Friday, February 05, 2016

Mate in one

I'm training "Find all checks" (FAC) lately. How could I have sunken so low? It all started with the analysis of my errors with tactical improvement. That brought to light that there are two main area's of improvement. The first is that I loose a lot of time with useless activities. I calculated that I can speed up about 6x when I am able to cut that specific waste of time. Inventing and internalizing a thought process would discipline the mind, and do exactly that: cut the waste of time.
The second weakness I noticed, is in the visualization-II of future positions. Before I start working on my thought process, I want to find out if I can improve in the visualization-II of future positions. I found that this visualization leaned heavily on a certain set of subtasks. The adage being "you have to have perfect vision of the current position first, before you can improve at the future positions".

And so I happened to arrive at mate in one, in an attempt to improve my vision. Mate in one is the ultimate current position, since the position has no future. Incidentally, I started with mate in one "hard" (M1-h), since I didn't know the difference with "easy"(M1-e).

The discussion about M1 is a bit obscured by the fact that we don't have an absolute calibration point. We don't know how fast a FM does this M1 exercise, and the chain of assumptions to come up with an educated guess is cluttered with noise and tolerances. "You have to become 6x faster than you are" is a bit vague, especially when the improvements of the first week are ignored. So I have to work with my gut feeling in stead "is there room for improvement here?", and that probably is good enough for me.

When investigating the subtask of M1-h, I stumbled upon the subtask "Find all checks". Thomasz tried it, and he found that he was way slower in FAC than in M1-easy. So apparently he has found a way to find the mate without looking at all possible checks. Now that is stunning. Has his unconscious mind worked its mythical magic, or is there something else going on? I think the latter. The subtask FAC stems from M1-h, not M1-e. The "hard" mates are derived from a database with compositions. In a composition, the composer usually builds in a lot of "temptations". That are checks that looks like mate, but which aren't. This makes that looking for checks and control if it is actually mate or just a scam, are subtasks that are overly abundant in M1-h, but not in M1-e. Although they happen every now and then in M1-e.

I think that the skills that can be acquired by training the subtasks of M1-h, are very well transferable to overall tactics, so I stick with that for the time being.



  1. FAC is not that easy element to be really fluent at it! It requires additional subskills. The most difficult to me: LONG moves, pins, quite crowded positions, backward moves and horizontal (long) moves with Queen (for example Queen is at a4, can give check from b4 and h4 - if the King stands on e1).

    Without a really good method to work out the positions quite efficient way - it will be VERY difficult to break 33-35 FAC. It is much more demanding (and complex) task than most people think (imagine). My present best score is 31.80 - I can find all the squares quite fast - EXCEPT the last one. If there are 8 squares - I can find just 7, etc. And finding the last square (solution) takes me about 15-30 seconds (!). What should be done to improve it? ;) :)

  2. In my eyes FAC has 2 important subskills:

    Find a check
    Find an other check ( dont use the same checkidea several times )

    As more checks you have already found.. as more difficult to find the next one IF you use a chaotic thinking system. If you jump randomly from piece to piece and move to move you need to keep track what you already have "calculated" and skip that the second time it comes to your mind.
    So like many of these boardvision exercises this is a thinking process exercise too
    Thats the reason why i did suggest to do the Lucas Chess -> chess for zebras -> find all moves first

  3. About calibration: Its not necessary to calibrate by FM's! We can calibrate it with our own performance: Tomasz ist better than Munich, Munich better than me, i am better than you and there is no doubt you are better than a beginner.

    M1 is the most easy possible tactic. If we cant improve at m1.. we can forgett it.

    1. But how do we now that Thomasz hasn't reached FM level already? He might wasting his time with exercises he is already good enough at.

    2. I am convinced the FM level is about 40-45 MPM (M1-e), IM level 45-48, GM level - 48-55 and SGM (superGM) is 56 and above.

      If I could reach 44 MPM it is highly probable I have already achieved FM level (or I am quite close to such level).

      Ask chess master (FM or IM) if he is able to find a checkmate in 1 (together with clicking the mouse to the correct square) faster than 1 second. The answer could be "rather not". You can compare this in a smaller scale when looking at bullets played by FM, IM or GM calibre players. Most often they need some time to recognize the best move and execute it (I exclude so called "pre-move" from consideration). I have played (and watched) dozens of thousand of blitzes and bullets and I saw how faster the masters may be. Anyway if they get unknown set of puzzles - they HAVE to be slower as they are not accustomed to the new task and do not have a database of 100K checkmate positions (patterns).

    3. Aox: "M1 is the most easy possible tactic. If we cant improve at m1.. we can forgett it."

      We can improve at it, but we need to find the key elements and a few volunteers who wants to practice this task extensively. Now I have just solved another 2280 M1-e and my average score was 43 MPM. Take notice that I could solve such amount in ONE session! It has been such comfortable that I simply could not resist to solve "as much as possible untill I start feeling tired".

      And I am curious how fast at such puzzles most solid GM or SGM would have been. I feel there might be some kind of "eternal limitation" related to the speed of solving - for example 65-70 MPM. And if some GM would have scored 60 MPM after solving 500-800 puzzles - it may not be possible to "make a significant improvement up to 40%". What do you think about it?

    4. There is of course a limit given by the speed of thougts (signals) in our brain ;) The problem with m1-e is that your startspeed is to high to measure an improvement propper. m1-h is slower and would measure your improvement with less "noise"-effects

    5. Tempo said
      "But how do we now that Thomasz hasn't reached FM level already? He might wasting his time with exercises he is already good enough at."

      I did check my performance on easy and hard puzzles in the past it was ( years ago ) balanced according to the findings of Elo. I did check this with several other players. Some players start getting worse at complex puzzles ( either because they are impatient, impulsiv, they loose vision or have a bad thinking system ?? )
      There are no big imbalances in the performing in different tactical pattern. A experience tatctician is not like 1300 at forks and 2300 at attraction

      Tomasz is twice as quick than me in easy tactics which makes him an "expert" in easy tactics. His performance should be reached ( and maybe more ) if we want to become CM.

      And you are right : If Tomasz would like to become a stronger chessplayer or even just a stronger tactician a further improvement in m1 is possibloy already not efficient for him, because of the effect of deminishing return.
      As more we like a task as better we get in this task and as better we get in a task as more we like it.
      As less we like a task as worse we get in it and as worse we get as less we like it
      So by this rule of "nature" we usually learn the wrong things ( without a coach or some other form of guidance )

      But it is ok to do things you like.. Chess is a hobby

    6. @Aox

      "There is of course a limit given by the speed of thougts (signals) in our brain ;) The problem with m1-e is that your startspeed is to high to measure an improvement propper. m1-h is slower and would measure your improvement with less "noise"-effects"

      Of course this limitation is the ultimate one. You cannot conciously express/click the right answer if the signal in your brain does not come oustide ;).

      I understand that I have been too fast to solve M1-e, that's why the system could not show how good I may be ;) :). Nice to know! And I hope M1-h show better calibrated results as I am extremally slow at this kind of excercises. To say it simply - I have not solved more than 1-2K of such puzzles. And to show the HUGE difference - I have already solved about 150K puzzles rated as M1-e (including repetitions of the same ones). That's why the difference between M1-e and M1-h may be extremally big.

      I can agree that "...a further improvement in m1 is possibloy already not efficient for him, because of the effect of deminishing return." I have simply done too much (M1-e) puzzles and that's why I am an expert at these. Now it is time to move on and see how much progress I could make at more challenging ones (mate in 2 or mate in 1 hard, etc.).

      Nowadays I am ready to see how difficult it will be to solve tactics that I do not enjoy (m1-h) so much as the easiest ones (m1-e). What I know for sure and tested it to the full - it is the fact that these m1-h requires THINKING hard and analyse as efficient as possible - to score great performance. It is simply TOO MUCH information (pieces) and TOO COMPLEX environment - to analyse everything into details. You have to work out ONLY the most important info and do it efficient (and really fast).

  4. I recommend to do M1-e first, not M1-h. I started with M1-h first (because back then M1-e was not available). It was pretty difficult, and not much fun, either.
    Then (when it became available) I did M1-e for a while. I could improve ~20% in speed here. Then I tried M1-h - and I scored much better here then, too. Without having trained M1-h. So there was an obvious transfer from M1-e to M1-h.
    The speed improvement in M1-h was about twice as fast (~100%!), but this is mainly because the numbers are so low.
    Initially I solved about 3-4 M1-h per minute, and when I came back after the M1-e training, then I managed about 5-9 M1-h per minute.

    Aox is right - if we cant improve at M1... it might not be worth to do any tactics other at all. Though my impression is, that a 20% speed gain in M1-e can cause a much better result in M1-h. So the impression that a 20% gain is not so much is maybe underestimating the transfer to more difficult tasks.

    It would be interesting to know if Tomasz tried M1-h, too, and if his M1-e training (where he improved a lot) was then beneficial for M1-h, too?

    1. There is transfer from M1-e to M1-h. Why? It is because some M1-h are actually the SAME (type) as the easy ones (M1-e). However most of the puzzles at M1-h are "quite hard" and that's why the effect may not be better than 20% (one of the five "hard" puzzles are easy ones - that's why 20% figure is very good estimation).

      As you noticed - your improvement M1-h of 100% (twice as fast as starting point) simply because your initial number was too low. Try solving 2000-3000 puzzles (as best as you can) and AFTER that write down your score. And at such point - try to make an improvement of 30, 40 or 50%. It will be really impressive if you can do it!

      As your impression "...that a 20% speed gain in M1-e can cause a much better result in M1-h. So the impression that a 20% gain is not so much is maybe underestimating the transfer to more difficult tasks." has been explain before. I hope you agree on it (if not let me know why).

      I have solved just a few hundreds M1-h puzzles (at Aox website, now accessible at Tempo's) and about 1000-1500 puzzles of the same type (chess composition and study - not "too artificial ones with hundreds of pieces") from paper books. But they most contained about 5-7 pieces. That means I am quite "non-infected" with the hard (m1) puzzles.

      As for now I can tell you that M1-e puzzles has been beneficial to solving M1-h puzzles mostly as the some "hard puzzles" are actually the simple ones (but with a tiny difference). It is the same as you would have to solve 100 math puzzles and in one of the five - you would have a huge hint ["(5+6*7+9) *0" - score mutliplicated by zero is ALWAYS zero].

      Let me know if everything I explained is understandable to you. And if you have some questions or comments - go ahead!

  5. Some of the readers may ask: HOW can we know if Tomasz words are true? Are there any proofs he REALLY reached 44MPM (at mate in 1 easy) or is it just his imagination or expectations? ;) :). Below you can see the screenshot (of course you can always claim - everybody can make such picture with today's technology, but it is not my problem).

    Actually I wanted to score 45MPM, but I could not do it. I had 44.60 or 44.75, but not 45 MPM. Anyway I set it now as my ceiling because I am not able to break 45 MPM (at constant speed not just for a few seconds speed) with the use all of my present skills. That's why I need to test another tool to improve my missing parts (to be able to break 45 MPM). What I know for sure is - you have to solve ALL the puzzles with "one shot". What do I mean? You must NOT think about the position, you have to KNOW it! If you start thinking about the position you are NOT the player who are above 45 MPM speed (measured by the website's speedometer).

  6. I think FAC has a positive influence at the puzzles we have been solving. I do not know how big it is, but has to influence the speed of solving because if you cannot see all the LEGAL checks - the positions you have not memorized cannot be solved instantly (in our case - most of these as the database is TOO big to memorize, at least I have not memorized more than a few thousand of positions).

    If you memorized the position you do not have to recognize all the checks, because you can give mate with "just the weapon you know" (checkmate move you have to play and you memorized).

    @Tempo - would you be able to write a post related to 'how to recognize all the checks' when the position is crowded? I am especially interested in reading about the long and 'hidden moves' - it is a shame we cannot PERFECT this exercise (such a trivial one!). By the way - what is your best score at FAC?

  7. I have just solved another 5K chess puzzles (M1-e) and I observed that I could find the solutions with less effort. It is hard to say how much FAC has influenced on the average, but I felt most puzzles were simply EXTREMALLY easy. I have never had such experience unless I felt such things very deeply. What strucked me the most? After 5 minutes of solving I scored 43 MPM and the score has been restored unless I finished solving 5000 puzzles. And I did not feel much fatigue or I was not extremally tired while solving the puzzles! (try to imagine solving 5000 puzzles not being very tired with such number!).

    1. The questions ( for me ) are:

      1)Did you improve in FAC?
      All of these exerxises have an initial phase of "quick " improvement which is usually just the effect that you get used to the task, the user Interface and so on. It takes a few hundred ( 1000? ) puzzles
      After such a phase there is a ceiling ( no flow , sticking around, nasty feeling "holes" ) or still some improvement to a high(er) speed with some type of flow.. it gets automatic,its done by itself

      2) Did this improvement in FAC help you in m1-e or >> m1-h?
      You are alread very fast at m1-e maybe this prog is not good im measuring your improvement, but m1-h should be able to show it ! ?