dailysudoku.com

[daj95376]

I've had something on my mind for awhile now, and decided the only way to get beyond it is to share it. I doubt if it's anything new, but I don't recall seeing anyone else express this opinion.

First off, let me say that I use and am a strong supported of AICs. I find them to be a convenient and effective way to present logical relationships. Second, there are those who support AICs but will bad-mouth Forcing Chains in the same breath. I'm writing this to put a damper on their position.

[Myth Jellies]

The difference lies not in what is discovered, but how it is discovered.

An AIC is a meta-pattern.

I can find an AIC by finding sub-patterns.

These sub-patterns (bivalued cells, bilocated digits, ALS's, URs, etc) always possess the same weak and strong link charateristics no matter what puzzle I work on.

Thus, I can assemble these sub-patterns into an AIC meta-pattern and learn something new about the puzzle without ever making an assumption other than that the puzzle-maker was following his rules.

This is very similar to the process most people use to find a naked pair. One usually sees the first ab-bivalue cell sub-pattern and then looks for an appropriately located second ab-bivalue cell to complete the pattern. No assumptions are needed during the entire process. Not surprising really since a naked pair is an AIC loop.

A forcing chain relies on you making assumptions to start it off. Comparing an AIC to a forcing chain is the same as comparing someone who recognizes a naked pair and uses it to make a deduction to someone who has to assume both cases of one of the naked pair bivalue cells and sees what happens to find the same result.

Thus, an AIC maker is looking for patterns that fit a pre-existing theory, while the forcing chain-ist is using ad hoc assumptions to prove things on the fly.

In addition, an AIC maker can learn to recognize and use new sub-patterns and gain the sense of satisfaction that that brings. The forcing chain-ist is limited improvement-wise to picking a good start point, which is very similar to the dilemma faced by those who utilize Ariadne's Thread-type algorithms to solve puzzles.

I'm not meaning to bad-mouth forcing chains. Rather I am just pointing out that solving puzzles in that way leaves little room for improvement, and thus, hardly requires a discussion forum.

[daj95376]

I abhor getting into a discussion about semantics. An AIC says nothing about how the user found the final relationship. All it says is that starting at one end and proceeding to the other end leads the reader through all of the sub-patterns used. If anything, a forcing chain at least gives the reader an idea of where the user started.

For example, an XY-Wing is a very common sub-pattern that's described (in this forum) as vertex, pinchers, and eliminations -- in that order. This is definitely a forcing chain description. However, this description can't be written as an AIC. Does this mean that everyone who starts by looking for the vertex in an XY-Wing is wrong ... or that they can't transcribe it into an AIC later?

Moving along and using the AIC from above ...

[wapati]

[Myth Jellies]

The issue is that when you describe an AIC construct via a forcing chain, you give the very natural impression that you plugged in one truth value and followed its implications along, then you plugged in the other value and did the same thing; and where the two agreed you had your deduction.

The reality is you may have found it like I do, and made no such assumptions; but try explaining that to a bunch of newcomers. I have many times and it is just not easy when their primary exposure to the concept is "if you assume x then it leads to y, and if you assume not x it leads to y, therefore y"

[storm_norm]

[Myth Jellies]