The New Sudoku Players' Forum

[JPF]

For those who like that sort of thing, here are three pearls that I fished on the way
...

Yes, well done.

This one is a pearl too :

Code: Select all

100200003045000060000000070000302001000060000800104000050000000070000450200009008 #43 = 9.7 , JPF


JPF

[TTHsieh]

• ...the exact start time [GMT].
• ...we have a few 10+ puzzles.
  
  • OK, GMT.
  • I will propose a symetrical pattern that have many puzzles with ER > 10.0 for the next game.

  I suggest to reduce this period of time to 2 days.

  I agree.
  TTHsieh

[JPF]

I will propose a symetrical pattern that have many puzzles with ER > 10.0 for the next game.

How do you know that

The spirit of the game is that the dealer starts without any real advantage compared with his competitors.
It's especially important if ERs>10+

If not, it's a fool's game

Could we agree to select a random pattern* (with symmetry or not) as soon the game is open by the dealer ?

* in a large database, for instance.

JPF

[TTHsieh]

How do you know that

The pattern (game #2) give me a strong hint.

This one should be easier :

Code: Select all

 4 . . . . . . . 5
 . 9 . 7 . . . 2 .
 . . 2 . . . 4 . .
 . 8 . 1 . 4 . . .
 . . . . 6 . . . .
 . . . 2 . 5 . 1 .
 . . 1 . . . 2 . .
 . 7 . . . 8 . 6 .
 5 . . . . . . . 8



21 clues, all singles, 16 steps.

JPF

400000005090700020002000400080104000000060000000205010001000200070008060500000008, 1.5, JPF
400000005070300060002000400080104000000070000000205010001000200030008050500000008, 7.1, m_b_metcalf
100000002030400050002000600070306000000050000000802040006000700040007090900000008, 9.0, JPF
700000003090400080001000500080905000000030000159069080002080600050001000600090403, 9.8, m_b_metcalf
700000003020400080001000500080905000000030000000208060005000100040009020300000007,10.7, JPF

• The rule says that the dealer must prepare a new pattern before a new game. agree ?
• I design a new pattern base on the pattern as above and try to generate some puzzles.
• I find many puzzles with ER > 10.0
• Do I violate the rules ?

Could we agree to select a random pattern* (with symmetry or not) as soon the game is open by the dealer ?

I agree, but no dealer.

TTHsieh

[JPF]

Do I violate the rules ?

TTHsieh,
Don't misinterpret what I was trying to point out.

You are doing a great job in this thread ; you are a formidable and regular competitor.
I love it.

Actually, the rules are more unbalanced since I proposed to calculate the score as the sum of the # of the posts ; the 20 pts penalty doesn't solve the problem.

Ideally, the dealer should not play, as wintder did... but it's a shame that someboby can't play.

Could we agree to select a random pattern* (with symmetry or not) as soon the game is open by the dealer ?

I agree, but no dealer.

OK, what are the other players'suggestions ?

JPF

[TTHsieh]

My conception :

• Build a patternbase that has 100~500 patterns, open to public.
• The champion assign someone to select patterns list (5~10 patterns) from patternbase.
• And assign someone else to select a pattern from the patterns list.
• When a pattern is selected, The new game starts.

TTHsieh

[wintder]

I would be willing to be a starter.

I don't have the ability to check many times a day but I do check several times a day.

I have a few patterns already that balance the time between generation and favoring minimal.

My thought is..you let me know when you want a new pattern and I will post it. I won't play and the patterns will be "new".

As for tens, if someone has a pattern to generate 10s everyone would use it. I have not heard of such.


I can post scores and such once a day for sure, more often is not likely, so for accounting you are all faster than I. (And better at math.)

[gsf]

My conception :

• Build a patternbase that has 100~500 patterns, open to public.
• The champion assign someone to select patterns list (5~10 patterns) from patternbase.
• And assign someone else to select a pattern from the patterns list.
• When a pattern is selected, The new game starts.

TTHsieh

the real problem is a dealer who starts with a stacked deck
(a cache of puzzles matching the rules for the dealt pattern before dealing)
publishing all patterns will at least level the playing field for a deck-stacker by allowing
all players to stack their own decks and lay in wait for games for which they have cached decks

it would be fairest for the dealer to not participate in the game
in that case it would make sense to have a different dealer each game
the dealer could then start with a stacked deck
and offer hints like "a 10.1 is possible" or "I'm still waiting for SE to crunch one of mine"
and the rules could even specifiy a time-related constraint on dealt patterns, like

• the dealer certifies that with no a priori knowledege of the pattern it is possible to find/generate at least N valid entries in at most T
  

where T is in some unit of elapsed time

I'm checking now to see if there is a way to glean the average player elapsed time between submissions
explicitly omitting dealer elapsed times just in case (but keeping them anonymous)

if that works (i.e., it can be scripted and done without manual intervention)
then we may be able to (subjectively) correlate player submission elapsed times to good games

[m_b_metcalf]

I'm checking now to see if there is a way to glean the average player elapsed time between submissions
explicitly omitting dealer elapsed times just in case (but keeping them anonymous)

That may be futile, even if it appears to 'work'. Given that the players are in different time zones, sometimes go to bed for eight hours (or go to work, fetch children from school etc.), the intervals are subject to large variations other than those inherent to the game itself.

Regards,

Mike Metcalf

P.S. I'm happy with any rules that are (a) very simple, (b) involve no other external program than SE.

[JPF]

I'm still trying to find a 60 th rating... just for the fun...

I didn't succeed
For me, the game is over.

OK, what are the other players'suggestions ?

Thanks TThsieh, wintder, gsf, m_b_metcalf for your contribution.

The smart solution to the problem is not easy.
I will propose something in the next coming days.

For the next game (#38), TTHsieh you are the dealer.

I suggest to reduce this period of time to 2 days.

I agree.

JPF

[TTHsieh]

New Game (#38) will be started in GMT 2007/12/03 12:00 pm ~ 13:00 pm.

TTHsieh

[gsf]

I'm checking now to see if there is a way to glean the average player elapsed time between submissions
explicitly omitting dealer elapsed times just in case (but keeping them anonymous)

That may be futile, even if it appears to 'work'. Given that the players are in different time zones, sometimes go to bed for eight hours (or go to work, fetch children from school etc.), the intervals are subject to large variations other than those inherent to the game itself.

I'm more interested in strings of short gaps rather that large gaps

also
it would be nice to specify invariant and otherwise unused strings for the first and last post of a game
with the game number if the first post

[TTHsieh]

New Game (#38)

Code: Select all

 1 . . | . . . | . . 2
 . 2 . | 1 . . | . 3 .
 . . 4 | . . . | 5 . .
 - - - + - - - + - - -
 . 3 . | . 6 . | . . .
 . . . | 7 2 5 | . . .
 . . . | . 4 . | . 1 .
 - - - + - - - + - - -
 . . 6 | . . . | 4 . .
 . 8 . | . . 3 | . 9 .
 2 . . | . . . | . . 6   ER=1.5



Code: Select all

100000002020100030004000500030060000000725000000040010006000400080003090200000006

TTHsieh

[m_b_metcalf]

Off we go:

Code: Select all

 1 . . . . . . . 2
 . 2 . 9 . . . 3 .
 . . 4 . . . 5 . .
 . 9 . . 6 . . . .
 . . . 7 2 5 . . .
 . . . . 4 . . 1 .
 . . 6 . . . 4 . .
 . 8 . . . 3 . 9 .
 2 . . . . . . . 6  SE 8.3

Regards,

Mike Metcalf

[TTHsieh]

Code: Select all

 1 . . | . . . | . . 2
 . 2 . | 1 . . | . 3 .
 . . 4 | . . . | 5 . .
 - - - + - - - + - - -
 . 1 . | . 2 . | . . .
 . . . | 4 6 5 | . . .
 . . . | . 3 . | . 7 .
 - - - + - - - + - - -
 . . 5 | . . . | 6 . .
 . 3 . | . . 8 | . 2 .
 6 . . | . . . | . . 4   ER=7.2



TTHsieh