The New Sudoku Players' Forum

[tarek]

I have actively been playing game3.

The way I extract ratings from the result.txt file is horrible. I would be interested to see how everybody else does it (I will not say how I do it because it is very very very basic)

tarek

[g.r.emlin]

::: game 0004 start at 2007-12-20+23:00:00-0000 :::

Code: Select all

dealer status

candidate patterns (via private message to g.r.emlin)

   3 JPF       89 g.r.emlin

past patterns

   1 m_b_metcalf     1 ravel          1 wintder    2 Mauricio
   2 g.r.emlin       12 JPF         15 TTHsieh


[m_b_metcalf]

::: game 0004 start at 2007-12-20+23:00:00-0000 :::

At which time I'll be en route to JPF's delightful country. Merry Xmas folks.

Mike Metcalf

[JPF]

I have actively been playing game3.

The way I extract ratings from the result.txt file is horrible. I would be interested to see how everybody else does it (I will not say how I do it because it is very very very basic)

I hope I understood your question :

In the result.txt file of SE, every puzzle :
starts with a line like

Code: Select all

Analyzing Sudoku #3

and is rated in the last line :

Code: Select all

Difficulty: 9.4

Even with a short Excel VBA program you can extract #3 and 9.4

JPF

[JPF]

At which time I'll be en route to JPF's delightful country. Merry Xmas folks.

You can play in France too

Bon voyage et Joyeux Noel.

JPF

[TTHsieh]

You can play in France too

No, he cannot.
But he can play on the airplane

TTHsieh

[TTHsieh]

The game rule was updated by g.r.emiln

TTHsieh

[JPF]

The game rule was updated by g.r.emiln

... I need a suggested value ...

IMO,
SCORE_PEARL_WEAK = 25
SCORE_PEARL_STRONG = 50

I agree.
g.r.emiln or g.r.emlin, whoever, still keeps SCORE_PEARL_STRONG = 100

JPF

[TTHsieh]

I personnaly think that SCORE_PEARL_STRONG=100 is too high.

this adds a bit of strategy

This adds many strategies. If the SCORE_PEARL_STRONG=100.
I think it will lead the ratings game to perals game.

g.r.emlin, Do you hope so

TTHsieh

[Ruud]

It would be fun to see this game turn into a real strategy game. At this moment it's mainly a game of chance (finding puzzles with the right SE rating), CPU power and programming skills to optimize the search routines.

Some ideas to introduce strategy elements:

Penalty for a submission lower than your previous submission. You then have to decide when to claim that high rating puzzle, risking it being claimed by another participant if you wait too long.

Bid for SE ratings at the beginning of the game. Sort of like bridge. You will then only score points for the ratings that you claimed.


Adding some complexity to the rules does not harm, as long as it makes the game more challenging. People still play bridge, chess, cricket, snooker, American football, baseball, tennis, WoW...

Ruud

[gsf]

The game rule was updated by g.r.emiln

the first post (the rules post) is the only one not generated by g.r.emlin
the coding isn't that good!

[gsf]

I personnaly think that SCORE_PEARL_STRONG=100 is too high.

this adds a bit of strategy

This adds many strategies. If the SCORE_PEARL_STRONG=100.
I think it will lead the ratings game to perals game.
g.r.emlin, Do you hope so

It would be fun to see this game turn into a real strategy game. At this moment it's mainly a game of chance (finding puzzles with the right SE rating), CPU power and programming skills to optimize the search routines.

Some ideas to introduce strategy elements:

Penalty for a submission lower than your previous submission. You then have to decide when to claim that high rating puzzle, risking it being claimed by another participant if you wait too long.

Bid for SE ratings at the beginning of the game. Sort of like bridge. You will then only score points for the ratings that you claimed.

the idea is to retain the best parts of the patterns game, but like ruud says, add some
strategy, so that making a submission takes some thought

the penalty for a submission lower than the highest submission by the same player sounds interesting and workable

in general I think any strategy that has a risk/reward for holding onto submissions
should make it more interesting to play because it will downplay brute force strategies

bidding would be a complication for the referee

I didn't modify SCORE_PEARL_STRONG yet because no one has made a pearl submission in this thread

what we can do is play another game or two, look at the results, and change the
SCORE_PEARL_* values post-facto, based on what the consensus deems fair

what I can also do is add a SCORE_LOWER adjustment for ruud's suggestion
and show how the results of the previous games would change with SCORE_LOWER in effect

[m_b_metcalf]

Bid for SE ratings at the beginning of the game. Sort of like bridge. You will then only score points for the ratings that you claimed.

I think this unworkable. With bridge you know your cards after the deal at the beginning of the game. One of the delights of the Patterns Game is finding a high-valued puzzle well after the beginning. An example of this occured in just the last game with JPF's last two or three submissions. Anyway, what's to stop someone bidding everything from 1.0 to 12.0? A fine for not playing them?

Regards,

Mike Metcalf

[m_b_metcalf]

Penalty for a submission lower than your previous submission. You then have to decide when to claim that high rating puzzle, risking it being claimed by another participant if you wait too long.

This aready exists. If I generate a very high value in the first few minutes, when do I play it? Immediately for 5 points, or wait a while for more points only to see someone else play it first.

Regards,

Mike Metcalf

[m_b_metcalf]

It would be fun to see this game turn into a real strategy game. At this moment it's mainly a game of chance (finding puzzles with the right SE rating), CPU power and programming skills to optimize the search routines.

Adding some complexity to the rules does not harm, as long as it makes the game more challenging. People still play bridge, chess, cricket, snooker, American football, baseball, tennis, WoW...

I agree about your general aim, but note that, over the whole of the first game series, there were perhaps in all only less than a dozen players, with only half or less of that number in any given game.

Are there any comments from other non-players about how to attract more participation?

Regards,

Mike Metcalf