Home » Forum Home » Analysis

Topic: Pente - Understand vs Memorize
Replies: 122   Views: 260,340   Pages: 9   Last Post: Nov 17, 2010, 10:26 AM by: zoeyk

Search Forum

Back to Topic List Topics: [ Previous | Next ]
Replies: 122   Views: 260,340   Pages: 9   [ Previous | 1 2 3 4 5 6 7 8 9 | Next ]
jasonb

Posts: 105
Registered: Jan 3, 2010
Home page
Re: Pente - Understand vs Memorize
Posted: Oct 8, 2010, 5:31 AM

"once we get further down the thread ide be interested to see someone try and create a summery of the valued information found in this thread from all players involved."

Funny you should mention that Zoey, I was thinking the same thing. I'd be happy to take a stab at it.


zoeyk

Posts: 2,071
Registered: Mar 4, 2007
From: San Francisco
Age: 43
Home page
Re: Pente - Understand vs Memorize
Posted: Oct 8, 2010, 5:38 AM

Image and video hosting by TinyPic

Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
zoeyk

Posts: 2,071
Registered: Mar 4, 2007
From: San Francisco
Age: 43
Home page
Re: Pente - Understand vs Memorize
Posted: Oct 8, 2010, 6:41 AM

Some will say a 2 jump (G10) is stronger than a special jump (J7). both sides of the argument have yet to prove which is the truth in a general forum. so, this is just my opinion that J7 is more logical. my reason is the 1 jump pair seems to have stronger Tempo than the Draw move (pair).

G10 and O10 = Level 3.5 strength
J7 = Level 4.0 strength

Image and video hosting by TinyPic

Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
jasonb

Posts: 105
Registered: Jan 3, 2010
Home page
Re: Pente - Understand vs Memorize
Posted: Oct 8, 2010, 7:06 AM

Wow, that drawing is exactly the kind of thing I was thinking would help this thread . . . a pictorial of potential lines and an evaluation of those possibilities. Excellent! Those lines are what I'm starting to learn to visualize . . . slow going though.

By "1 jump pair", did you mean K10,J7 same as "special jump"?

zoeyk

Posts: 2,071
Registered: Mar 4, 2007
From: San Francisco
Age: 43
Home page
Re: Pente - Understand vs Memorize
Posted: Oct 8, 2010, 7:48 AM

Here we are looking at the N8 or "Allen's" opening.
and for P2's second we are looking at the basic key stone attack. very similar in principle to the O9 opening shown above.

keep in mind you are not limited to using K12 as your 3rd move as Player 1, but if you are going to use it as your 3rd, these are some second moves as P1 that you might consider for logical reasons.

G10 = Lvl 2.0
N12 = Lvl 3.5
N11 = Lvl 4.0



Image and video hosting by TinyPic

Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
zoeyk

Posts: 2,071
Registered: Mar 4, 2007
From: San Francisco
Age: 43
Home page
Re: Pente - Understand vs Memorize
Posted: Oct 8, 2010, 8:06 AM

by 1 jump i mean K10 M8.

or X _ X


a 2 jump is X _ _ X

a 3 jump (or a "Post") is X _ _ _ X

a Post 3 is X_X_X




a special jump is

_ X _
_ _ _
_ _ _
X _ _





a knight's jump is

_ X _
_ _ _
_ _ X




a knight's draw "A" is

X X _
_ _ O
_ _ X



or a knight's draw "B" is

_ X _
_ _ O
_ X X







a pawn broker is

_ X _
_ _ O
X X X





a upper T is

X _ X
_ _ _
_ _ _
_ X _




a lower T is

_ _ X _
_ _ _ _
X _ _ _
_ _ _ X



a one jump triangle is


X _ X
_ _ _
X _ _



a two jump triangle is

X _ _ _
_ _ _ _
_ _ _ _
X _ _ X



a H is

X _ X
X X X
X _ X



a pair is

X X


or


X _
_ X



a draw move is, but not limited too this shown example.
(too many examples to type here, use imagination)
a draw move is a pair positioned such as to tempt
the opponent into a capture trade.


X X _
_ _ O
_ _ _
_ _ O




if you need any other shape terms shown let me know.

Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
zoeyk

Posts: 2,071
Registered: Mar 4, 2007
From: San Francisco
Age: 43
Home page
Re: Pente - Understand vs Memorize
Posted: Oct 8, 2010, 8:31 AM

i have a feeling that a good test for a "grand master to be" would be for him to use a level one move once per game in his P1 opening while rated and win 90+% of the time. and no i doubt i'll be attempting that any time soon but ide have fun watching nosovs try that. and im sure he could pull it off.

Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
zoeyk

Posts: 2,071
Registered: Mar 4, 2007
From: San Francisco
Age: 43
Home page
Re: Pente - Understand vs Memorize
Posted: Oct 8, 2010, 9:00 AM

here is an example of opening conversion to reduce P1's 12 2nds to lower than 12. this consolodates the amount of seconds thus a reduction in openings to study and learn and remember. working smarter, not harder. this is only a tip, but not too overly important. there are other ways to go with the K7 open of equal strength. it just means more data to retain for openings. becuase one extra second, equals 30 to 40 more 3rds to learn. makes sense?


Image and video hosting by TinyPic

Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
jasonb

Posts: 105
Registered: Jan 3, 2010
Home page
Re: Pente - Understand vs Memorize
Posted: Oct 8, 2010, 6:50 PM

That makes complete sense Zoey. Your examples are excellent and demonstrate visualization in a clear concise manner. I am sure they will help many that are trying to become better at this game. As Classy noted, we all have different learning styles; drawings like the last 3 you posted make learning easier, IMHO.

http://www.pente.org/gameServer/forums/thread.jspa?forumID=27&threadID=4641&start=32&tstart=0
http://www.pente.org/gameServer/forums/thread.jspa?forumID=27&threadID=4641&start=34&tstart=0
http://www.pente.org/gameServer/forums/thread.jspa?forumID=27&threadID=4641&start=37&tstart=0

zoeyk

Posts: 2,071
Registered: Mar 4, 2007
From: San Francisco
Age: 43
Home page
Re: Pente - Understand vs Memorize
Posted: Oct 8, 2010, 7:05 PM

i'll try to add a methodical check list of thought process that players can use for any style when i have time. the moves ive shown here in the pictures are not the be all and end all of moves to try. i didn't post every thing that is doable because it would just be a confusing clutter of moves i think. some might worry that if im just handing the answers the players will fail to think of why are they good. understanding why is more important than knowing what. so i'll try to explain the whys.

Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
zoeyk

Posts: 2,071
Registered: Mar 4, 2007
From: San Francisco
Age: 43
Home page
Re: Pente - Understand vs Memorize
Posted: Oct 8, 2010, 8:34 PM

ok step 1.

before we even get into the whole deal you will need to make your first move as P1. actually your second stone sinse K10 was your first.

it could be a multitude of choices. some are better than others of course.

any how, so P2 makes a move, could be any thing right?

now we need to look at something. we need to see how it Angeles to attack K10. we must understand that K10 is required to support P1's heavy advantage.

so when picking your 2nd move here are the 2 things your going to try and see.

u look at a potential move that you like for you second, and you ask your self, "if P2 attacks my K10 stone, what 2nd and 3rd and 4th, will help me to do one of the following two things;

either defend K10 from capture,

or

transfer K10's energy into the 4th stone and there by allowing K10 to be captured.

if you allow K10 to be captured, and its energy (or Tempo)
is not correctly transferred to the 4th stone,
one of 2 things will happen.

either;

you will lose tempo, (like having dead weight chained to your leg) thus slowing down and complicating your resolve,

or

you will simply be in a sure loss position.


for this i will need to post some visual examples im sure.
but any ways, this is the first step. because if you cant get out of the starting gate with out tripping on your shoe laces, theres usually little point to continue running to catch up. keep your shoes tied.

(note that; K10's energy can be split 50-50 into 2 stones at a further point in some situations.
and also note that; there comes a point where you can let go of K10's energy all together. to put this into an analogy;
think of a rocket ship going into space, it needs the big heavy bulky fuel tanks to get into orbit, but once in orbit the rocket can let go of the fuel tanks, and begin using a different form of perpulsion. this new form of fuel, and when it can start being used will be explained later. this is not always the case, but it does occure.
and, P2 is the gravity trying to pull the ship down, also P2 is the asteroids gliding out in space as obsticals to avoid. some moving faster than others )


now that this has been explained (minus visual of course),
there are actually more things to consider when choosing your second move.

to be continued.....


Message was edited by: zoeyk at Oct 8, 2010 3:11 PM

Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
alisontate

Posts: 157
Registered: Nov 27, 2008
Age: 30
Re: Pente - Understand vs Memorize
Posted: Oct 11, 2010, 9:37 AM

zoey - I am glad you liked those two paragraphs
I like your drawings - very snappy.

Some thoughts on your checklist idea Jason.

Seems to me what we are talking about with this checklist is actually something like an algorithm or procedure. You know - 'if this, then do that'.

I see this as a decision tree, branching out into various considerations depending on the prevailing situation. I think it would be useful to structure it in this way for the reason that obviously not all situations warrant consideration of all factors.

Following on from Watsu's interesting comments, I thought it would be useful to quantify this checklist concept in AI programing terms.

This 'checklist' approach (although not called that) has been tried in Chess programs and is known as a 'top-down' approach. These programs are generally unsuccessful for an AI program when compared to bottom-up methods, but humans tend to use top-down and bottom-up combined.

Such a program would need to have the analysts perform some sort of quantification and classification of the items on the checklist and relate them in some way to game situations. The program would look for particular situations in a pattern matching approach, and then choose from a list of Factors (considerations) deemed as relevant to that particular Situation, and not consider things that are not relevant.

If I may wax mathematical for a moment....
There could be a set of factors F = {F1, F2...Fn) where each of F1..Fn are factors to consider. And there could be a set S = {S1, S2...Sn) where S1..Sn are situations where some or all of the members of set F could be considered.

For any situation S there could be bound a subset F of S looking like this FS1 = {F1, F5, F10) meaning that factors to be considered in situation 1 (S1) are factors F1, F5 and F10. Finally a set B being the set of bound Factors and Situations B = {Bfs1, Bfs2...Bfsn).

Once these are all defined you can set up a decision tree accordingly. Then work up some code to suit.

Using Pascal and a couple of programmers we wrote a top-down program to play Othello and pitted this against a bottom up program written by another team. Our program could make good decisions about 80% of the time but would invariably make a clanger along the way. We tweaked it and tweaked it, adding ever more factors to consider and situations to match but all this did was slow it down and did not improve its play that much. The problem was that rules of thumb are just that - imprecise. There are always unique factors, or at least a unique balance of factors to consider, and you just cant cover all of the possibilities.

In Chess programs the situation is more complex than Othello, and top down approaches have generally failed badly. Bottom-up search wins hands down.

So, what would happen with Pente with such a program? I suspect that when compared with Chess which in my opinion is more complex than Pente, a top-down Pente AI would do better against an average human player but be crushed by a Master, and it would not do at all well against a bottom-up AI program. If anything, the situation in Pente would be worse than with Chess because the search depth and width of the game tree are much smaller and it is conceivable that an AI for Pente could exhaust the game tree to 14 ply and therefore past the length of 90% of perfectly played games. In this situation a top-down rule of thumb approach would likely lose every game.

So, what can we conclude from this as human players? My thinking is this: By all means have a checklist, understand it, use it and make it part of your play-by-play process, but do not rely on it as it will inevitably fail you if used on its own. It is clear to me that to be a Master player, there is no substitute for knowing a large number of opening lines and their responses, a comprehensive grasp of shapes and how to use them, an undertanding or tempo and initiative and a sound tactical and positional understanding of the game, and last but not least - play lots of Pente. This approach is true of Chess and I expect is true of all zero-sum abstract games.

Alison

jasonb

Posts: 105
Registered: Jan 3, 2010
Home page
Re: Pente - Understand vs Memorize
Posted: Oct 11, 2010, 8:13 PM

Alison,

I guess my point of asking the experts for thier checklist, rules of thumb, thought process, etc. before making a move is I know it evolves as we learn. I have several things on my checklist that weren't there, or even part of my knowledge base when I first started. For instance, protecting keystones, when to extend, "fukumi"s, and traps. I am certain there are many more on the master's lists that I have yet to ponder; some of which are better explained with diagrams, such as Zoey's recent posts.

Your decision tree sounds interesting, perhaps it would work . . . I don't know.

zoeyk

Posts: 2,071
Registered: Mar 4, 2007
From: San Francisco
Age: 43
Home page
Re: Pente - Understand vs Memorize
Posted: Oct 11, 2010, 10:12 PM

you can make a basic check list. but by its self doesn't answer every question when mastering the game i don't think. there are exceptions to rules. and there are exceptions to the exceptions, and so forth. and there are things that could be placed into a check list, but with out the experience as to what they really mean, they will not be fully helpful. some of it is visual, and some of it is concepts of time and space and energy transferring.
sometimes the more complex you make a subject the more you lose, but sometimes the more simple you make it and again the more you lose.
there is a 3 to block, block it, but check for higher threat levels first that not be apparent, but then don't block it because there is a way to block it with out blocking it, but don't do that because there are other surrounding stones creating a unique situation that prevents this because of blah blah blah ext ext...
this stuff is hard to put into a easy to follow check list when the understanding isn't there to follow it.

i suppose you could make a list, and there would need to be new terms created with values to incompass many things that are understood in concept but have yet to be labeled. but with out training to know these new terms well they will be jibberish.

Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
zoeyk

Posts: 2,071
Registered: Mar 4, 2007
From: San Francisco
Age: 43
Home page
Re: Pente - Understand vs Memorize
Posted: Oct 11, 2010, 11:24 PM

here it is black to move. there are no captures. white has a 3. what is the thought process check list for P2 to proceed? and each move going forward, what will be both sides thought process check list? how will it play out if done correctly? who wins in perfect play here?
keep in mind i threw this together in about 20 minutes.

i think black plays L10 now, and plays M6 after that. but why, how did we figure that out? and yet, even if we know that to be the case, how do we proceed from there? how do we know who has the advantage? what are the tempo levels at?

lets see if someone has a good answer here. maybe alisontate? or up2ng? or watsu? or some body?

and i even challenge nosovs or richardiii or rollie tesh to tell me who wins here...


BRAIN FREEZE by zoeyk
no captures yet, Black to move, who wins?
Image and video hosting by TinyPic




Message was edited by: zoeyk at Oct 12, 2010 12:52 AM
I added move M5 to fix a problem.

Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
Replies: 122   Views: 260,340   Pages: 9   [ Previous | 1 2 3 4 5 6 7 8 9 | Next ]
Back to Topic List
Topics: [ Previous | Next ]


Powered by Jive Software