= 27d567bc2d48d9f40e9ccc20ea370b15)
lukeprog comments on The Fabric of Real Things - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (305)
Mainstream status:
The introduction to causality is intended to be bog-standard. All departures from mainstream academia are errors and should be flagged accordingly.
I can't particularly recall hearing anyone suggest that a universe is a connected fabric of causes and effects in the Pearlian sense of causality. Since Special Relativity there have been many suggestions that a universe is a connected fabric of 'events', or points in spacetime.
I can't particularly recall reading that you can only meaningfully talk about things you can find by tracing causal links (this theory will be developed further in upcoming posts).
Once again speaking as the author of Eliezer's Sequences and Mainstream Academia...
Like Eliezer, I haven't seen anyone suggesting that a universe is a connected fabric of causes and effects in the Pearlian sense. Nor have I seen anyone suggest that one can only meaningfully talk about things that one can find by tracing causal links.
If some young philosopher wants to publish some great, attention-grabbing papers in Nous early in their career, I suggest they write up the philosophy-paper versions of the core ideas in The Useful Idea of Truth and The Fabric of Real Things (citing Eliezer, of course).
What would it even mean to say that the universe is a fabric of cause and effect "in the Pearlian sense of causality"? Pearl originated a new method of causal analysis, not a new concept of causation.
I guess it means causation as manipulability, as opposed to e.g. causation as contrafactual?
How is causation as manipulability "Pearlian"? Pearl's whole point is that it's possible to determine causality without manipulating.
In so far as Pearl has a theory of causation, it is a manipulability theory. Sure, Pearl provides a method for drawing causal conclusions from data about correlations. But what he means by causation is cashed out in terms of how interventions on one variable affect another variable, not just in terms of correlations. Correlations are indicators of causation for Pearl, but they don't constitute causation. The structural equations Pearl uses to represent causal relationships are meant to capture not just de facto mathematical relations between variables, they support counterfactuals. They tell us what would happen to the variable on the left hand side of the equation if the variables on the right hand side were manipulated.
See section 6 of the SEP article for connections between the manipulability theory of causation and Pearl's formalism.
But first you need to know what causality is. While the general idea that "causality is about manipulations" was around before Pearl, he certainly popularized it, and his causal bayesian networks and the do-calculus made it something that could be studied mathematically.