View Single Post
Old 05-23-2019, 02:52 PM
Voyager's Avatar
Voyager is offline
Join Date: Aug 2002
Location: Deep Space
Posts: 46,786
Originally Posted by Half Man Half Wit View Post
I'm not a proponent of IIT, but, as far as I understand it, the issue is: you can calculate (approximately, at least) the integrated information between the parts of a physical system (actually, what to consider a system's 'parts' is only determined, via a minimization, upon calculating---essentially, you split the system up such that the integrated information is minimal), essentially calculating a quantity that's somewhat similar to the mutual information. For a brain, that gets you a high number, for a computer, a low one.

So in that sense, the integrated information is a physical quantity that's present in a system, but not, in general, in a system that's simulating that system---just like the mass of a black hole isn't present in a simulation of said black hole (which I gather is rather a good thing).
I think it would be hard to calculate this number based only on the architecture of a computer, not what it was running.
Information, unlike mass, must be present in a simulation of something with that information. If you simulate telephone traffic, say, you don't need switch hardware but you do need the contents of the calls. This is simulation, not modeling, where you can describe the traffic mathematically without the information in them.
That's information, not integrated information of course. I did read at your link but found it less than interesting.