Sunday, May 29, 2011

Chess960: SP301 - An extreme test on opening theory

Thanks Mark for that wonderful example: Watch out for KRBN!. As pointed out the SP301 combination doesn't happen often in Chess960 (roughly 2-3% of all games). So I decided to take a look at this extreme example from the point of view of a possible general opening theory in Naming the Pawn Moves.

What I found is pretty interesting! The move that the CCRL engines did not play very often was 1)....e5! an absolutely striking example that I'll examine a bit more. I think what happens with the 960 engines and I see it a lot in Rybka. They get fixated on a specific line and don't come off it. With human intervention it is amazing to see how they then recalculate their lines as if the engine says "oh right ok, I didn't properly evaluate that idea thanks human!". You even see that change of mind in quite a few mid-game situations that are positionally oriented.

SP301: Black plays according to solid theory
The main point is that the idea 1...e5!? actually follows opening theory. I'll explain it below:

1) Ng3 ... e5!? According to the theory, this one of two "free pawn moves"
2) Nf5 ...      White gifts a tempo to black by moving the knight twice
Note: if you want to know why 2) Nh5? doesn't work...I've got the answer hidden somewhere on this page. It is both hilarious and beautiful. I dream that one day in the future people will primarily play Chess960 for this kind of joy... :-)
       ... g3   Black plays a gift "free pawn move" from the spare tempo
3) Nh6 ...      White gifts another tempo to black by attacking the bishop
       ... f5   Black plays another gift "free pawn move" from the spare tempo
4) NxB ... RxN  A neutral tempo exchange. It will be hard without the bishop.
5) c4  ...      White plays their first theory allocated "free move"
       ... Nf7  Black plays a knight before his second allocated "free move"
6) Nc3 ...      White plays out a knight according to theory
7)     ... a5!  Black plays their second theory allocated "free move"
                Black has a playable game well done!

I think that a5 is a great move. What it does is prevent the coming of a massive pawn storm that white will unleash on the c-side after white castles g-side and black castles c-side (formely known as queenside). It also gives the black queen an alternative development path. (I actually doubt that black needs to castle at all but that is another story).

But the real point is that black has played according to theory. Black played out a total of four free moves but two of them were freely gifted to black by white in the tempo that white expended chasing down black's metropolitan bishop (Naming the Bishop Pairs). Not only that, but black played out a knight before his second theory allocated "free move" (the other "free moves" were not forced and were temporal intermediate opportunities).

The pawn storm that I am talking about is listed in the alternative lines above but you can see the result of it here if black does not subdue it with his free move 7) ...a5!

SP301: White pawn storm on the c-side king (black to move) 

In summary, I do think that these 2-3% of positions are about black finding a way to draw. But heck does it matter? It's a lot of fun in the meantime considering that there are 960 different starting positions. There are some surprising resources that black actually has! Haven't even touched on some of the others.

Enjoy 960

Answer: 2) Nh5?..g6!, Ng7+?...Ke7, e4...Kf6!, Nh5+...gxh5, f4...Kg7!, fxe5...Ng6!

Thursday, May 26, 2011

Chess960: Naming the Pawn Moves and Insider Information

(Image: Tool Pawn and Trade)
Ok here is a very simple start to this post:

There is such a thing as Chess960 opening theory...

How do I know that? Basically you have to read the most comprehensive book on Chess960 that has ever been written. What is that book and where do I buy it? It is the chess960 positions themselves position by position all 960. Reading that book is like experiencing a general lesson on chess that no other book comes close to. Read it sequentially position by position page by page. Each consecutive number changes the start position by just enough to make you think, but not enough to confuse you. 

I can say that after having read a few hundred pages of the 960 pages in that book, Chess960 opening theory does exist. How can such a theory be framed and how can it be useful? The approach that I think works best is to craft a "superannuation" or investment theory of Chess960 opening play. Such a theory is useful when time is short and players have to think quickly and efficiently....

So here we go on a basic way to play the Chess960 openings under quick time controls (forget about classical time controls for now, where Chess960 theory is much more elaborate and detailed) .....

In the Chess960 opening we look for "theories" that do a very similar thing to investing. What theory provides the best return over all 960 positions? The theory might not be the best theory, but it provides a steady return over the long term experience of 960 different openings. The theory might actually fail in some small set of circumstances, but all in all it is a solid investment. Some better theories would provide a higher return over a smaller set of the 960 positions and some exceptional theories pay off even better. The great book of Chess960 has various continuous sequences of positions that definitely do suggest that there are specific exception theories that yield a high rate of return. However those theories collapse over the entire range of 960 positions. In fact, some theories are way too conservative as well! The point is that in Chess960, we are working probably 25% harder than we would be in classical Chess, because we are having to think from the first move. Therefore a well invested opening principal will save a lot of energy that we will need for the real heart of the battle in the mid-game.

So now that I have framed Chess960 theory in terms of financial risk and investments, why not continue that argument into the most fundamental unit of investment in Chess960, the movement of the pawns? Each time you play a pawn move you are making an investment in tempo, space and initiative. You must make the investments wisely, because each investment is linked to it's other investment partners and if not played carefully, the whole industry of your pieces collapses into ruin (don't despair, learn from it, take a break, think it over and start another game!)

In standard chess, we see only one type of opening pawn move! But if you read the great book of Chess960, you find that there are actually at least three types of pawn moves in the opening. I will ignore two pawn move types that are obvious (a good pawn move and a bad pawn move :-)

Naming the Chess960 Opening Pawn Move Categories:
  1. A Free Move (a theoretical thematic allocation)
  2. A Gift Move (a temporal thematic opportunity)
  3. A Barter Move (a mutual necessity)
  4. A Money Move (a positional exploitation)
In a strict sense, these pawn types only occur during the immediate opening development phase of play:

Pawn Move 1: A "Free" Move
What do I mean by a free pawn move? Well that's really simple. All the pawn moves in standard chess SP518 are actually "free":

The Free Pawn moves of SP518
Each pawn move in green both by white and black are "free" in the sense that the player can play these pawn moves according to their own personal free choice based on thematic ideas they hold. There is a cost to every pawn move, but that cost might never have to be payed back. When you want to understand the concept, think of the pawn moves in Chess SP518. They are all free.

Pawn Move 2: A "Gift" Move
A gifted pawn move occurs when the opponent has gifted you a spare tempo to play a "free pawn move" outside of your theoretical allocation of them. This often occurs when the opponent moves the same piece twice because of a development problem that they have discovered and that must be sorted out by giving away tempo to you. Another situation might be when an opponent begins a premature attack during the opening phase that costs them excessive tempo to execute. In such cases, a "gift pawn move" might well be the winning move because a single such move can increase space as well as make it much easier to complete development. A "gift pawn" move is a "free pawn move" because not only can you execute the move without cost to tempo, but you can play it according to your free choice of thematic idea. You have to be careful about whether the pawn move is actually "gifted". It may turn out that in some SP's it is highly desirable for both players to move the same piece twice during the opening phase and as such the tempo is not actually a free gift to you. If that is the case then have been tricked and you must pay for the "gift".

Pawn Move 3: A "Barter" Move
What does it mean to play a "barter" pawn move? A barter pawn move is one that is equally exchanged with the opponent. It is a pawn move that both sides are compelled to play essentially to survive or that is a commonsense rational requirement. It is an equal trade of tempo and space to accomplish the undeniable necessities of existence . Hence they are called barter moves!

A "Barter" Pawn Move of SP192
The moves in yellow are not compulsory by any means, but are so compelling on a simple rational basis that both sides simply exchange them with each other in a barter like fashion. The deal will probably favor one side more than the other, but in every barter transaction there is some inequality! Barter pawn moves are not always as obvious as the example above. If you have a look at the Svidler/Leko example below, you see that these great grandmasters realized that there was a rationally compelling barter pawn move that BOTH black and white should play (e3/e6) but that the move was far from obvious.

Pawn Move 4: A "Money" Move
The analogy of pawns and exchange and investment carries on one step further. In Chess960 there is a pawn move that is not actually free in the sense that we feel compelled to play it for a specific reason. It is not a "bartering" move either. The hope is that the pawn move will pay off much more than what it could cost us in time, tempo and adverse pawn structure. We make a risk assessment and sense that there is a chance to exploit an opportunity that the opponent has allowed and so we make "just one more pawn move!". The move is called the "money pawn move" and it is the one that costs a lot in terms of structure, tempo and possibly material as well, but we cannot resist to play it because of the potential pay off in raw cash!

The "money move" is a feature in quite a few Chess960 SP's because of the piece layout, particularly with regard to white's opportunity. The "money move" is not seen nearly as much in standard chess SP518, because the way that the pieces most naturally develop does not facilitate attack by the money move as much as other SP's. "Money moves" of a type do occur in SP518, but the strict definition of a money move is that it occurs during the immediate opening phase of play. One particular form of the "money move" is in the Chivalry Knight starting positions (Naming The Knight Pairs) where one money move is played in order to facilitate the threat of a second money move in the form of a pawn fork on that knight combination.

A "Money" Move of SP192
The red arrow is a very crude example of the money pawn move. Black would probably not play like this, but it does highlight the money move very well. The theoretical idea behind the "money move" is that if the pawn must expend one tempo, then the the net result is that the opponent must play more tempo escaping the subsequent money moves, or simply that their possibility to develop out of the opening is more impeded than the money move player. 

If this were a subtle position, white should be pressing on with development. Castling has not been completed and the rooks remain unconnected. The bishops are hypothetically already developed as is the case with the Primary Patriarch Bishops (Naming the Bishop Pairs) but white has issues with developing his remaining knight that need to be sorted out as well as his queen. In theory, the money move can also cause structural dilemma's in our setup (but not in this case). Simply put it is that white cannot resist to play "the money move"! In this case the move d5 does pay off, but in more subtler examples it most definitely will not! Interestingly in this position, white has played two barter moves and one free move before the opportunity to exploit the money move. Just like risky financing, in Chess960 it can turn out that one money move leads to another!

An Example of Pawn Move Types - SP51: Svidler/Leko Mainz 2003

How many types of pawn moves does white play in the main line example? (Svidler's Opening)

  1. One theory allocated free move at move 1
  2. One barter move at move 2 (because both sides will have to play it)
  3. One barter move at move 4 (because both sides will have to play it)
  4. One barter move at move 5 (e3/e6 is compelling for both sides to play)
  5. A second theory allocated free move at move 7 (Note: There is a nice discussion that sometimes takes place between what is a genuine barter move verses what is a free move? For example in the example above, could Svidler have played b4 instead of b3 and if he did would that be a barter move or a free move?)
Notice that the line that is showing is different to what Svidler actually played in the tournament (see the side line). What he actually played in the tournament was two free moves before developing a knight (it did not work as well as the above alteration in the main line where a knight is played out between Svidler's two free moves)

A Beginners Guide on How to Play the Free Pawn Moves in Chess960 Openings:
The proposal is that in general Chess960 opening theory, white get's two possibly three "free pawn moves" but black get's just two. As per the example above, the proposal is that it is generally good to play out one of your two free moves followed by a knight and then working towards the second "free pawn move". For roughly 10% of Chess960 positions there will be severe exceptions to this general principal of opening play, but that is a matter of experience!
  1. Learn to distinguish between the three types of pawn moves
  2. Allocate yourself two Free Moves in your opening plan (possibly three if you are white)
  3. Find one good Free Move to start (that promotes your thematic idea)
  4. A genuine Barter Move does not count as a free move
  5. A Money Move does not count as a free move (be extremely skeptical of it)
  6. A Gift Move does not count as a free move (be extremely skeptical of it)
  7. Try to play out a piece that influences the centre, before the second genuine Free Move
  8. Calculate if the opponent has given you an opportunity to play a Gift Move above your theoretical allocation of Free Moves
Do you see how this theoretical breakdown of the pawn move types in Chess960 helps? Looking at the example above, Svidler played barter moves as well as free moves but it about the net number of free pawn moves that Svidler played which did not exceed two such moves and is thus in accordance with theory.

The Juicy Bit
Ok now onto the juicy bit where I get to tell you some insider information on playing the Chess960 opening. If you have read this far then you deserve the reward! In Chess960 we count the total number of actual free moves that are distinguished from barter moves.

If you want to win at Chess960 in situations where there is limited time and you want the best rate of return on your pawn move investments over the entire 960 positions, then you have:
 2 FREE MOVES (possibly 3 if you are white!);
As black, try to conserve free moves by finding ones that release two pieces;
You possibly have up to 3 BARTER MOVES (calculate carefully!);
Either 0 or to as many MONEY MOVES as the opponent will give!
Either 0 or as many GIFT MOVES as the opponent will gift you!

 To play the money move, you have to be VERY confident that you can exploit a situation that would not normally have taken place with correct play. As such the money move actually PAYS OFF. Most of the time the money move does not pay off because the dividend is only a temporary gain.....If you are a novice player it is usually far better to continue developing your pieces no matter how tempting it is to pull the money lever and play the money move....

Essentially what I am saying is that for novice level play in Blitz or Rated categories, there is a specific way to play the Chess960 opening according to principals that does not pay off with the highest rate of return, but does pay off with a solid position where the opening did not take long to play. You will most likely not have played the best line theoretically, but your opening style will be much more adaptable to all 960 positions and the mid-game position you get will be solid in any case.

....Even more importantly because you played according to solid principals, you will be feeling confident and have plenty of time. Matter of fact even more importantly, because you have played according to solid principals, you will be calmer of mind and be thinking more clearly positionally and tactically in the many unfamiliar situations that Chess960 tends to throw up. Even more importantly, because you have played according to solid principals, you have a principal that you can learn how to break when necessary, and thus will form a tree of principals rooted in one solid theory.....

More on that in future posts, but for now remember that in Chess960 you have two free moves that can be distinguised from barter moves and possibly a third free move if you are white. It is usually better to play out one free move, a knight or a piece that takes some control over the center squares, and then another free move.
In that sense this Chess960 opening theory is basically the same as Standard Chess SP518 except that because the pieces are non-optimally organized, there can be that third free pawn move that makes all the difference as well as intermediate barter, money and gifted pawn moves.

Here are more examples of the pawn move types: 

Enjoy 960.

Wednesday, May 25, 2011

Chess960: Amazing variations SP189

Chess960 throws up some really unusual examples of how a good position can compensate for being down in material. Here is an absolute corker example. Black is down a rook but up a knight and a pawn. Both players have problematic positions. Black has both knights on the back row and doubled up pawns under attack. But even worse is white's trapped queen. The variation that follows shows how amazingly black's pieces cooperate in such an efficient and elegant way.

SP189: Black fights back!

The part I cannot get my head around is how black manages to bring both knights back into active play AND maintain an initiative that does not wither away (like usually happens!). It is nice to watch how black get's the two knights active in the time it takes for white to release the trapped queen. A beautiful variation (in a non-classical sense).

Enjoy 960.

Sunday, May 22, 2011

Chess960: How to practice it - part one

The thing about Chess960 is that there is no official way to practice it yet. That suits me perfectly because I started playing chess very late when I was twenty years old and so I have no entrenched investment in Chess that would have happened had I started very early. As soon as I became aware of Chess960, I simply dropped Chess altogether because the variety in the Chess960 openings is so much more enjoyable. I have played only Chess960 since, and have never looked back. Chess960 is truly a wonderful pastime because you do not narrow yourself down with unnecessary knowledge. Everything you do in Chess960 is enjoyable and the game is totally balanced in terms of skills and enjoyment in all phases from the opening to the middle to the end. There truly is little point in playing Chess when you can do everything that you can in Chess with Chess960 and win and loose just as much!

Here are my thoughts on playing Chess then how to play Chess960. Chess is a combination of these skills:
  1. Pattern recognition that flows on into an instinctive awareness of basic tactics (under-defended squares are especially important for Chess960)
  2. High level calculation when pattern recognition is no longer good enough on it's own
  3. Very vivid positional awareness (the true art)
  4. Memory skills
The only major difference between Chess and Chess960 is step four! Chess960 requires great memory skills just like chess. What is being remembered are chunks of familiar themes and clusters of piece structures rather than trees of opening lines. Chess960 can even name these chunks as I am beginning to demonstrate in posts like Naming the Knight Pairs and so Chess960 can develop huge volumes of theory based upon these categories, just as traditional chess has done with it's opening tree. People with exceptional memory will even remember specific lines.

So how do I specifically practice Chess960 in the absence of official theory? Here is the high level method:
  1. Learn a basic principal of playing the Chess960 chess openings. The standard chess opening principals are a good place to start.
  2. Study all the basic tactical principals of chess
  3. Study a core group of standard Chess games to get an idea about positional play
  4. Every endgame you get into, you later study with official theory (learn as you play)
Here is the specific "day to day" method that I use. The central point is that in Chess960, we have to be acutely tuned into analyzing the position on the board even at move zero. So the question is how do you truly analyse "position" when the positions in Chess960 are highly unusual at times relative to chess? At the same time there is no-one around in on the Chess960 scene that is readily available to help. Here's an introduction to how I do it:
  1. Practice through the Chess960 positions sequentially and each position both as white and as black. Literally start at 001 and end at 960 and then repeat. Each game is played on quick time controls but analysed in retrospect very slowly. You do this because it is the best way to develop your own ideas on the chess960 opening. If you play through the positions sequentially, the positions change only so much so that you can still compare them with the last position. It is a beautiful journey to take. Whole sequences of chess960 numbers share common patterns that you can study and learn from as they pass before you. It is like reading the most fundamental book on Chess that has ever been written! Do it! If you just randomly play Chess960 starting positions, your mind struggles to unify and formulate any principals that are meaningful to you. We need graduated progression to formulate ideas, and that is exactly what happens when you play through all 960 position sequentially. People think it takes too long to do this. But one journey through the 960 positions is not even two thousand games. Have a look at how many games people knock up in standard chess! It doesn't take much. Four quick games of Chess960 a day translate to an hour of actual game play followed by one hour of deep study afterwards. Two hours a day total time is all that it takes and you will have gotten through the 960 positions in approximately fifteen months.
  2. Practice tactical pattern recognition. In Chess960 knight tactics are the first priority
  3. Practice being tuned into the "moment" where a vivid awareness of position is developed
  4. Play a quick 960 game (my favorite for Chess960 is 15min games) preferably both as white and black.
  5. Then play through the quick game incredibly slowly afterwards with a chess960 engine. Ignore what you're opponent played. Just focus on what you played. The idea is that you are not trying to learn about the absolute nature of the position, but your own perception of the position. You use a chess engine to give you ideas on that (I'll explain more about that later). Even though the game is finished and regardless or whether you lost it or won it, you study the game in detail in terms of the position on the board comparing it to what you thought at the time, how long it took you to think it, and what you think about it in retrospect.
More on that last point next post. Essentially the art is how to lesson the amount of "blind spots" that are apparent in our mind both on tactics and positional understanding.

Wednesday, May 18, 2011

Chess960: The chess clock - when it should start ticking

Typical for me, I go onto a forum, passionately request something, use too many words, loose everyone's attention, and at the same time they don't understand what I am proposing (whatever: my intention is good!). I can sense that there is an opportunity to do something unique and special that not many people have been thinking about as they develop Chess960. I am not the only one to think it. Here is Mark Weeks on the subject:

"I'll close this post with another plug for Chess960 Jungle. HarryO wrote this week about The chess clock - when it should start ticking. It's an important subject that's given short shrift in the implementation of online chess960. I imagine that most chess960 software developers are building on an existing implementation of traditional chess and reuse the same clock specifications that have been developed for the traditional game. The considerations for chess and chess960 are, however, a world apart in the opening phase. 
In traditional chess you're basically on autopilot for the first few moves because you've seen the start position thousands of times. In chess960, you're on your own from the very first move. It is fundamentally unfair to let White consider the first move without using clock time, then start the clock for Black as soon as White has moved. You could say that the unfairness applies to all players 50% of the time, but if you're playing a chess960 game for an important prize and you happen to have Black, that 50% argument doesn't help you. I could say a lot more about this last topic, but the Sicilians beckon. Maybe next week..."

The simple request is that at (and also more generally), white's timer should start counting down at the beginning of the game. Here is my forum post on it where I did not seem to get my point across (according to the one reply I got): [feature request] Chess960 clock change

Here is the critical point from that request:
  • White's clock should start counting down the instant the Chess960 board is displayed to both players and they are ready to move (real life or internet)
  1. Because white has a natural initiative advantage at the start of the game as well as having a tempo advantage. The compensation for that is that white should have to pay with time.
  2. The problem is made worse because if white's clock is stationary while white thinks through his first move, it will also be stationary while black makes their first move. White's time only starts to be penalized after two whole half moves have already been played.
  3. Black only has a marginal chance of being able to plan a response to white's first move while white is thinking. If white plays an unexpected first move, black must recalculate from scratch while white has no such uncertainty. White has played their move and can already begin to calculate all black's replies with much greater probability of correct calculation at zero time penalty.
  4. Theoretically a Chess960 game is already a blend between a very high level mid-game scenario AND an opening phase at move zero and thus the player with the move should be paying for that on their clock (as if the game were already in progress when the players arrive at the board at move zero :-)
What are the benefits?
  1. For the first time in the history of Chess there is a chance to compensate black for being down in initiative and tempo! It is not much compensation, but at least it is something. It will shift the odds of white winning slightly back into black's favor, and that is a good thing.
  2. It will make Chess960 more exciting because it will give black better chances to win.
  3. It will cause interesting first move clock strategies: white might choose to play the first move quickly on the belief that his black opponent will suffer on the clock on their first move. White might choose to play slowly on the belief that concrete analysis of the opening will cause undeniable benefits that black will not be good enough to counter, even if it has cost more time.
What are the issues?
  1. Black would get too much advantage? People say that there is possibly a high level Zugswang in Chess960 and white suffers by playing at all! But I think that is just not true to any practical degree.
  2. It would cause confusion because white is not used to the clock ticking on the first move? Yes that is true, but any change always causes some turmoil initially. It will quickly settle down once people get used to it.
  3. At this early stage in Chess960's evolution people are so unskilled at it that it puts too much pressure on white and will discourage people from Chess960? That may be true initially, but only for the white player! The black player will like it. But over a reasonably short period of time the balance of the result table will naturally re-emerge according to the reality of white's initial initiative and tempo.
  4. It could cause a clock synchronization issues across networks in internet play? That may be true, but there must be a reasonably creative technical resolution. It just needs a bit of clever thinking.
If you agree with my thoughts, could you help me to convince ChessCube by supporting me on the request forum? The issue could possibly repeat itself if start implementing their Chess960 live code. It would be a tragedy I think if were to leave white's clock stationary at move one - thanks

Monday, May 16, 2011

Chess960: The territory of the Chivalry Knights

Now that I have got some name for the knight pairs in Insider information on the Knight pairs, it's really useful to look at them in detail. First came the Territory of the Military Knights and now the Chivalry Knights. Together with the Military Knights, they make up almost 50% of all Chess960 games so if you fundamentally understand these two pairings, you should have a good chance to win half your Chess960 games. Even without considering the Military Knights, the Chivalry's alone cover roughly 33% of all Chess960 games. This is because they exist on their start squares in two forms (NxN and NxxxN).

The Chivalry knights are great gallopers that can quickly confuse and demoralize an enemy. You can imagine that their speed is fantastic but because of it they have blind spots directly adjacent to them. They start on the same color squares and this is what really distinguishes them from the other knight pairs. When developed to the third rank, in just one move they can theoretically cover an exact 50% of the area of the board ((35sq-3sq)/64sq))  in a perfect 7x5 rectangle (except their blind spot squares):

The Chivalry Knights (~33% all games): 7x5 Territory (50% of the board)

The green squares show the squares that they cover straight away. The yellow squares show the squares that can be covered in a single move. The red squares show where the Chivalry Knights have their major blind spot. 

Like all knights, they have trouble really close in but in the case of these knights, the weakness is minor. They are very "bright" high energy knights because of the way they work together on the same colors, rapidly switching from one color to the other cooperatively.

The Chivalry Knights:
  1. Cover all of the opposite colored squares to themselves in an area of 7x5 squares (except their blind spots)
  2. Cover any same colored square in that area in one move (except their blind spots)
  3. Outside of their 7x5 area, in one move they can only cover a same color square
  4. Have only only three blind spots between them
  5. Have a significant weakness to a pawn fork at their "Chakra" square
  6. Can move forward and lock together in one move when needed
  7. When "locked" they have pseudo rank and file control (see diagram)
  8. Can retreat from being "locked" back to their Chivalry form to cover maximum area
  9. They can in one move form a confusing orthogonal Military Knight stance (see diagram below)  that can catch out unsuspecting enemies that do not appreciate how much area they cover in that configuration
  10. Cannot comfortably defend any of their pawn protectors ever
  11. Work best cooperatively and if they must split apart into independent operation,
     it better be for a good reason

The "Locked" Chivalry Knights area of influence in one move:
The green squares show the squares that are immediately covered. The yellow squares show the squares that can be covered in one move. The red squares can only be covered at the very least by two moves. So their primary area of influence is a bit like a light bulb shape (yellow lines) with some warning "electricity bolt type "zigzags" either side (red lines). When they are locked like this, specific calculation of what they can do is required as well as a general feel for their area of influence.

The "Locked" Chivalry Knights rank and file influence in one move:

You can see that within a single move window, they can actually cover three whole ranks and almost two whole files in an "electricity pole" shape (yellow lines). Since when the knights are locked together the coverage area is not a simple shape, specific calculation is required as well as a general feel for what they can do. As you can see there is a lot of territory that they can cover in one move.

Another particularly confusing arrangement that the Chivalry Knights can quickly posture into is an orthogonal Military Knight stance. When you are looking down the barrel of that stance, it can get optically confusing and can startle the enemy into a tactical error:

Chivalry Knights in Orthogonal Military Knight Position -
a 5x6 "Split Dome of Protection"

Although they are extremely vulnerable to rook attacks because of their severe vertical file weakness, in this orthogonal Military Knight posture they cover the green squares immediately, the yellow in one move and the red in two moves. Therefore their green area is almost a complete 5x6 area dome of protection other than a slice right down the middle. Without knowing this, our human eye tends to think that they have many more weaknesses than this arrangement actually shows them to have (we assume that they have the same incomplete coverage as a single knight). In this posture their adjacent square weakness is only a minor weakness and once again they have trouble getting to their long diagonal squares which is the main similarity with their single knight form.

Here are some practical examples:

Black to move: 5 seconds to find the right move
....Qf7 why? For every move, try to find at least two things that the move accomplishes
  1. Qf7 stays out of the 7x5 Chivalry Knight coverage area
  2. Qf7 stays off the dark colored squares that the Chivalry knights are on and is therefore immune to a forward attack by the knights. This is because outside of their 7x5 area, the Chivalry knights can only attack the same colored square that they are on.
  3. Qf7 attacks f3
  4. Qf7 clears the g-pawn

The Orthogonal Military Knights startle the enemy and cause a blunder!
Why am I going to all this trouble? Well with us humans, we both recognize patterns and we calculate. In future generations when kids have grown up playing Chess960, they will instantly see the coverage area of the Chivalry knights and they will instantly see what squares those knights can attack outside of their area. At the same time they will still calculate specific lines. It comes down to efficiency of cognition. Kids of the Chess960 future will build opening concepts, themes and lines around the various opening configurations of the chess pieces. All of the Chess960 openings will be familiar to them conceptually.

Realize this about our old beloved SP518 classic chess position.  If you look at the knight configurations, we do not often see Chivalry or Military knight configurations. Play around with SP518 knight movements and you can see it for yourself. There is a whole tactical set of themes missing in the SP518 setup that we only see if we play Chess960. Realise that Chess960 includes SP518 and other positions that are probably just as beautiful and just as deep. 

Friday, May 13, 2011

Chess960: Amazing variations SP175

Check out this amazing amazing midgame/endgame that SP175 threw up in post game analysis! Chess960 can produce some bizarre variations that are for want of a better word....breathtaking!

A  Queen+Rook sacrifice variation that emerged from SP175:
What can I say! Remember that black is down in material and so is on the back foot but at least puts up an amazing fight. Note that when the comments say "singular move" it means that the move played is the only one that produces a significant gain or is required as a significant save. Rybka helped me find this line....

*Some* of the ridiculous highlights!:
(Ok you think you see a mate in one, but think not about what you see, but how white constructed what you see!):
  • 63.       ... Rf1+!!  += (else mate in #5)
  • 64.       ... c8/Q+!! += (else mate in #10)
  • 65. Ke2!!             += (else black -+)
  • 65.       ... Qd1+!!  += (else white +/-)

The variation shows an incredible merging between mid-game and end-game themes combined with an amazing clustered pawn promotion. White set's up a complete home on the second rank and migrates his king to the third! The bank rank is completely abandoned and white seems to ignore the promotion threat only to return to it just in time by combining it with a mating threat. Finally white must demonstrate that there is a win without a queen and yet another pawn promotion threat!  Almost every possible chess resource, idea and phenomenon that can be imagined is seen in this variation.

Realise that such a variation is improbable in Chess SP518 because of black's trapped suffragan bishop and white's own hemmed in bishop.

Wow. Enjoy!

Chess960: Gambits and Stalemate

Played a couple of games the other day from SP175. The first game ended in stalemate with a king and pawn verses a king. I should have lost it and the feeling is undeniable. It made me re-look at the arguments about the stalemate rule in chess and the latest discussions on it:

It is a brilliant discussion on so many levels. On reflection, I think the rule is great and some of the people in the forum above made some excellent points. Chess is about material AND position. If you do not demonstrate an advantage in both, you don't deserve to win! That is what the stalemate rule actually says. Beautiful!

One of the comments in the forum above is relevant to Chess960:
"If k+p vs K was always a win, I think most people would adopt a more conservative strategy: don't throw a pawn away in the opening for some initiative, because that pawn is going to come back to haunt you. My own game is based on a lot of semi-sound gambits, which I play with confidence because I recognize which pawn endings I can draw and which I can't - so when my gambit fizzles out, instead of resigning, I start maneuvering towards a drawn endgame. Removing stalemate would make that strategy untenable, and would mean that gambits were much less playable. I'm pretty convinced that's a bad thing."

That is a good point. Thankfully in Chess960 we do have stalemate! If we didn't have a stalemate rule, a gambit line in Chess960 would be even less likely than it appears to be at the moment. As it stands gambits are already rare in Chess960 as far as I can tell. I think that it is because without previous knowledge of the results of a gambit line, players like myself are unlikely to try them. That will change with time as people start to get comfortable with the conceptual thinking that Chess960 encourages.

Here is an example gambit line in the one and only time that I have played one:

SP039: A Queens Gambit Opening without a C1 Bishop!

The line produced a beautiful open game and an aggressive attack possibility on the g-side. It is very funny considering it's initial similarity with the SP518 version of the Queens Gambit line. But the point is that this gambit does not guarantee a likely return of the pawn as does the SP518 version. If that is true, then indeed would such a gambit be played if there were not a stalemate rule?

Sunday, May 8, 2011

Chess960: The territory of the Military Knights

Now that I have got some name for the knight pairs in Insider information on the Knight pairs, it's really useful to look at them in detail. First up are the Military Knights. Together with the Chivalry Knights, they make up almost 50% of all Ches960 games so better watch out for them! The military knights have a decent probability of moving beyond their start squares to sit on the third rank and then from there they may stay together for a long time. They occur in 25% of all Chess960 games and hardly ever in standard Chess SP518. Once they are on the third rank, they cover this territory:

The Military Knights (~25% all games): Territory Diagram

The green squares show where the knights can go and the red squares show where they cannot. If you have this diagram hardwired into your head, you are a lucky Chess960 player because you will win all of the 216 Military Knight starting positions. Notice the blind spots of the pairing? Imagine the diagram as if two real knights were charging along at a quick gallop. Where would their weaknesses be?

The Military knights are:
  1. Good forward and backward two squares out
  2. Good diagonally one square out
  3. Form a hat shaped shield forward and rearward
  4. Totally blind sideways even two squares out
  5. Weak diagonally two or more squares out (like a single knight)
  6. Weak immediately adjacent one square out in a cross pattern
  7. Two moves to cover the distant diagonals but they do so from an undefended square (hint hint)
  8. Two moves to cover each other (hint hint)
  9. One move to cover all the other weaknesses (hint hint)
  10. Only cover one extra file relative to their single knight form
  11. Weak to an enemy knight that can sit inside their undefended territory and attack two of their undefended squares (relevant to the Chess960 opening hint hint)
  12. Strong to enemy knights that can sit inside their undefended territory but never attack them
  13. Weak to all long range attacks unless protected by pawns
  14. Can partially defend their pawn protectors
  15. Totally impenetrable by a king if they are protected by pawns because the only access path is laterally and that is blocked by their pawn protectors. A king cannot even get access to their pawn protectors ever
And so if military knights can move out into position on the third (fifth rank) and hang in there until all the long range attackers are dead, they could hypothetically move forward to the fifth rank and guard a tremendous amount of territory (but always weak to their opposite knights).

A single knight is:
  1. Weak immediately adjacent in all directions
  2. Weak diagonally in all directions and all distances
  3. Three moves to cover the distant diagonals
  4. Two moves to cover the adjacent line squares
  5. One move to cover the adjacent diagonals
  6. One move to cover the distant line squares
  7. A single knight need not go beyond two squares out to cover all it's weak squares
  8. Weak to all long range attacks unless protected by a pawn
  9. Cannot defend it's own pawn protector.

How The Military Knights split apart or wall defend
The combination in the corner can be really tricky for beginners! Normally you would think that the corner knight is out of play, but it is not about the corner knight, it is about the pair of knights together. This corner combination of military knights are keenly active. They can move vertically up the wing. They can strongly defend a king that castles c-side and the b1 knight can split away and reach the traditional f3 position in two moves that is normally only accessible by the Hereditary Knight positions on e1/g1.

SP 095: A Sample Military Knight Game