Hey everyone!

Just finished up giving a lesson with a student who is reading this excellent book. He had a few questions about some of the positions in the book, so we paused the regularly scheduled topic to look at some of his questions more in-depth. In every case, the student would posit the problem number to me, I'd put it into an XG without hitting "analyze", and I'd answer the question and give my explanation without looking at the provided solution. I'd then have XG analyze it and check Bill's explanation.

What I was surprised to find were a few situations: ones where the 2nd best play was only off by a very small amount (less than .005) and problems where ++ disagreed with the book solution. In the first situation, this makes sense -- I'd get the right answer on my own, often because the "small difference" plays weren't things I'd consider because they were similar to the "best play" but had something wrong with them that made them strictly dominated by the correct answer in an obvious way. A good example of this is problem #52, where if I'm going to split, I'm definitely playing 13/11 to unstack the mid.

Ultimately, my student, who would select an inferior play, would come away from it thinking they blundered, only to be wrong by a tiny bit in many cases. I think most players view the book as "bathroom reading", so it's unlikely most would sit down to check with XG to see if there are reasonable alternatives. I get that the conceptual reasoning is what's important, but something like "many of the splitting plays are close" would be helpful. I guess it lists it first, so that points it out as priority? The key to Problem #123 (clear cut right answer for this one) is one of the last things mentioned in the explanation. In Problem #106 (clear-cut right/wrong, easy one for open players), everything said is true, but the harrassment value of keeping the checker back to confront oppo's slot and muddy his priorities with 3s gets no mention at all.

As for ones where ++ disagrees with the right answer: what's the rollout level on the problems in the 2017 edition? I know this is a recent update, were all problems taken to a higher court on the update? Were explanations re-written for ones where the answers changed, or were they largely left intact from the prior edition? Are there any addenda out there/lists of problems worth rolling out for the newer version? I found a list from the previous edition on Timothy Chow's site, but don't know if those are relevant to this version.

Thanks in advance.