Chapter 4. Where the rules come from, and how they were tried
The limits of the whole process
I want to be honest about what this process is and is not.
- The readers were AI readers. They were not outside human readers.
- The tests were run on the cases the readers were given. They were not run on every passage.
- Many rules are marked as not retested after my rulings.
- The rounds on the long argument test my long argument. They do not test the Bible.
- Counts from round to round are not like for like. A drop or a rise does not by itself mean improvement or harm.
- No result here is proof. A passed rule is not a proof either.
So do not trust the tests because of how many readers there were. Trust nothing here that you have not checked. If you find a quote that is off, tell me. If you find a rule used outside its limit, that is a finding too.
The honest claim for a passed rule is small: no contradiction found in the passages checked. A small claim, honestly kept, is worth more than a big claim that cannot lose.
Now the tests themselves, one at a time. We start with the most common mistake: hearing the same words and assuming the same event.
Three questions
- What two things can make a rule stand, and what can never make one stand?
- What is a split, and why is it useful to find one?
- What does it mean that a human decides every status?

