Tree G: theory. Summary in plain language
The question
Before we measure anything, we should say what a result would mean. If we ever find a system that does more than we paid for, what follows? And what would a sceptic say?
What the work looks like
Reading and writing, with no code. The output is one paper of about six thousand words. It restates Levin's idea as a single measurable claim. It explains how we count effort and competency, and why that counting has known limits. The no-free-lunch theorems say every advantage is paid for somewhere, so a free lunch only ever exists relative to a set of tasks. The paper also states in advance that a positive result would prove nothing non-physical. It would only show that our bookkeeping missed a term. Finally it gives a one-page checklist that every claim of an "emergent mind" has to pass, ours or anyone else's.
Other people are most likely to use the checklist. It asks five questions. Did you change the measurement and see if the effect survived? Did you compare against an untrained copy of the same system? Did you say in advance what would surprise you? Did you let a critic rerun it? Did you rely on what the system says about itself?
What the outcomes could be
- A paper that makes the rest of the programme legible and answers the two standard attacks before they are made (that the idea is unfalsifiable, and that it is dualism in disguise).
- A checklist page cited by people who never read the paper.
- Possibly a better unit for "effort" than compressed size, if constructor theory turns out to give one.
What kind of research this is
Philosophy of science aimed at a specific engineering target. It is cheap and quick. A small group can do it better than a large lab, because a small group has no result to protect.
Timeline
The sections that constrain Tree A come within the first month, and the full draft by the end of the quarter.