Chess theory

Sort:
Ziryab

The only expert on the topic is a professor of computer science at SUNY Buffalo. He has produced mounds of evidence for rating deflation. You can Google it.

TheOldReb
Ziryab wrote:

You do know that ratings 30 years ago were inflated in comparison to those today? If you were 2100 then, you would be 1700 today with the same skill.

I hope you dont actually believe this hogwash ??   Undecided

 

Sqod
eastyz wrote:

How do others approach this question?

 

Exactly what is your question?

(I had a lengthy response that got deleted accidentally. Maybe I'll rewrite it from scratch if you provide a provocative answer to that question.)

 

Martin_Stahl
eastyz wrote:

....  As to 1200, TT defaults back to that rating when you do not use it for a while. ...

 
No it doesn't. You can reset it yourself, but if you get a score and never do another tactic again, you will keep that rating. (barring a system bug or something).

zborg

Ratings in Standard Chess on this site have "deflated" about 300-400 points over the past two years.  Ditto with Blitz.

In the case of Standard Chess, the site decided to arbitrarily "boost" everyone's rating from 100-400 points, depending on your initial rating.

"Too easy to use engines," some would say.  I would say my Blitz rating fell 300-400 points for "unknown reasons."

Nonetheless, I kept my Standard rating @300 points higher -- largely because I stopped playing Game in 15/5, just as the "ratings deflation" kicked in.

The cheat forum discussed it, ad nauseum.  Results were inconclusive, AS USUAL.

LIfe Goes On, in our sleepy little chess town.

Ratings fluctuate, sometimes quite a lot.

eastyz

Martin, you have no idea what you are talking about.  That is what happened.

drawingdroidfish
[COMMENT DELETED]
eastyz

Sqod, I was asking how do others approach the question of finding tactical opportunities over the board?  In other words, is there a method they employ?  I once had a Soviet coach.  He used to give me tactical exercises to solve but he also said that my over the board approach was poor.  He used to knock about with world champions in the Soviet Union.  He once said that I am the most gifted tactical player he ever came across.  He even once accused me of using engines.  I didn't, of course.  They were very primitive then.  I gave up chess soon after for a career but it has intrigued me since as to why that sort of ability does not automatically translate to over the board results (I am talking about serious games rather than blitz).  So in recent times I have been looking for clues in both what great players have said in the past and in what the current crop say about their approach to the analysis of concrete positions.  What surprised me is that there seems to be little in the way of a systematic approach to analysis amongst the current crop.  They just analyse.  The past great players can't be quizzed in the same way.  My question is whether a systematic approach can improve over the board analysis.  It can certainly improve solving puzzles as I have found.  But what about over the board?

Martin_Stahl
eastyz wrote:

Martin, you have no idea what you are talking about.  That is what happened.

 
While yours may have glitched and reset, that isn't something that normally happens. It makes absolutely no sense that it would either.

 

The rating system used here has a built in Rating Deviation which will change the longer you are away from the site, which will cause wide rating swings if you start in again (though I would have to check to verify if TT uses RD too).

Ziryab

Martin is correct. The TT rating does not reset due to inactivity.

Sqod
eastyz wrote:

Sqod, I was asking how do others approach the question of finding tactical opportunities over the board? 

OK, I thought you had the tactical aspect mastered so now you were asking about a similar system for positional analysis, which I know more about. I believe that "chess science" is in a surprisingly primitive state now, despite all the computers and databases, compared to what it could be, so I haven't heard of any great shortcuts in tactical analysis. As far as I know, all tactical methods boil down to: (1) a standard checklist of things to look for, especially checks, captures, pawn pushes, and pawn sacrifices; (2) positional analysis that looks for anomalies like hanging units, overworked units, badly aligned units, more attackers than defenders, etc.; (3) interchanging the order of moves in a considered combination; (4) pruning the tree of possibities using heuristics, memory, more forcing moves first, etc.; (5) common tactical themes, such as clearance, obstruction, attraction, true x-ray moves, etc.; (6) brute force calculation everywhere else.

The responders here are surprisingly negative. Your claim of a terrific tactical shortcut sounds a little grandiose, but it still sounds plausible to me. For example, at one point in the '90s I came up with a system of positional analysis that used priorities of heuristics that required almost no thought for finding the best move, but I never got around to writing it up or exploring it further. I believe it started to break down when the position became freer or more tactical, though, since it arose from my study of defense in one very marginal opening where only one clear-cut best option existed at each step. With my math background I'm also very aware of invariants everywhere in nature, so that even complicated systems can be shown to have some absolutely guaranteed invariants existing in them. Therefore I'm more open-minded about such analytical systems existing; I can imagine a similar system as mine except inverted in a sense for tactical application rather than positional/heuristic application. If you have such a system you definitely should publish it in some way, whether to make money or to make a name for yourself since I'm pretty sure nobody else is aware of such a system.

eastyz

Ziryab, you insist on reinventing history.  That is what happened more than once.

eastyz

Sqod, I am interested in what you say (I had actually forgotten what my original post said in view of some of the nonsense written here which distracted me).  As to my tactical system, my only claim is that it works for me and I believe it would help others.  It needs hard work to learn it but I believe it is worth it.  I am still learning things about it.  By that I mean that I have to learn to recognise certain specified elements in the position and before applying the system.  It sounds mechanistic but it is not as simple as it sounds.  However, what I can say is that I remember how long it used to take me to solve tactical puzzles of similar nature and this system improved my speed significantly.  If nothing, it forces me to look at things I probably would not have looked at until I became entirely frustrated with a position and started looking for wild solutions.  Of course, objectively speaking, there is no such thing as a wild solution but we are trained to avoid certain ideas and often enough that is where the solution lies.  In summary, nobody will become world champion using the system but it will help them improve tactically significantly and even dramatically as they got better at the system.  But the system does require the ability to moves pieces in one's head without confusion.  The system helps in that sense as it lessens the confusion.  But it cannot help a beginner with learning to move pieces in their head which only comes with practice.  Getting back to your system, it looks very tactical to me rather than positional.  Having said that, the more I work on my system, the more it seems to me that the difference between tactical and positional chess is conceptional rather than anything else.  We call a position "positional" because it does not lend itself to concrete analysis as there are too many possibilities and for that reason it seems to me that my system is more feasible than yours.  That said, being able to narrow down the possibilities and then allowing the player to choose would be a great tool.  Are you going to write up your system?

Sqod
eastyz wrote:

Getting back to your system, it looks very tactical to me rather than positional. 

...

Are you going to write up your system?

That list I gave above is for standard, conventional methods of analysis of a position for *tactical* possibilities, and for carrying out the needed *tactical* calculation when necessary. My system was only for *positional* heuristics.

No, I don't have time to write up that method of mine. I don't think it was very complicated and maybe not even very novel. It just used larger chunks to consider--heuristics--rather than individual moves. Basically you had to know all the heuristics for the position, things like:

"You can never use ...b6 with the intent to trade pawns as an option."

"...Na6 is the standard defense, but don't use it if you can't back up your knight with another piece."

"If White's bishop doesn't protect the c5 pawn, then don't play Na6."

Since it was clear at a glance what the situation was, it was clear which heuristics applied and which didn't, and therefore it was immediately clear which move to play. Whether it equated to a system of IF-THEN statements, prioritized heuristics, or something else, I'm not sure, but I was struck by how accurate and yet how simple it was, and I noticed that whenever I mastered a given position in any opening, I was using the same system, just with different heuristics.

Somebody posted a thread on this site about 1-2 months ago that hinted that they worked out a similar system that was based on the idea of simple mechanics of blocking and of open lines, though they described it differently. The player was fairly low rated, and I don't believe they worked out the details of the system, but that's another sign that chess players everywhere are occasionaly noticing deep underlying patterns and/or logic that makes it easier for them to find the best moves.

eastyz

Tactics

1219
eastyz

Sqod, I was hoping you had something of general application.  If I have not misunderstood you, your system works on a particular opening and perhaps similar positions from other openings.  Otherwise you have to learn the tabiya of each opening.

Sqod
eastyz wrote:

Sqod, I was hoping you had something of general application.  If I have not misunderstood you, your system works on a particular opening and perhaps similar positions from other openings.  Otherwise you have to learn the tabiya of each opening.

You're right: each tabiya had its own set of heuristics that had to be learned. However, whatever method of representation my brain was using was the same for each tabiya I mastered, and occasionally some heuristics would be the same between positions.

The application of the heuristics was more algorithmic, not heuristic, since there was no hedging ("Well, 80% of the time, x will be a good move."): the proper move was clear-cut every time, with no other candidates ("Move x will work 100% of the time here, and no other move will work.") The heuristics were there, but already built into the chunks that were being manipulated. Maybe the only differences between positions I knew vs positions I didn't know were all the exact heuristic chunks, and what those heuristics had to be for the likelihoods to all be 0% or 100%, never inbetween. That took a lot of experimentation. The main thing I would have to figure out to write this up would be the exact representation I was subconsiously using: rules? IF-THEN statements? mechanical object manipulations? I just don't know and I just don't have the time to work on it.

So was your system a general system for any position, including unfamiliar positions?

eastyz

My system has general application for positions that can come out of real games rather than composed problems as such although, while it was not designed for them, it was also useful for them.  Familiarity with the position is not essential and what I found can be a distraction because it results in prejudice in terms of variations to look at closely.  By algorithmic, are you saying that you are able to find the right candidate each time?

Sqod
eastyz wrote:

By algorithmic, are you saying that you are able to find the right candidate each time?

Yes, for a given tabiya, the "algorithm" (system) gave one and only one candidate, it was always the correct one with 100% certainty, and it was very easy to use since it required noting only a few characteristics of the position. I suppose the cost was the amount of experimentation and analysis that had to be invested to produce those heuristics. I'd like to reach that level of mastery in every opening I use, but since then I've changed openings and I am still short on time. It's remarkable that brains can come up with systems like that on their own, and it would be interesting to figure out what data structure is being used, but since those data structures are being used only in one narrow domain (chess), it might not be valuable in general to figure out what data structure it is. Brains apparently do the same thing subconsiously with other topics, especially language, especially when learning that there exist exactly 10 sentence patterns (in English), and what those patterns are, without having to learn those patterns in a college level English class.

eastyz

The subconscious use of data structures is a tricky thing for chess.  Because chess is not a "natural" exercise, the data structures (which are based on perception) get distorted unless one is taught the right things about a position.  This is what I mean about familiarity sometimes being an impediment to solving a tactical problem.  Chess intuition is not always a good thing.  There might be that minor detail in the position that is different which makes all the difference to the solution.  As to your algorithm, maybe it was pitched too narrowly.  I had the same problem at one stage and then ended up with the opposite problem.  Then I chopped and changed until I simplified the system enough to make it usable while remaining of general application.  I did not think it possible at one stage but it was.  Maybe that is what you should be aiming for.