Exactly. So why bother saying gzip is an approximation of KC? (I assume: because KC is a well-known theoretical object with good properties, and one shouldn’t let the fact that none of these properties carry over to gzip ruin one’s chance of getting published cheaply.)
Because gzip is being used to measure complexity. That’s literally the reason they used gzip and not, I don’t know, rot13. It’s an explanation of the causal role that gzip is playing in the whole process.
Exactly. So why bother saying gzip is an approximation of KC? (I assume: because KC is a well-known theoretical object with good properties, and one shouldn’t let the fact that none of these properties carry over to gzip ruin one’s chance of getting published cheaply.)
Because gzip is being used to measure complexity. That’s literally the reason they used gzip and not, I don’t know, rot13. It’s an explanation of the causal role that gzip is playing in the whole process.
No.
“gzip is being used to measure complexity” is an explanation of the causal role that gzip is playing in this study.
“gzip is an approximation of KC” is either 1) flatly not true, see edit to grandparent or 2) not relevant to the study at all.
And while we’re at it, why do we even care about Turing Machines, it’s not like we could ever build one anyways. ;-)
goes back to tending his potato garden