Appendix E. Where these tests came from, in short and honest
The diary, in order
Two days of splits. The death-sense rule taught me that I needed a place to stop. Chapter 8 teaches that rule. One clause in it is about when two sayings are tied together. That clause went through six wordings in two days. Each fix gave the next readers something new to split on. They split on who counted as the second speaker. They split on what to each other meant. They split on whether the narrator's words were speech.
8 November 2026. I froze the text of that rule and gave it the label Passed, limited. On the same day I chose a stopping rule. Here it is in plain words. I change a wording only when a reader shows, from the text in front of them, one of three things.
- A wrong label.
- The answer the text does not tell where the text does tell.
- Two readings that print two different labels on one passage.
If two readings print the same label, I do not change the wording. I write the split beside the rule as a limit, with the passages it touches. And a part that is a choice I made changes only when I choose. It never changes because readers split on it.
I wrote down the guards too, because a stopping rule can fool its maker. The same label can hide two different reasons. A frozen wording can look settled when it is only tired. A choice I made could spread to parts that should be tested. And a stopping rule can look chosen to favor a result. So I wrote it down before I read the next result.
11 November 2026. Three days later, I did choose. I changed one other clause of that rule twice. It is about who counts as speaking for God when a formula naming God is said. No reader has tested that clause since. The rule says so every time it runs. Its label rests on the tests that came before.
17 November 2026. This day was about the nearness-length rule, which chapter 7 teaches. Older wordings came first, and chapter 12 tells how some failed and some were replaced. Then came three rule rounds of fresh readers. There were six reader tests in all. Each one ended with the same words: it passes in part. Two reader disagreements stayed open. On 17 November I gave four rulings.
- I allowed a narrow exception for the words you will see. It rests on one case, and it needs judgment.
- I dropped Luke 19:11 as a bound.
- I ruled that a later text can give a length only if it carries a word of saying or promising.
- I ruled that a length built on Hebrew words is reported as resting on the NET's English.
Then I gave the rule its status. My word was limited. The fresh readers did not read the wording again after my rulings, and the rule's limit says so.
The second stopping rule. I used a second stopping rule for the long argument. It is simple. Stop after two argument rounds in a row with no new confirmed error.
20 November 2026. On the morning of that day, an AI editor found a misquote that no reader had caught. The count of clean rounds went back to zero. Chapter 12 tells that story, and the other mistakes we caught. Later that same day, we counted two clean rounds in a row, and the stopping rule was met. After the rules were updated, we counted two more clean rounds, and it was met again.

