DonM wrote:Your nrc notation is not by any measure a standard for manual solving and never has been any more than other alternative methods such as Bill's OR Chains (from the old Eureka forum) or eleven's freehand method. (On that score, it is not surprising that you and eleven have formed a mutual admiration society on this subject.) Nor has nrc notation been demonstrated to be a realistic alternative to Eureka notation for manual solving.

To me, absurdity is your promotion of your notation as a manual method when virtually every solution you have posted on this forum has been computer output.

nrc notation hasn't become a standard on the late Eureka or Sudoku Player's forums, mainly because of their being dominated by a clan of Eureka extremists who wouldn't hear of anything else.

As for the rest, you are just repeating

ad nauseam your old confusing claims. But a notation should be directed to the reader. It should be non-ambiguous and compact (but not overly compact). Whether examples of it are generated by hand or computer is totally irrelevant.

As for the general idea of "alternating inference", I defy you to find a better than nrc notation for it.

I'll let anyone judge for himself which of the following two options is clearer.

My nrc notation, with my general chain pattern:

{x1 Y1} — {x2 Y2} — {x3 Y3} — ... ==> not Z, where the xi are candidates and the Yi are candidates in the simplest cases but may be full patterns (to be defined within the {} or in separate lines) in more complex cases

— means linked, in direct contradiction, logical nand, i.e. it corresponds to your "weak links", a place where nothing really happens;

{x X} corresponds to your "strong links"; the { } notation provides space for where interesting things happen.

Concrete examples from my previous solution:

whip[3]: c3n8{r1 r4} - r4c9{n8 n5} - b3n5{r1c9 .} ==> r1c3 ≠ 5

biv-chain[4]: r3c1{n5 n6} - r2c1{n6 n4} - r2c3{n4 n1} - r3n1{c3 c9} ==> r3c9 ≠ 5

Or examples taken from the above "solutions":

9r9c3 =UR= 3r7c8 - (3=4&6)r147c7 - (4|6=137)r2c279 - (1=9)r279c3

[(9r9c3=3r7c8)ur:45r79c38-(3=456)r147c7-(46=137)r2c279-(1=9)r279c3]-45r9c3

(9=45)r79c3 -UR- (45=3)r79c8 - (3=456)r147c7 - (46#1=17#1)r2c7 - (137=4)r2c239 - (45=9)r79c3 => r9c3 = 9

(9=45)r79c3 -UR- (45=3)r79c8 - (3=456)r147c7 - (46#1)r2c7 = (6,4)r2c1,r2c3 - (45=9)r79c3 => r9c3 = 9

(45)r27c3 = (3)r7c7[AMSLS133445567:2n2379,7n3,147n7] - (3)r7c8 = (9)r9c3[AUR45:r79c38] => r9c3 <> 4

(9=45)r79c3 -UR- (45=3)r79c8 - (3=456)r147c7 - (46#1=17#1)r2c2 = (137=4)r2c239 - (45=9)r79c3 => r9c3 = 9

DonM wrote:I bought your book (right after it was published) based on its promotion as a new manual solving method. Never have I felt more misled by a sudoku book since it became very apparent that it was more appropriate for some niche of academia and/or a new approach to computer solving.

My book (I suppose you are speaking of HLS1, 2007) was never promoted as a manual solving method but as a complete theoretically based resolution model, based on pure logic. It also introduced new kinds of chains, at the pure logic level. Once more, whether such chains are found by humans or computers is irrelevant to their definitions.

Since the beginning, you have been allergic to oriented chains. That's your right and I respect it. But I can't accept that you consider that they are not for manual solvers. Many solutions proposed in this section of the forum use oriented chains (or even nets), pretending not to be aware of previous work and thus re-inventing the wheel.

You also constantly ignore the difference between AND branching (propagating the consequences of one step) and OR branching (bifurcating because there are several possible conclusions). Some of my chains have AND branching (seen in the t-candidates) but no OR-branching.

Indeed, there is some limited form of OR-branching in g-whips, but it is totally wrapped in the notion of a g-candidate.

DonM wrote:The most recent example is below where at least 3 solvers posted one-step solutions while your nrc-based computer solution was by any comparison long and obscure.

My solution was based on several very easy steps that anyone can understand without much effort. Most of them are bivalue chains (i.e. basic AICs). Only one is oriented, the only whip[3].

The pretendedly one-step solutions are totally artificial ones that no normal player would ever consider. They are so complicated that they generated three full pages of discussion and misunderstandings, part of which is due to under-specified AIC notation and part of which is due to uncertain logic.

The fact is, this whole section of the forum has become a monstrosity. Notice, I consider anyone is free to look for such artificially complicated solutions if that's what they like. But please don't try to make us believe that this is any kind of standard that a normal player should take as an example.

DonM wrote:Perhaps David Is too diplomatic to mention it, but I have no problem doing so:

He asked you specifically '1) Did you find your solution manually? We are discussing notations suitable for human players.' Your answer was ,' Whether a notation is suitable for a human player depends on its intrinsic properties; it should be usable by both human solver and software.'

That you evaded the question so flagrantly speaks volumes.

One more absurd claim. My resolution path started with "*** SudoRules 20.0.s based on CSP-Rules 2.0.s, config = W ***"

Isn't it explicit enough? Should I write it in red bold size 60 capital letters for you?