.org/pdf/1105.5450.pdf.
Perfect descriptions of reality are unattainable, unnecessary, and too costly for learning organisms, including humans. But workable descriptions are indispensable. So knowledge systems, like maps, are a complex blend of realism, flexibility, usefulness, and inspiration.
-- David Christian, Maps of Time: An Introduction to Big History
Someone once quoted Shakespeare to the philosopher W. V. O. Quine: "There are more things in heaven and earth than are dreamt of in your philosophy." To which Quine is said to have responded: "Possibly, but my concern is that there not be more things in my philosophy than are in heaven and earth."
Hi. I'm Thomas Colthurst. I will be doing a visiting fellowship at the Singularity Institute this summer.
Prior work, in the form of a twitter joke: https://twitter.com/thomascolthurst/status/1032345388605431810