Visualization Journey

Starting November 1, 2023, I began actively training chess visualization, with the goal of eventually playing blindfold games. Exactly 50 days later, on Dec 20, 2023, I played my first blindfold game.  It was against a bot and I had infinite time, but nevertheless, a major milestone.

This post documents my training history.  Every day, I wrote a small note about what I trained, which I have included in full.  I have also summarized it, so you can get a sense of what I did, and then you can check the detailed notes for more information if you care.

Summary

For the first month, I worked with single pieces on an empty board.  My goal was to improve my "logical infrastructure," as my friend Aiden Rayner would call it.  For example, if I place a Bishop on a board in front of me, it takes no effort to see the diagonals and where it can go.  I wanted to get the same clarity in my mind (and for all the pieces).

I did this in three stages: dark-squares and their associated diagonals, then the light-squares, and then Knight moves.  After ten days, the dark-diagonals were seared into my mind; roughly the same for the light-diagonals.  Knight-moves took longer to get comfortable, and honestly, I find a Knight on an empty board much harder than just playing a regular game.

Starting in December, I began starting from move 1 on the full board.  This was much easier, despite the large jump in pieces; I have the initial position memorized, and things like 1.e4 e5 2.Nf3 require no effort to "see".  For the first 10 days, I did pure blindfold analysis, either making up moves on my own or recalling old games.  For the following 10 days, I reviewed complete games, reading the notation and following along as best as I could.

On December 20, I actually felt ill and took the day off work.  I had more or less recovered by the afternoon and thought, what the heck, let's see what happens.  I played a game against Stockfish level 1 and, though slow, I managed to not only play blindfold but to to play fairly well.  I was fortunate to mate the computer on move 17; a longer game would have surely taxed me much more.

What Do I "See"?

When I first started, I didn't see anything; I simply had a "knowing", if that makes sense.  If you ask me 5x5, I just know it is 25; I don't have to count.  Similar, I don't need to "see" 1.e4 to know that it opens up the Bishop and Queen, it controls f5 and d5, etc.

During the first month, I spent a lot of time just moving Bishops on my imaginary board.  The more I did this, the more my "knowing" translated into something "visual".  Parts of the board began lighting up. In this regard, my decision to focus on the dark-diagonals first seemed quite smart. By focusing on half the board, it came into sharper focus, and the other half stood out by virtue of its negative space. Working half the board worked the full board, in other words.

I also discovered the power of working with smaller chunks of the board.  Visualizing an 8x8 board took effort, but visualizing a 3x3 board was effortless.  A 4x4 board was also pretty easy ... and the full 8x8 board is just four such boards, each one a quadrant.  By focusing on each quadrant, I could then "fit" the full board together.  Spending time just moving pieces around a 16-square 4x4 board was thus fairly easy and allowed me to expand into the full board by putting them together. 

Example of the 4 Quadrants being put together.

Sometimes the full board is available and fully lit; othertimes one quadrant at a time is clearly visible, and I have to mentally connect them.  I have found the quadrants have diminished in importance over the last week or so as I've gained more practice.

By the end of November, I definitely saw a full board most of the time.  2d, in colour, white exclusively on the bottom.  I can sometimes "rotate" the board, so I can see it from any of the four main angles. This is a terrible image but it shows the idea:

According to Aiden Rayner, this is somewhat unique, as a good chunk of chess players do not visualize this way. This was definitely a learned skill, but it also came about naturally, if that makes sense.  The more I thought about chess without a board, the more the board appeared in my head. I spent a large chunk of time working on the "mental scaffolding" part, on constructing the empty board, which definitely helped.

Final Thoughts

The empty board is harder to see than a full board, oddly enough. I'm not sure I can play an endgame with an confidence.  The fewer pieces, the fewer geometrical relationships on the board, and the more likely something gets lost.

Currently, memory is the biggest challenge.  I can see the position well enough, I guess, but I frequently forget which square a piece is on.  For example, White almost always plays Nf3, but if that Knight delays and sits on g1 for a long time, I start assuming the Knight must have moved and place it on f3 in my mind.  Other times, Black plays ...Bf5 and then doesn't touch that piece for 20 moves, in which case it gets lost in the fog.

The mental stamina builds fairly quickly. I jumped into 30min training sessions, struggled for the first week, and then was more or less fine. Earlier on, when I was "running on empty", my brain basically shut down; in December, it was rather a slow down.

As for next steps? I will likely play several more training games against the computer, then graduate to timed games, and then take the plunge with games against real people.  As these will all start from the initial position, I will probably do some isolated endgame practice sessions, to keep working that empty board. I will record every game I play; not sure if it will be YouTube content, but having a record is always fun.

My Daily Log

Below are my training notes for each day.  It's in pseudo-point form, and it might only make sense to me, but I've included it for reference.  If nothing else, you will get a sense of my mental state and motivation throughout.

Edit: I have stopped updating the below chart.  If you want to "see" my improvement, check out the following YouTube playlist.  It is unlisted, so it won't appear via regular searching, and it shows my annotating my games in real time.  After the third video or so, I figured out how to include a live board, so you aren't just watching a blank chess board for 20min.

1. Nov 1, 2023: 30min total, 20min reviewing diagonals (placing Bishop on a square, clearly seeing the diagonals, and then moving one square forward, seeing the new diagonals clearly, etc); my brain then felt tired, so spent the last 10min on the Opera Game.
2. Nov 2, 2023: 30min total, 22min reviewing diagonals (all dark diagonals, working Black's half of the board); brain went to mush; last few minutes on seeing the quickest mates possible
3. Nov 3, 2023: all 30min on dark diagonals; strong for 28min!
4. Nov 4, 2023: full 30min, np!  Could have done more.  Played on the long dark diagonal: Bishop on a1, see diagonals; move to b2, see diagonals, then back to a1; then a1-b2-c3, etc, slowly stepping forward, always seeing the diagonals clearly before moving on.  Only d4 and e5 gave any trouble, rest were relatively clear!
5. Nov 5, 2023: full 30min; same drill in reverse, start at h8 and move to a1, repeating every diagonal as I went along.  d4 and e5 clearer today.
1. Bonus! At the risk of TMI, did extra visualization practice during my bath.  Went through some opening lines very slooooowly, striving to see the position clearly on every move.  Oddly, the full board is much easier to "see" than just a lone Bishop on diagonals.
6. Nov 6, 2023: 30min, much harder today; brain not in it it seems.  Still working on dark diagonals, this time placing a Bishop on every dark square on the outside of the board (a1, a3, c1, etc) and visualizing the diagonals.  Last four min I switched to simple opening positions as a break of sorts.
7. Nov 7, 2023: 30min, last 4 was a struggle; "traveling diagonals" exercise: put Bishop on a3, see all the diagonals, and then move the Bishop to each of those squares and see those diagonals; then put Bishop on a5, repeat for all dark-coloured "edge" squares
1. My goal is to get a firm handle on all the dark diagonals, have them fairly strong in my head, and then add the light diagonals.  Taken together, that amounts to the entire board.
8. Nov 8, 2023: Wow, felt amazing today.  The board felt crystal clear, and it stayed that way for the first 15min or so.  Afterwards it was no longer crystal but still clearer than anything I've every felt.  The diagonals were absolutely jumping out.  It feels like all the drawing of diagonals in the last week has built a foundation that I was able to jump off of today.
1. For the record, 30min, still focusing on dark diagonals.  Put Bishop on every dark square on White's half of the board and drawing all the diagonals.  This builds off itself, because a Bishop on d2 is identical to Bc1+Be1 (or at least half of each).
9. Nov 9, 2023: 30min, though not as clear as yesterday.  Did the same thing as yesterday but on the Black half of the board.  Had a busier day at work, so that may have drained my mental reserves, or perhaps yesterday was simply an outlier.
10. Nov 10, 2023: Full 30min.  First 12min I did the "Bishop's Tour", starting on a1 and moving to any square afterwards, always seeing the full board / all diagonals before moving on.  Last 18min I had a tactical position I had memorized (from THIS YouTube video), and I went through various lines.  I had memorized the solution, so I didn't solve the position blindfold, but I did explore other lines and had a decent view of the mental board.
1. Celebration Time: Ten days in a row at 30min each puts me at 5hrs of visualization practice!
11. Nov 11, 2023: Full 30min, switched over to light diagonals.  This was surprisingly hard!  I think it's because everything is inverse, eg, the dark long diagonal runs up the board from left to right, and the light diagonal goes left to right.  Once I get a handle on this inversion, I think I'm fine.  Spent last 12min or so on the "Colours Reversed Stafford Gambit", exploring variations after 1.e4 e5 2.Bc4 Nf6 3.Nf3 Nxe4 4.Nc3!? Nxc3 5.dxc3, which was quite fun and fairly successful.
12. Nov 12, 2023: Full 30min, again on light diagonals.  Much better today, but not as good as the dark ones yet (no surprise, I've only done two sessions!).  Similar to Nov 4th, I played on the long diagonal: start on h1, see diagonal, move to g2, see diagonal, then back to h1, g2, f3, etc, slowly working up, revisiting every square.
13. Nov 13, 2023: Full 30min; same as yesterday, but starting on a8 and working down towards h1.  White side is easier to see, for some reason; h1-e4 are very easy, whereas a8-c6 are tricky and d5 is especially opaque for whatever reason.
14. Nov 14, 2023: Full 30min; brain somewhat slower today.  I blame work.  Same as Nov 6th but on the light diagonals.  Only managed to get one side of the board done (h1, h3, h5, h7).
15. Nov 15, 2023: Something about Wednesdays?!  Excellent session.  Board was alive.  I moved the Bishop effortlessly across the diagonals for the first 24min, with the first 15min being almost too easy.  I experimented with two Bishops at the very end, keeping one stationary and moving the other, making sure diagonals stayed fresh.  Felt great.  Full 30min.
16. Nov 16, 2023: Another great day, almost as good as yesterday ... for about 22min.  Then I hit the wall and slowed way down, but still decent.  A continuation of yesterday's "edgework", putting Bishop on edge of board and then moving to every available square, seeing the diagonals on each move, etc.
17. Nov 17, 2023: Full 30min; 15min spent doing the "Bishop's tour", like on Nov 17th; other 15min spent trying to analyze the Muzio Gambit in my head.  I knew the double Muzio was the way to go, and I could see up to the tenth move fairly clearly, though it was very slow and my mental reserves were clearly draining.
18. Nov 18, 2023: Full 30min ... barely.  Brain very tired, possibly most tired I've felt while doing these exercises.  Ended up doing Rooks, which I figured would have been easy!  The visual board itself was, but it took effort.  Also spent time trying my "quadrant" idea on the center (eg, the 16 squares from c3-f6).
19. Nov 19, 2023: Again, barely 30min.  Did a little bit of Rook visualization, and then tried doing the RK v K mate.  I made it super simple: Black king on back rank, Rook on seventh, so all I needed to do was bring the White King closer (and not hang the Rook).  Took a lot of effort to keep the position in my mind.
1. As a bonus, I discovered something about these endgames!  Basically, once you have the opposition, all you have to do is make a passing move with the Rook.  Your Knight now just shadows the enemy, a straight line to checkmate.  (Before, I'd do the "slow Rook shuffle," getting the King to the corner before trying to mate.  This is much faster.)
2. 
   
    
20. Nov 20, 2023: Full 30min, divided into two 15min sections.  First, I did some simple Rook maneuvers; seeing the Rook on my mental board is now extremely easy.  I tried using two Rooks, moving one after another, which provided more challenge.  Second, I started visualizing Knights.  Moving Knights is easy, because I know, in the same way I know 5x5=25, that a Knight on b1 can go to a3, c3 and d2.  I know that, but seeing it was extremely challenging.  I predict I will have to spend a large amount of time drilling these Knight maneuvers.
1. Celebration Time: 20 days in a row!  That's 10hrs of visualization practice!
21. Nov 21, 2023: Full 30min, with all Knights.  I found a way to make visualizing Knights easier: Bishops!  That is, a Bishop on d4 controls all of the same squares as a Knight on d1, so imagining the diagonals helps make me see the Knight moves.  Much easier, in fact ... though still not that easy.
22. Nov 22, 2023: Full 30min, all Knights.  I am placing the Knight on every square on the rim (a1,b1,c1; a2,a3,a4, etc), because here, the Knights have the fewest moves.  I'm thinking if I can clearly see the Knight moves on a4 and h4, respectively, then placing a Knight on d4 is effectively combining these two "rim patterns".  Simply put, the stronger I get at seeing a Knight with only four possible moves, the easier it should be to see Knights with all eight moves.
1. and the "Bishop diagonal" Knight hack is really powerful.  Much easier to see Knight moves.
2. 
   

   

   Placing two Bishops lets me easily see all possible Knight moves.
   

   
   
23. Nov 23, 2023: Full 30min (but work was busy and had to fight to squeeze it in!)  More of the same, Knight moves.  Similar to the Bishops, I'm working just in Quadrant 1 right now, placing the Knight on each square and focusing on the moves, then move to another square.  The range is hard: a Knight on c3 can jump to both d1 and d5, and seeing that clearly, at the same time, takes enormous focus at present.
24. Nov 24, 2023: Full 30min; basically same as yesterday, placing my Knight on all 16 squares in a quadrant and striving to see the resulting jumps.  I'm getting better, but still a challenge.
25. Nov 25, 2023: Full 30min; same as yesterday, but using the White Kside quadrant (my default is the White Qside, the square a1-d4).  Plan to do the same for all four quadrants, and then maybe just the center quadrant as well.
26. Nov 26, 2023: Full 30min, but change of plan.  I focused on the smallest possible board (3x3) and looked at Knight moves.  Very clear.  I then expanded to 4x4; less clear ... but more clear than normal. I think I will keep doing this until it is comfortable and then try 5x5, which is when the full 8 moves become possible.
27. Nov 27, 2023: Full 30min, fully explored the 4x4 board with all pieces, though focusing on the Knight.  Clearest the Knight has felt, though fog crept in the last 6min or so.  (As an aside, the last two sessions have felt very fast.  Surprised when my timer rings telling me I'm done!)
28. Nov 28, 2023: Full 30min, but terrible.  Could barely see, had trouble focusing, etc.  Not my finest moment.  Not sure if it was better than when I started, but that might just be disappointment talking.
29. Nov 29, 2023: Broke the streak! Busy day at work, then helped someone move. Back at it tomorrow.
    
30. Nov 30, 2023: Full 30min, very good!  Started on a1, put the Knight there, saw the moves, moved to b1, saw the moves, etc.  Did that for the entire White half of the board.  I rushed a little, so I wasn't completely "seeing" the moves, but I certainly knew them, if that makes sense.
31. Dec 1, 2023: Full 30min. Switched gears and analyzed "weird" openings completely in my head.  Stuff like 1.e4 c5 2.d4 d5!?, or the Exchange French with early c4 and ...c5 from both sides.  Not bad!  Had to re-play the positions from the beginning several times to make sure it was there, but I could do it.  Checked my analysis on a real board afterwards, and most of it was on point.  No way I could do this a month ago.  I will continue to do this and see how deep I can go.
32. Dec 2, 2023: Full 30min.  Not as easy, but pretty decent.  Went through some funky positions with 1.d4 f5 2.Nc3!? and 2.f3!?
33. Dec 3, 2023: Full 30min.  I tried to recreate some of my favourite games.  I didn't know the exact move orders, but I did a decent job and got to around 12 moves.  I could have gone further if I were just exploring the positions, but I wanted to get it "right" so didn't go further.
34. Dec 4, 2023: Full 30min.  Successfully went through one of my all-time favourite personal games from memory, blindfold. It's the final game in my "Smithy's Minis" course, ftr.  I had analyzed this game fairly recently, so it was relatively fresh.  Replaying the moves was fairly good, but not sure how much I could have calculated different moves with any accuracy.  Still, going 27 moves with decent clarity, I will take that!
35. Dec 5, 2023: Full 30min.  Went through my favourite King's Gambit game (definitely got Black's move order wrong, and couldn't fully see the various tactics, but I "felt" the general flow of the game well).  Also explored some funky lines in the French Advance, just for fun.
36. Dec 6, 2023: Full 30min.  Started randomly looking at lines.  Eventually landed on the Philidor, and then the Philidor Counterattack with 3...f5.  I analyzed the following line perfectly, if not quickly.  I struggle with King hunts normally, let alone blind, and I have to thank my intuition my guiding me to the right path. (Can't link to an analysis board, so here's a gif of a game that goes down the same line.)
1. 
   
   
   
   
37. Dec 7, 2023: Full 30min.  Recalled a complete game from my Smithy's Minis course. I'm definitely following the train tracks, so to speak: I can go forward on the preset path, and I know what the moves are, but I have little ability to veer off and explore other lines... yet! That said, I'm astonished how well I can see the board; it's not just a vague knowing, and I can "look" around and see certain things.  It just takes effort and concentration, something not normally required when glancing at the board.
38. Dec 8, 2023: Full 30min.  Same as yesterday, went through more miniature games from memory.  Went extremely well; easiest it has ever felt.
39. Dec 9, 2023: Full 30min.  Brain foggier today, but I also went through a much more complex game, and I tried to at least look at alternate moves at certain junctures.
40. Dec 10, 2023: Full 30min.  I ran out of games I had memorized; the games I tried stopped halfway through, as I couldn't remember the continuations.  
1. Tomorrow, I will switch to reviewing games I don't know via scoresheets, seeing how far I can follow them and if I understand what is happening.
41. Dec 11, 2023: Full 30min.  Completed this game with surprisingly good clarity! Much smoother than I would have expected.  I saw the threats quite distinctively, such as the threatened ...Nb4 fork, but I also lost some of the Qside pieces towards the end.  Excellent first step, though, very happy!
1. Celebration: 20 total hours to visualization!
   
42. Dec 12, 2023: Full 30min.  Completed this game.  Much more complicated, more material discrepancies, more tactics.  I followed the general flow of the game, and I saw the game-ending tactics, but the murky middle section was less clear.
43. Dec 13, 2023: Full 30min. Went through this game, but did not fully complete it. The first 12 or so moves were crystal clear, possibly the clearest my internal vision has ever been; shortly afterwards, I misplaced a piece, got mixed up and never fully recovered.  Even discounting that, I missed a number of important elements that were obvious when I reviewed the game with pieces.  Still, a step forward, I feel.
44. Dec 14, 2023: Full 30min.  Got 34 moves through this game. Position generally made sense and I could see it, but I kept forgetting about my Bc4.  Pretty good.
45. Dec 15, 2023:  Full 30min.  I somehow completed this game, even though I only got 3 hours of sleep last night. I was going to skip training, but ... I couldn't sleep, so why not? Did very well; only one gap (didn't "see" that 23.exf6 also attacked the Queen). I'd be more excited if I were more awake.
46. Dec 16, 2023: Full 30min.  Went through 41 moves of this game. The blocked nature made it difficult at times; I definitely lost the thread a few times, but I'm forgiving myself, for this was probably the wildest game I've visualized so far.
47. Dec 17, 2023: Full 30min. Got to about 38 moves of this game, but very badly.  Possibly my worst attempt this week.  I've noticed that a) if a piece is "normally" developed early but isn't [eg, Nf3], my mind assumes it has been developed, and b) if a piece doesn't move for a long time [eg, the ...Bf5 in this game, which is developed on move 6 and doesn't move against until move 25] I forget it exists.
48. Dec 18, 2023: Full 30min. Pretty lousy; got to move 27 in this game, but had to restart once because I completely lost the position.  Second time was much cleaner, but not my best outing.  (My tactics warm-up were also very slow, so not just a visualization thing.)
49. Dec 19, 2023: BONUS 33min, because I was only a few moves away from finishing this game and decided to just go for it. Not as clear as my best days last week, but still very decent. I consider it a very good sign if I can predict the move before I read it, which happened several times with this one (even if the clarity was a little rough and effort needed to focus was quite high).
50. Dec 20, 2023: I DID IT! I played a full blindfold game! It was against Stockfish level 1, infinite time, but I still did it!
1. Link to the game, to my video proof and to my celebration post.
   
51. Dec 21, 2023: Played second blindfold game, 22 moves, almost all of which I would have played had I full sight (just much faster!).  Found a pretty nice mate in two at the end, always nice..  Link to the game, and link to video proof.
52. Dec 22, 2023: Leveled up! Third blindfold game, 35 moves, against Stockfish lvl 2 this time.  A much harder game, with some significant visual gaffs.  I kept most of the position, and almost all of the key items, and I played most of the best moves (even if I did so for the wrong reasons at times!).  Link to the game, link to the video proof.
1. I am recovering from a slight cold, so you will hear some laboured breathing at times.
2. Bonus Game! I was just playing around, attempting to troubleshoot why my recording was not getting the live board to auto-update, and I happened to mate the computer with a great opening attack.  Took 4min... though I didn't troubleshoot my problem, so I guess the game was a failure :D
3. That said, I did figure out how to do it.  In OBS, the secret is to have it as a Browser capture, not a Window capture.
53. Dec 23, 2023: Played a fairly decent but frustrating Ragozin against Stockfish level 2.  Difficulty was less in seeing the board than in playing the resulting position!  Link to the game, link to the video proof.
54. Dec 24, 2023: Terrible game.  Computer played stupid moves like f3, c3, d3, g3, h4 all in the first 8 moves or so.  Reached a position that was difficult to break through even with sight.  I actually dropped a piece (due to a tactic, not necessarily because of blindfold), but the computer didn't take it and blundered thankfully into a mate in 1.  Not my happiest experience.  Link to the game, and video proof.
    

Future Training Idea

Dec 11 or so: go through actual notation

• Light diagonals
• Rooks
• Knights (can do more work isolating these; Knight endgame analysis perhaps?)
  
• Queens
• Simple Mates (KQvQ, trying to mate in less than 10 moves; double Bishop mate)
  
• "paint the board", start at corner, slowly fill in the board, stop at every square