The New Sudoku Players' Forum

[Maria]

Original puzzle

Code: Select all

///
 |  *      *    6      |  *    *    *   |   *     *     *9    |


Much better now?

close but no cigar -- that extra 9 in r9c9---

and i too would prefer dots for empty cells, thus --

Code: Select all

 . . 7 | . . . | 6 2 .
 . 5 . | . . 1 | . . .
 . 4 . | . . . | . 7 .
-------+-------+------
 3 . . | . 5 . | 1 . .
 1 . . | 7 . . | . . 8
 . . 2 | 9 . . | 4 . .
-------+-------+------
 . . 9 | . 3 . | . 5 .
 8 . . | 2 . . | . 1 .
 . . 6 | . . . | . . .


Thank you, you did it for me!

[Maria]

It is not just a matter of personal preferences, the most important reason is that many people use computer programs to analize whatever you ask about and these program require a specific format to accept the data, which was defined long ago as to require dots to replace empty cells (you can write the candidate on that too, just write them one after the other whithout any soace in between them).

Then, here is mine, this is very correct, now if somebody has an easy explanation for me to follow? I followed previous advise, but what is the technique name so that I can check and learn more? :

Code: Select all

 . . 7 | . . . | 6 2 .
 . 5 . | . . 1 | . . .
 . 4 . | . . . | . 7 .
-------+-------+------
 3 . . | . 5 . | 1 . .
 1 . . | 7 . . | . . 8
 . . 2 | 9 . . | 4 . .
-------+-------+------
 . . 9 | . 3 . | . 5 .
 8 . . | 2 . . | . 1 .
 . . 6 | . . . | . . .


[aran]

Maria, this is the grid as per your post

Code: Select all

 *--------------------------------------------------------------------*
 | 9      38     7      | 3458   48     458    | 6      2      1      |
 | 2      5      38     | 6      7      1      | 389    48     349    |
 | 6      4      1      | 38     29     29     | 358    7      35     |
 |----------------------+----------------------+----------------------|
 | 3      679    48     | 48     5      26     | 1      69     27     |
 | 1      69     45     | 7      246    2346   | 25     369    8      |
 | 57     68     2      | 9      1      368    | 4      36     57     |
 |----------------------+----------------------+----------------------|
 | 47     12     9      | 148    3      4678   | 278    5      246    |
 | 8      37     35     | 2      469    45679  | 79     1      469    |
 | 457    12     6      | 1458   489    45789  | 23789  48     2349   |
 *--------------------------------------------------------------------*

If you are familiar with Deadly Rectangles, then one way to begin is to examine the rectangle 69 in r45c28.
To avoid the deadly pattern, either r4c2=7 or r5c8=3 (otherwise there is a Deadly Rectangle).
If you are familiar with notation, this can be written 7r4c2=3r5c8.
If you are familiar with little logical chains (called AIC), then this is a promising start for such a chain giving :
7r4c2=3r5c8-(3=6)r6c8-(6=9)r4c8
from which it can be concluded that 9 can be excluded from r4c2.
This has brought a focus to cell r4c2 and indeed there is now, as it happens, a one move solution from there.
This again uses AIC logic.
7r4c2=7r4c9-(7=5)r6c9-(5=3)r3c9-(3=8)r3c4-8r1c456=8r1c2-(8=6)r6c2-(6=7)r4c2
from which it can be concluded that r4c2 must be 7, usually written at the end of the chain as : =>r4c2=7.
This solves the puzzle.
(by the way, that solution would have worked even without the Deadly Rectangle move).
If this seems obscure to you, not to worry, we all had to begin and learn.

[Maria]

Maria, this is the grid as per your post

Code: Select all

 *--------------------------------------------------------------------*
 | 9      38     7      | 3458   48     458    | 6      2      1      |
 | 2      5      38     | 6      7      1      | 389    48     349    |
 | 6      4      1      | 38     29     29     | 358    7      35     |
 |----------------------+----------------------+----------------------|
 | 3      679    48     | 48     5      26     | 1      69     27     |
 | 1      69     45     | 7      246    2346   | 25     369    8      |
 | 57     68     2      | 9      1      368    | 4      36     57     |
 |----------------------+----------------------+----------------------|
 | 47     12     9      | 148    3      4678   | 278    5      246    |
 | 8      37     35     | 2      469    45679  | 79     1      469    |
 | 457    12     6      | 1458   489    45789  | 23789  48     2349   |
 *--------------------------------------------------------------------*

If you are familiar with Deadly Rectangles, then one way to begin is to examine the rectangle 69 in r45c28.
To avoid the deadly pattern, either r4c2=7 or r5c8=3 (otherwise there is a Deadly Rectangle).
If you are familiar with notation, this can be written 7r4c2=3r5c8.
If you are familiar with little logical chains (called AIC), then this is a promising start for such a chain giving :
7r4c2=3r5c8-(3=6)r6c8-(6=9)r4c8
from which it can be concluded that 9 can be excluded from r4c2.
This has brought a focus to cell r4c2 and indeed there is now, as it happens, a one move solution from there.
This again uses AIC logic.
7r4c2=7r4c9-(7=5)r6c9-(5=3)r3c9-(3=8)r3c4-8r1c456=8r1c2-(8=6)r6c2-(6=7)r4c2
from which it can be concluded that r4c2 must be 7, usually written at the end of the chain as : =>r4c2=7.
This solves the puzzle.
(by the way, that solution would have worked even without the Deadly Rectangle move).
If this seems obscure to you, not to worry, we all had to begin and learn.

Thank you for yr carefull explanation, I have followed you though I will never be able to find such clues in the future - let me check it once more - yours

[aran]

Maria
Don't be afraid to ask any question that will help you understand better.
It might be the meaning of a word, or of a symbol, or how to read an AIC chain, or what it all means, or how the conclusions follow from the chain, or whatever.

[JeJ]

Maria

Your grid is correct now, well done.

The program that I use to play is Hodoku and it classifies your puzzle as Unfair with a rating of 2242, meaning it is a very complicated puzzle requiring very advanced techniques to solve it.

[Maria]

Maria

Your grid is correct now, well done.

The program that I use to play is Hodoku and it classifies your puzzle as Unfair with a rating of 2242, meaning it is a very complicated puzzle requiring very advanced techniques to solve it.

Thank you

[Pat]

now if somebody has an easy explanation for me to follow?
I followed previous advise, but what is the technique name so that I can check and learn more?

you may like this "forcing net"
( matching advice posted by daj95376 ) --

if r2c3 = 8
then r4c3 = 4, r5c3 = 5, r6c9 = 5, r3c9 = 3, and r2 has no 3



Code: Select all

 *--------------------------------------------------------------------*
 | 9      38     7      | 345    48     458    | 6      2      1      |
 | 2      5      38     | 6      7      1      | 389    48     349    |
 | 6      4      1      | 38     29     29     | 358    7      35     |
 |----------------------+----------------------+----------------------|
 | 3      679    48     | 48     5      246    | 1      69     27     |
 | 1      69     45     | 7      246    2346   | 25     369    8      |
 | 57     68     2      | 9      1      368    | 4      36     57     |
 |----------------------+----------------------+----------------------|
 | 47     127    9      | 148    3      4678   | 278    5      246    |
 | 8      37     35     | 2      469    45679  | 79     1      469    |
 | 457    127    6      | 1458   489    45789  | 23789  48     2349   |
 *--------------------------------------------------------------------*

[eleven]

Maria,

maybe it helps you, when i say, that i dont like this puzzle. After all daj's 6 cell xy-chain is the easiest way to solve it. But to find it manually is extremeley boring, because you never know, where you should start. In this grid there are so much possible xy-chains, including at least two 5-cell xy-chains, which do eliminate a candidate, but are no help. So you can spend a lot of time for finding this useful one. (And there are are a lot of other techniques, you can try here without success.)

You saw, how aran solved it, which looks straightahead, but he also was lucky to find a useful starting point in the cell, where he eliminated a candidate.

So dont be frustrated, when you dont find a solution for such a puzzle. There are nice one's and much more boring ones. Unfortunately i dont know a source, where the nice ones are the majority. The problem is, that the creators dont solve themselves, what their programs generate.

[PIsaacson]

Maria,

Using the starting PM posted by aran plus his DP and the techniques list posted by JasonLion, I arrived at the following solution path:

Code: Select all

 *--------------------------------------------------------------------*
 | 9      38+     7     | 3458-  48-    458-   | 6      2      1      |
 | 2      5      38-    | 6      7      1      | 398-   48+    349    |
 | 6      4      1      | 38+    29     29     | 358-   7      35     |
 |----------------------+----------------------+----------------------|
 | 3      679    48+    | 48-    5      26     | 1      69     27     |
 | 1      69     45     | 7      246    2346   | 25     369    8      |
 | 57     68-    2      | 9      1      368+   | 4      36     57     |
 |----------------------+----------------------+----------------------|
 | 47     12     9      | 148-   3      4678-  | 278+   5      246    |
 | 8      37     35     | 2      469    45679  | 79     1      469    |
 | 457    12     6      | 1458-  498+   45798- | 23798- 48-    2349   |
 *--------------------------------------------------------------------*

Applying X-Coloring to the digit 8, the above parity map can be produced when starting from 8r1c2.
7 Eliminations can be applied starting from 8r1c2 and following conjugate pairs as follows:
a) Starting from row 2 conjs r24c3 & r29c8 - eliminate 8r2c7
b) continuing from col 4 conjs r3c47 & r4c34 - eliminate 8r179c4
c) continuing from row 9 conjs r19c5 & r29c8 - eliminate 8r9c67
d) finish with col 6 conjs r19c5 & b5x19 (or r6c26) - eliminate 8r1c6

Code: Select all

 *--------------------------------------------------------------------*
 | 9      38     7      | 345    48     45     | 6      2      1      |
 | 2      5      38     | 6      7      1      | 39     48     349    |
 | 6      4      1      | 38     29     29     | 358    7      35     |
 |----------------------+----------------------+----------------------|
 | 3     *69+7   48     | 48     5      26     | 1     *69     27     |
 | 1     *69     45     | 7      246    2346   | 25    *69+3   8      |
 | 57     68     2      | 9      1      368    | 4      36     57     |
 |----------------------+----------------------+----------------------|
 | 47     12     9      | 14     3      4678   | 278    5      246    |
 | 8      37     35     | 2      469    45679  | 79     1      469    |
 | 457    12     6      | 145    489    4579   | 2379   48     2349   |
 *--------------------------------------------------------------------*

There is a type 6 UR r45c28.<69> as aran pointed out. It immediately leads to 9 <= r4c8, r5c2 since the 9s are conjugate in both rows/cols and cannot occur in r4c2/r5c8.

Code: Select all

 *--------------------------------------------------------------------*
 | 9      38     7      | 345    48     45     | 6      2      1      |
 | 2      5      38     | 6      7      1      | 39     48     349    |
 | 6      4      1      | 38     29     29     | 358    7      35     |
 |----------------------+----------------------+----------------------|
 | 3      67     48     | 48     5     *26     | 1      9      27     |
 | 1      9     *45     | 7     *246    346-2  |*25     36     8      |
 | 57     68     2      | 9      1      368    | 4      36     57     |
 |----------------------+----------------------+----------------------|
 | 47     12     9      | 14     3      4678   | 278    5      246    |
 | 8      37     35     | 2      469    45679  | 79     1      469    |
 | 457    12     6      | 145    489    4579   | 2379   48     2349   |
 *--------------------------------------------------------------------*

The above starred cells result in an ALS-XY/WXYZ-Wing pattern, or a type of Disjoint Locked Set called an ADLS since the 2s are contained in 2 sectors instead of only one. Since there is only one "outlaw" candidate and the 2r5c6 sees each 2, it can be eliminated by either ALS rules (WXYZ-Wing als 1 r4c6 -6- r5c357) or ADLS rules.

Code: Select all

 *--------------------------------------------------------------------*
 | 9      38     7      | 345    48     45     | 6      2      1      |
 | 2      5      38     | 6      7      1      | 39     48     349    |
 | 6      4      1      | 38     29     29     | 358    7      35     |
 |----------------------+----------------------+----------------------|
 | 3      67     48     |*48     5      26     | 1      9      27     |
 | 1      9      45     | 7      26-4  *36+4   | 25    *36     8      |
 | 57     68     2      | 9      1     *36+8   | 4     *36     57     |
 |----------------------+----------------------+----------------------|
 | 47     12     9      | 14     3      4678   | 278    5      246    |
 | 8      37     35     | 2      469    45679  | 79     1      469    |
 | 457    12     6      | 145    489    4579   | 2379   48     2349   |
 *--------------------------------------------------------------------*

This results in a type 3 UR r56c68.<36> forming a naked locked pair with cell r4c4 eliminating 4r5c5.
The resulting locked pair <26> in box 5 also eliminates 6r56c6.

Code: Select all

 *--------------------------------------------------------------------*
 | 9      38     7      | 345    48    *45     | 6      2      1      |
 | 2      5      38     | 6      7      1      | 39     48     349    |
 | 6      4      1      | 38     29     29     | 358    7      35     |
 |----------------------+----------------------+----------------------|
 | 3      67     48     | 48     5      26     | 1      9      27     |
 | 1      9     *45     | 7      26     3-4    | 25     36     8      |
 | 57     68     2      | 9      1      38     | 4      36     57     |
 |----------------------+----------------------+----------------------|
 | 47     12     9      | 14     3      4678   | 278    5      246    |
 | 8      37    *3+5    | 2      469   *4679+5 | 79     1      469    |
 | 457    12     6      | 145    489    4579   | 2379   48     2349   |
 *--------------------------------------------------------------------*

The starred cells form a W-Wing. Consider the conjugate 5s in row 8. Depending on which is a five, either r1c6 is a 4, or r5c3 is a 4. Therefore the peer intersection at r5c6 cannot possibly contain a 4.

Singles from there on.

So, you need X-Coloring, UR type 6 and 3, WXYZ-Wing or ADLS and W-Wing in order to solve this without using chains. I don't know if that solution path is any easier, but I find patterns easier to locate manually than chains. It took about 15 minutes to execute the above using Simple Sudoku to assist in coloring, locating the URs and the W-Wing. I actually found the ADLS before I noticed that it was also an ALS-XY or WXYZ-Wing pattern - probably the same one as JasonLion. So, I'm not certain that I agree 100% with eleven that "to find it manually is extremely boring...", but then I had the tips from aran and JasonLion to guide me, so perhaps his point is valid. I usually give up after about 30 minutes of manual effort since my brain is at the bursting point by then.

My point... I don't have a point, especially since I'm from the camp that believes chains are patterns and patterns are chains. My best advice is to study, learn techniques and use something like Simple Sudoku at the very least for tracking candidates and for producing PM grids that are postable.

Cheers,
Paul

Sudoku - it's not just a hobby, it's an affliction...

[eleven]

Nice solution.

However on paper after some steps i would see a puzzle like this one:

Code: Select all

 9  38 7 | .  48 . | 6  2  1
 2  5  38| 6  7  1 | .  48 .
 6  4  1 | 38 29 29| .  7  35
---------+---------+---------
 3  #  48| 48 5  . | 1  69 27
 1  69 45| 7  .  . | 25 #  8
 57 68 2 | 9  1  . | 4  36 57
---------+---------+---------
 47 12 9 | .  3  . | .  5  .
 8  .  . | 2  .  . | .  1  .
 .  12 6 | .  .  . | .  48 .

Here the marked DP 69 forces a 6 in r6c2 or r6c8 (the only cells, where 6 or 9 can be in the 2 boxes), both leading to r4c8=9 and r5c2=9 and eliminating 6 in r6c6.

Code: Select all

 9 *38 7 | .  48 . | 6  2  1
 2  5  38| 6  7  1 | .  48 .
 6  4  1 | 38 29 29| .  7  35
---------+---------+---------
 3  .  48| 48 5  . | 1  69 27
 1  69 45| 7  .  . | 25 .  8
 57 68 2 | 9  1 *38| 4  36 57
---------+---------+---------
 47 12 9 | .  3  . | .  5  .
 8  .  . | 2  .  . | .  1  .
 .  12 6 | .  .  . | .  48 .

Now in this grid its very hard to see, that there are useful coloring eliminations for 8.
More likely e.g. i would look, if there is an (extended) w-wing with the "new" 38 cell in r6c6, which could eliminate the 8 in r6c2, starting with the strong link for 3 in r13c3. This way i had a chance to find the chain r3c4=3->r3c9=5->r5c7=5->r5c4=4->r4c3=8, getting the elimination luckily in another way - with a chain very similar to the ones by daj and aran.

So on paper complex coloring eliminations are very hard (without writing many markers). Normally i dont even look for swordfish - if not the puzzle pattern itself is "swordfish-like" or i am stuck with a puzzle by Wayne Gould

[Maria]

Maria,

Using the starting PM posted by aran plus his DP and the techniques list posted by JasonLion, I arrived at the following solution path:

Code: Select all

 *--------------------------------------------------------------------*
 | 9      38+     7     | 3458-  48-    458-   | 6      2      1      |
 | 2      5      38-    | 6      7      1      | 398-   48+    349    |
 | 6      4      1      | 38+    29     29     | 358-   7      35     |
 |----------------------+----------------------+----------------------|
 | 3      679    48+    | 48-    5      26     | 1      69     27     |
 | 1      69     45     | 7      246    2346   | 25     369    8      |
 | 57     68-    2      | 9      1      368+   | 4      36     57     |
 |----------------------+----------------------+----------------------|
 | 47     12     9      | 148-   3      4678-  | 278+   5      246    |
 | 8      37     35     | 2      469    45679  | 79     1      469    |
 | 457    12     6      | 1458-  498+   45798- | 23798- 48-    2349   |
 *--------------------------------------------------------------------*

Applying X-Coloring to the digit 8, the above parity map can be produced when starting from 8r1c2.
7 Eliminations can be applied starting from 8r1c2 and following conjugate pairs as follows:
a) Starting from row 2 conjs r24c3 & r29c8 - eliminate 8r2c7
b) continuing from col 4 conjs r3c47 & r4c34 - eliminate 8r179c4
c) continuing from row 9 conjs r19c5 & r29c8 - eliminate 8r9c67
d) finish with col 6 conjs r19c5 & b5x19 (or r6c26) - eliminate 8r1c6

Code: Select all

 *--------------------------------------------------------------------*
 | 9      38     7      | 345    48     45     | 6      2      1      |
 | 2      5      38     | 6      7      1      | 39     48     349    |
 | 6      4      1      | 38     29     29     | 358    7      35     |
 |----------------------+----------------------+----------------------|
 | 3     *69+7   48     | 48     5      26     | 1     *69     27     |
 | 1     *69     45     | 7      246    2346   | 25    *69+3   8      |
 | 57     68     2      | 9      1      368    | 4      36     57     |
 |----------------------+----------------------+----------------------|
 | 47     12     9      | 14     3      4678   | 278    5      246    |
 | 8      37     35     | 2      469    45679  | 79     1      469    |
 | 457    12     6      | 145    489    4579   | 2379   48     2349   |
 *--------------------------------------------------------------------*

There is a type 6 UR r45c28.<69> as aran pointed out. It immediately leads to 9 <= r4c8, r5c2 since the 9s are conjugate in both rows/cols and cannot occur in r4c2/r5c8.

Code: Select all

 *--------------------------------------------------------------------*
 | 9      38     7      | 345    48     45     | 6      2      1      |
 | 2      5      38     | 6      7      1      | 39     48     349    |
 | 6      4      1      | 38     29     29     | 358    7      35     |
 |----------------------+----------------------+----------------------|
 | 3      67     48     | 48     5     *26     | 1      9      27     |
 | 1      9     *45     | 7     *246    346-2  |*25     36     8      |
 | 57     68     2      | 9      1      368    | 4      36     57     |
 |----------------------+----------------------+----------------------|
 | 47     12     9      | 14     3      4678   | 278    5      246    |
 | 8      37     35     | 2      469    45679  | 79     1      469    |
 | 457    12     6      | 145    489    4579   | 2379   48     2349   |
 *--------------------------------------------------------------------*

The above starred cells result in an ALS-XY/WXYZ-Wing pattern, or a type of Disjoint Locked Set called an ADLS since the 2s are contained in 2 sectors instead of only one. Since there is only one "outlaw" candidate and the 2r5c6 sees each 2, it can be eliminated by either ALS rules (WXYZ-Wing als 1 r4c6 -6- r5c357) or ADLS rules.

Code: Select all

 *--------------------------------------------------------------------*
 | 9      38     7      | 345    48     45     | 6      2      1      |
 | 2      5      38     | 6      7      1      | 39     48     349    |
 | 6      4      1      | 38     29     29     | 358    7      35     |
 |----------------------+----------------------+----------------------|
 | 3      67     48     |*48     5      26     | 1      9      27     |
 | 1      9      45     | 7      26-4  *36+4   | 25    *36     8      |
 | 57     68     2      | 9      1     *36+8   | 4     *36     57     |
 |----------------------+----------------------+----------------------|
 | 47     12     9      | 14     3      4678   | 278    5      246    |
 | 8      37     35     | 2      469    45679  | 79     1      469    |
 | 457    12     6      | 145    489    4579   | 2379   48     2349   |
 *--------------------------------------------------------------------*

This results in a type 3 UR r56c68.<36> forming a naked locked pair with cell r4c4 eliminating 4r5c5.
The resulting locked pair <26> in box 5 also eliminates 6r56c6.

Code: Select all

 *--------------------------------------------------------------------*
 | 9      38     7      | 345    48    *45     | 6      2      1      |
 | 2      5      38     | 6      7      1      | 39     48     349    |
 | 6      4      1      | 38     29     29     | 358    7      35     |
 |----------------------+----------------------+----------------------|
 | 3      67     48     | 48     5      26     | 1      9      27     |
 | 1      9     *45     | 7      26     3-4    | 25     36     8      |
 | 57     68     2      | 9      1      38     | 4      36     57     |
 |----------------------+----------------------+----------------------|
 | 47     12     9      | 14     3      4678   | 278    5      246    |
 | 8      37    *3+5    | 2      469   *4679+5 | 79     1      469    |
 | 457    12     6      | 145    489    4579   | 2379   48     2349   |
 *--------------------------------------------------------------------*

The starred cells form a W-Wing. Consider the conjugate 5s in row 8. Depending on which is a five, either r1c6 is a 4, or r5c3 is a 4. Therefore the peer intersection at r5c6 cannot possibly contain a 4.

Singles from there on.

So, you need X-Coloring, UR type 6 and 3, WXYZ-Wing or ADLS and W-Wing in order to solve this without using chains. I don't know if that solution path is any easier, but I find patterns easier to locate manually than chains. It took about 15 minutes to execute the above using Simple Sudoku to assist in coloring, locating the URs and the W-Wing. I actually found the ADLS before I noticed that it was also an ALS-XY or WXYZ-Wing pattern - probably the same one as JasonLion. So, I'm not certain that I agree 100% with eleven that "to find it manually is extremely boring...", but then I had the tips from aran and JasonLion to guide me, so perhaps his point is valid. I usually give up after about 30 minutes of manual effort since my brain is at the bursting point by then.

My point... I don't have a point, especially since I'm from the camp that believes chains are patterns and patterns are chains. My best advice is to study, learn techniques and use something like Simple Sudoku at the very least for tracking candidates and for producing PM grids that are postable.

Cheers,
Paul

Sudoku - it's not just a hobby, it's an affliction...

My friends, tomorrow morning I will check for a step lower and see whether I can move better, really too hard for me to follow - I will keep you posted if still stopped somewhere, thank you Paul

[Maria]

Nice solution.

However on paper after some steps i would see a puzzle like this one:

Code: Select all

 9  38 7 | .  48 . | 6  2  1
 2  5  38| 6  7  1 | .  48 .
 6  4  1 | 38 29 29| .  7  35
---------+---------+---------
 3  #  48| 48 5  . | 1  69 27
 1  69 45| 7  .  . | 25 #  8
 57 68 2 | 9  1  . | 4  36 57
---------+---------+---------
 47 12 9 | .  3  . | .  5  .
 8  .  . | 2  .  . | .  1  .
 .  12 6 | .  .  . | .  48 .

Here the marked DP 69 forces a 6 in r6c2 or r6c8 (the only cells, where 6 or 9 can be in the 2 boxes), both leading to r4c8=9 and r5c2=9 and eliminating 6 in r6c6.

Code: Select all

 9 *38 7 | .  48 . | 6  2  1
 2  5  38| 6  7  1 | .  48 .
 6  4  1 | 38 29 29| .  7  35
---------+---------+---------
 3  .  48| 48 5  . | 1  69 27
 1  69 45| 7  .  . | 25 .  8
 57 68 2 | 9  1 *38| 4  36 57
---------+---------+---------
 47 12 9 | .  3  . | .  5  .
 8  .  . | 2  .  . | .  1  .
 .  12 6 | .  .  . | .  48 .

Now in this grid its very hard to see, that there are useful coloring eliminations for 8.
More likely e.g. i would look, if there is an (extended) w-wing with the "new" 38 cell in r6c6, which could eliminate the 8 in r6c2, starting with the strong link for 3 in r13c3. This way i had a chance to find the chain r3c4=3->r3c9=5->r5c7=5->r5c4=4->r4c3=8, getting the elimination luckily in another way - with a chain very similar to the ones by daj and aran.

So on paper complex coloring eliminations are very hard (without writing many markers). Normally i dont even look for swordfish - if not the puzzle pattern itself is "swordfish-like" or i am stuck with a puzzle by Wayne Gould

Eleven - as I have told Paul I am reducing the level one step below and see what happens, will come back - have a nice weekend

[aran]

You saw, how aran solved it, which looks straightahead, but he also was lucky to find a useful starting point in the cell, where he eliminated a candidate

Eleven
I might put it like this : the deadly rectangle deduction draws attention to that cell, which isn't a bad reason for then considering it in the next move. The luck was of course that this proved decisive.