From Newsgroup: rec.puzzles

On 28/11/2025 02:07, Mike Terry wrote:

On 27/11/2025 20:42, Richard Heathfield wrote:

On 27/11/2025 18:30, Mike Terry wrote:

242 lines - understandable, but I'm applying a liberal axe.

On 27/11/2025 16:26, Richard Heathfield wrote:

On 27/11/2025 04:14, Mike Terry wrote:

On 26/11/2025 23:34, Richard Heathfield wrote:

<snip>

a4b
c7d
efg

Out of the six squares

a4b
c7d

a maximum of 4 can be black, so the 7 needs to find 3 more black squares from somewhere else,
and efg are the only candidates.

Ah, right - that's valid.a (I was misunderstanding the notation.)a With my notation that would be:

aa ..a .4a ..aaaaa ..a .4a ..
aa ..a .7a ..a ->a ..a .7a ..
aa ..a ..a ..aaaaa B.a B.a B.

Aha! Reminiscent of DOS TUI attribute bytes.

That's a genuine pattern beyond what the rules say directly!a The puzzles I looked at did not
require such advanced logic to solve...

It's tempting to cut code looking for patterns.

The overall feeling for the game is very like MineSweeper. Start from easily deduced squares,
and then gradually work out along the edges of the known area.a It's not like sudoku at all.

<nod>

For my environment (Windows with MS Edge browser) the right- click to set a square does not
work properly - it does set the square, but also still displays the standard (MS Edge) right-
click context menu.

KS may apprec a bug report?

Yeah, I'll look for a place to do that.

moc.srehsabniarb@emtctanoc is (I think) exactly wrong.

How about this puzzle I've just invented:a (I'll call the Sequidoku puzzles)

<snip>

NOT MUCH FUN, but no different really from Surroundoku. :(

I have yet to discover a way to frame the one obvious rule in a way that gives me the solution.

You mean for my Sequidoku puzzle?

No. I meant for SurrounDoku.

For Surroundoku the equivalent rule would be

1.a If the game constraints are directly violated, if a square is coloured anything
aaaa but one specific colour, then you can mark the square with that colour.

I don't see how that moves us forward.

Well, it allows us to fill in /some/ squares, right?a That's (limited) progress!

For example:

aa...aaa WWW
aa.0. -> WWW
aa...aaa WWW

(Making any of the W squares B would directly violate the constraint that the 0 square has 0
black squares as neighbors.)

Sudoku's pretty shit too:

+---+---+---+
|953|8.6|124|
|162|435|8.9|
|48.|921|365|
+---+---+---+
|398|56.|412|
|624|198|53.|
|5.1|243|986|
+---+---+---+
|816|.54|293|
|235|619|.48|
|.49|382|651|
+---+---+---+

You may be being facetious, but you are right /for the given puzzle/:a it can be completed by
applying one simple pattern, which is a direct application of the standard Sudoku constraints.a If I
saw that puzzle in Metro, I wouldn't even bother completing it,a It's pretty shit as you say.

The point is that the rule can (and is) used in solving /all/ sudokus, but it is not /sufficient/
for any worthy puzzle.a It may enable us to fill in a handful of squares then we're stuck - further
reasoning and other patterns will be required.a It is this variety that makes Sudoku puzzles enjoyable.

With Suroundoku it seems to me that the equivalent rule is [might be?] /sufficient/ to solve all
Suroundoku puzzles published on BB.a You're not convinced, and now the puzzles have gone so we won't
be able to confirm either way!

<..snip..>

I guess the only way to check here is for you to post a position where you're stuck, and we'll
see if there are any "direct constraint" patterns that you've missed.

Works for me... if KS ever sees fit to restore the puzzle to his site. (I presume he's had some
issues.)

Maybe lots of people said the right-click button didn't work :)a Let's see if they come back...

They're back!! But with a new name: SurroundNine

<https://www.brainbashers.com/surroundnine.asp>

It looks the same as the old game which was withdrawn, but the right-click bug has been fixed.

Regarding my claim that it seems all the puzzles can be solved by repeated application of one simple
mechanical rule - that still appears to be the case. Since some time has ellapsed and servers may
have dropped old posts etc., I'll summarise what I said back then:

Summary
=======

Reminder of game objective: Mark all cells in the given grid black or White, so that the constraint
below is satisfied:
- If a cell contains a number, that number must match the number of
Black cells appearying in that cells "adjacency block".

[The adjacency block is all the cells that are directly adjacent to the cell, either horizontally,
vertically or diagonally, plus the cell itself. So an interior cell has a 3x3 block of 9 adjacent
cells, while a cell on the edge has an adjacency block of 6 cells, and a corner cell has a block
with 4 cells. We could also describe it as all the cells reachable from the starting cell in zero
or one Chess king moves.]

Example puzzle with 3x3 grid:

. . .
6 . 3
. . .

Solution

B B W
B B W
B B W

My proposed mechanical solving rule is applied to a grid which is partially solved. (All cells in
the grid will be in one of the states B, W, . indicating the cell is Black, White, or Undecided.)

Rule(s):
- If a cell is currently n Black cells short of meeting its B-cell constraint, and
currently has exactly n Undecided cells in its adjacency block, then
MARK ALL THE UNDECIDED CELLS BLACK.
- If a cell currently has n Black cells in its adjacency block, and
has a B-cell count constraint of exactly n, then
MARK ALL THE UNDECIDED CELLS WHITE.

Those rules are completely obvious; they effectively say "if the cell has a constraint implying all
the undecided cells must be B or W to comply, mark the undecided cells appropriately". It took me
much longer to write the rule down precisely than to understand it!

Example rule application:

. . . B B . B B W
6 . 3 -> B B 3 -> B B W
. . . B B . B B W

[1st application is to the edge cell marked 6, which has exactly 6 undecided cells in its block, so
all those cells must be B. 2nd application is to the edge cell marked 3, which needs 3 B cells in
its block and at this point /has/ 3 B cells in its block, so all undecided cells in the block must
be W. Er, that's it!]

We've solved the puzzle only using applications of my mechanical solving rule. Yay!

My previous claim was that ALL the SurroundNine puzzles on the website can be solved in a similar
fashion. Since the rule is so simple, being just a single direct application of the given game
constraint, this makes the puzzles rather uninteresting. (Akin to WordGrid "puzzles" where you have
to find a list of given words in a given letter grid - we all know we can "solve" these puzzles by
just starting with the first word and scanning the grid from top-left to bottom-right, checking in
all directions until we inevitably find the word, then go on to next given word etc. 100%
mechanical, with no original thought required...)

Richard was unsure whether all the puzzles could be solved this way, but at that point Kevin
withdrew the puzzles, so we couldn't investigate further. Now the puzzles have reappeared, maybe
Richard will find one of the puzzles (on the web site) that cannot be solved just using my
mechanical rule? (If he finds one and gives me a link, I can check the puzzle. On several of the
puzzles I've tried, I /nearly/ decided that it couldn't be solved using just my simple rule, only to
eventually spot how the rule could be applied somewhere I'd simply missed!)

To be clear, I'm not suggesting /all/ SurrondNine puzzles can be solved with my rule - just all the
puzzles on the given web-site, so perhaps it's a feature of the automatic puzzle-generator the web
site uses?

Mike.

--- Synchronet 3.22a-Linux NewsLink 1.2