Memfis Posted February 18, 2018 I like chess but my big problem with the game is that once most figures enter the battlefield and the opening ends I simply don't know what to do. Unless my opponent is clearly a bad player, typically there are no obvious weaknesses that I can exploit, which puts me in a confused state where I just make small insignificant moves and wait for something to happen. I think because of this I tend to prefer defense, as when I'm being attacked I at least have some understanding of what I need to focus on. I also usually want to exchange as much as possible as quick as possible so that we can enter endgame where it's easier to see what my goal should be. Anyone else here struggled with this? What can you recommend? 2 Share this post Link to post
ShoDemo Posted February 18, 2018 When I play chess, I am a bad to average player. Usually I do what you said about small random moves and exchanges are a must for me, because they don't always need you to think a lot, about how you can do them. However, one friend of mine, for whom chess is a hobby, told me that you should do exchanges only if you don't know what else to do. In the meanwhile, he has read books and learnt chess openings from many sources (like videos) and he prefers setting the board in a way that he can remove pieces of the opponent, with no exchanges. Once you have an advantage from the opening or you are in equal ground with the opponent, because he countered the opening, you could probably use some tested moves, if the scenario is the correct one. And protect your pieces, but I don't know if protecting multiple pieces, with just one, is a good idea. Probably it isn't worth it, because if you accidentally move that one protecting piece or if it is taken out, bye bye protection. Essentially, look online for tips and maybe find a book or two, to see those essential moves that you must know (and maybe look at some beginner mistakes, to get an idea of what you should and shouldn't do - I may be doing lots of them when I play). 2 Share this post Link to post
Fonze Posted February 18, 2018 Chess is a lot of fun and I think it's great that you want to learn more about it memfis. A few pointers that i can give: Don't waste moves. Those small, piddly mobbes you're talking about; if they gain you a new position they are good, if you wind up back in the same position, well, you might as well have moved something else. This goes hand-in-hand with the last point of not wasting moves. Spend the beginning of the game developing pieces, not moving around 1 or 2. It is very common to see somebody move a queen around, only to allow their opponent to develop their pieces while constantly pressuring the queen. As you develop pieces, remember that knights and bishops favor closed boards while rooks favor things being open. This means that you should ge t your bishops and knights out asap in the early game and treat them as more expendable late game. Trading pieces is a mixed bag and it's very easy to miss opening a weak spot in your own setup when you feel bloodlust for captures. Defensive play is overrated; in chess, just as you don't want to waste moves, you don't want the opponent to be one move ahead of you, as that's all it may take for them to win. Proactive>reactive. If you have trouble developing a clear strategy to win, try playing for positions. Or in other words find squares your pieces have the most potential to be deadly and deploy accordingly. Closing the board on your opponent will not only make their game harder, it will also make it easier to make things happen for you, just don't close yourself in. Again think of knights and bishops as you decide on how much to close the board. Lotta other tips, but thatll do it from me for now. Gl and hf; chess is a great game :) 4 Share this post Link to post
Searcher Posted February 18, 2018 1. Play against a computer and see what it does to kick your butt. It is quite a learning experience. 2. Read chess books like Grazza's 3. Control the center of the board. 4. Even if you have to defend, make sure that move is also an attack. Never waste a move. 3 Share this post Link to post
fraggle Posted February 18, 2018 I'm no Chess expert but the advice I remember hearing is that a huge part of the learning curve is just learning not to make mistakes. A single blunder or bad move can lose you a game. Assuming you're still a novice, it might be better to focus on the basics like counting material value and basic pawn structure rather than worrying too much about higher level planning. Grazza can answer better but my understanding is his books are for more experienced players. I found Chess for Dummies was a useful intro that introduced all the basics. 2 Share this post Link to post
Fonze Posted February 18, 2018 I didn't know grazza wrote books on chess, but I'd love to read them if anybody can point out where to find/purchase them. 0 Share this post Link to post
MathsDevil Posted February 19, 2018 Hi guys, just happened to notice this thread, I am a keen chess player --- whether I am good depends on who you compare me with! Chess is such a deep game, there's always more to learn. Recently I have enjoyed this web site by Ward Farnsworth, which is accessible to many skill levels, including beginners: http://www.chesstactics.org/ After I knew the rules, there are some general pointers I try to follow: 1) When playing White, don't be arrogant. Black's opening may look slow but I must prioritise developing all my pieces unless there is a very obvious weakness to attack. 2) In the opening, be aware of squares your opponent wants to put pieces on and get those squares covered, usually by pawns or minor pieces. 3) Don't rely on a powerful/important piece, especially a queen or king, to defend a less valuable one in case the more powerful/important piece is chased away from its defensive duties by an opposition attack. 4) Be aware of checking moves at all times. Don't give check just for the sake of it, in case your piece is chased away with no profit for you. However, checking moves can give you a 'free' move because the opposition must spend a move saving their king. 5) Sometimes disabling opposition pieces is just as good as capturing them, such as hemming in a bishop with pawns. 6) When it is my move, I spend a long time looking for threats against me. Then I search for attacking possibilities and consider the threats against me which may result from them. My game thinking is 90 per cent defensive, bizarrely this results in better attacking play in the end. Above all, have fun playing against people at a similar level to you or a little higher --- over the board experience is a brilliant way to learn the game and have a good time too! 3 Share this post Link to post
Tristan Posted February 19, 2018 6 hours ago, Fonze said: I didn't know grazza wrote books on chess, but I'd love to read them if anybody can point out where to find/purchase them. Here's a list if that's of any help 2 Share this post Link to post
Urthar Posted February 19, 2018 I'm not a particualrly good chess player by any means, but my general strategy (which is more or less the same one used by most chess programs) is to increase the mobility/threat-projection of my own pieces, while restricting the enemy pieces. 1 Share this post Link to post
CARRiON Posted February 19, 2018 You'd think being a fan of card games, RTS games, and TBS games I'd love chess but honestly I kind of loathe it sometimes. Probably doesn't help that anytime I tried to play it with family or friends they'd be actually decent at it and show me absolutely no mercy whatsoever. Whatever though, Battle Chess was awesome and probably the only reason I ever took the small interest I had in it in the first place. 1 Share this post Link to post
pablogener Posted February 19, 2018 I play chess online against other ppl on chess.com. I use the 'free' service they offer. After the game ended, you're presented with a simple (not 'deep') analysis of your moves. On every 'blunder' or flat out 'mistake', you're given a '8-10 moves' alternative, showing you a better way to resolve a tight situation you made a mistake with. You could possibly identify that situation at future games and have in mind that alternative the computer showed you. Sometimes the computer can't 'understand' where you're going to, or what you're planning, and will declare a 'mistake' where maybe you were luring your opponent in or something, but pretty much it makes sense when showing you an alternative to your moves. I found this method to be very good for learning how to identify 'game stages' and things that happen routinely in a game and have a 'map in my head' of possible things to do to have an advantage over my opponent. that's my 2 cents. cheers! 2 Share this post Link to post
Grazza Posted March 10, 2018 (edited) A bit late to this, as you posted when I had just set off for a trip into the Patagonian wilderness. It sounds like the problem is that you are viewing the opening and middlegame as two separate units. That is, in the opening you are following the standard opening goals of getting your pieces out and controlling the centre. That's fair enough as a way to avoid an opening disaster, but may leave your pieces without any natural plans going forward. Further opening goals (which become more important as you become stronger as a player) are to develop purposefully with a view to possible middlegame plans. Carve out possible strong squares. Inflict weaknesses. Maybe building up a strong and flexible pawn-centre. Maybe keep the king flexible if it isn't yet clear where you will want to attack. If your opening play has been purposeful in this way, then your possible plans should become clearer. Do you have some strong squares? Then use them. Have you inflicted weaknesses? Put pressure on them. Can you advance your centre pawns and stifle the enemy pieces? Then, er, do so. You get the picture. And if you have a clear plan to attack the enemy king, and have still kept your king flexible, then determine how best to make use of that situation: maybe castle on the opposite side (assuming the centre isn't safe) and throw everything into an attack. But remember that it might be best to castle on the same side and attack mostly with pieces. Don't confuse "defence" with "being passive". Defence is rather a dynamic thing, as you are trying to counter the opponent's ideas by minimal means while seeking ways to counterattack, possibly using sacrifices of your own. Simply sitting there waiting to be attacked and hoping the opponent screws it up is not defence. Always when playing chess, spare a thought for what the opponent is intending. This will sometimes suggest your best move straight away, but don't always assume you must respond. You might be able to reply with threats of your own that render your opponent's ideas irrelevant. Or neatly parry the opponent's plan while doing something useful for your own position. Bear in mind that abstract considerations have limited value. You'll only really learn how to play better by playing, but you'll progress faster if you have a good idea of what you should be doing in general terms. Like going to Patagonia: you can read a guide book all you like, but it is no substitute for visiting. And once you have visited, a lot of that stuff in the guidebook will make a lot more sense. Edit: "I also usually want to exchange as much as possible as quick as possible so that we can enter endgame where it's easier to see what my goal should be." This is a really common error in pretty much all levels of play below master. Reducing tension (generally by initiating exchanges of pieces or pawns) is in general a concession. Only do so if you really gain something specific (or truly have no better option). Otherwise let the opponent initiate exchanges if he wishes (as long as he isn't gaining something specific, of course). Why? Well, by keeping your options open, you increase your range of possibilities, and this means that your opponent will have to be ready to meet all of them. There may be an ideal reply to each of them in isolation, but a reply that serves best in all cases may be more of a compromise choice, and allow you then to choose which option (e.g. exchange, advance or retain the tension) is most favourable to you. This is a real point, and not just a psychological one. A situation where you can resolve tension without conceding much (but the opponent cannot) is called favourable tension, and something that strong players seek to create. Yes, this is an advanced topic that will make more sense with experience, so feel free just to take my word for it right now. Edited March 10, 2018 by Grazza 15 Share this post Link to post
Memfis Posted March 26, 2018 (edited) Thanks for awesome responses. The idea of an exchange being a concession is a new one for me and I found it especially interesting. So you typically want to keep all options and put as much pressure on different parts as possible, this makes a lot of sense! Reading the opponent's thoughts is something I'm having a lot of trouble with actually. But some sites allow you to flip the board while playing, which is surprisingly helpful. :) Btw does anyone follow the Candidates Tournament? It's quite interesting, there are only 2 days left and yet literally anything can happen. So many strong players have almost the same score. What an exciting time. Also if someone here wants to play, let's arrange some games! I'm currently using lichess.org but I'll be okay with other options. 3 Share this post Link to post
Fonze Posted March 26, 2018 I'm always down for a game memfis :) Stewboy got me onto gameknot.com, and that's my main chess site now; but if need be I can always make an account elsewhere I suppose. Chess online is more fun when being played against people you are familiar with outside/before a chess match :) As much as I love being taken to school, somehow it's less special when against a random, heh. 1 Share this post Link to post
Soundblock Posted March 27, 2018 I played some chess as a child (1st game at 4 or 5), but gave it up before starting school (got demoralised by my uncle thrashing me in a handful of moves). Fast forwards 30 years and a friend of mine who was kinda into it wanted someone to play during a mountain hike (we were in a cozy cabin). Since it had been ages I asked him for pointers to refresh my memory during our first game, and we openly discussed every move made during that game, allowing me retreats if I made obvious blunders. On the second game we played normally and I figured I'd try a move I'd always wanted to do, but never pulled off while I was a kid - getting a pawn all the way to the other side of the board to make a second queen. So, while he was playing what I guess was his meticulous opening routine and not expecting much from me, I was exclusively moving up pawns that protected each other (a diagonal line starting at the far right of the board). By the time my friend realised what I was doing we had already traded queens and though he had a major advantage in the remaining pieces before I spawned the second queen, she quickly mopped the floor with what was left. That's the last game I've played, well enough to retire on I think. Anyhow, to answer your question more directly (it is a subject I also struggled with the short while I played) I suppose if I played again I'd try to: - not be so aggressive as to soon thereafter have to retreat/lose a piece moved forwards to the opponent half - play turtle style, try to get the opponet to overcommit - save towers for the endgame, spend knights & bishops early, if at all - if no defensive moves to save pieces are neccessary, shuffle up that diagonal pawn line I'm sure someone who plays actively can give you much better pointers 2 Share this post Link to post
geo Posted March 27, 2018 (edited) Wait until it looks like you're about to lose... then bam hit your opponent with a punch yelling check mate mutha fucka! Then flip the table and walk away to find the next person that will never see your strategy coming! I mean, that's why chess was invented, two nerds too afraid to fight head to head. This is why chess is calling for steroid testing. 2 Share this post Link to post
Grazza Posted March 27, 2018 On 3/26/2018 at 11:55 AM, Memfis said: Reading the opponent's thoughts is something I'm having a lot of trouble with actually. But some sites allow you to flip the board while playing, which is surprisingly helpful. :) Well, you're not looking to read their thoughts - just to understand their moves and what ideas might logically flow from them. If they are actually intending something illogical, then it is unlikely to go well for them. Yes, it can be useful to look at the position from your opponent's side of the board, though in an over-the-board game you shouldn't do this in a way that annoys the opponent. Don't stand right next to them, and definitely don't sit on their chair if they get up from the board! (No, I've never seen the latter, but sometimes the former.) 3 Share this post Link to post
VGA Posted March 27, 2018 A good book for improving players is this one, which is also published by Grazza's company. These opening systems and their simple plans are great t get started with and have an idea of what you are doing as White. https://www.amazon.com/Killer-Chess-Opening-Repertoire-enlarged/dp/1906454183/ref=sr_1_1?s=books&ie=UTF8&qid=1522194415&sr=1-1&keywords=Killer+Chess+Opening+Repertoire It has annotated model games so you can understand the plans. Although players that hang pieces or generally push pawns around withouts any idea what they're doing should just train tactics, solve puzzles, play longer games, watch educational chess videos on Youtube etc. 2 Share this post Link to post
Memfis Posted April 6, 2018 Pls add me on lichess: https://lichess.org/@/kuchitsu So far I've been playing 10 minute rapid, but we can try other time controls (or even no time control) if you want. 1 Share this post Link to post
Hellbent Posted April 10, 2018 (edited) I just had some pretty fun games in chess. I hadn't played in a really long time. I totally agree with your statement about the early-mid game doldrums. I have no idea what to do in the early game. I remember studying a little on chess.com and developing some early game strategy at one point, but there definitely was a large period of my chess "career" where I'd just do my best to get through the early game. A lot of time I'd just focus on creating as many supported developed pieces as possible: either get my bishops out with knight supports, or try to get my knights in the center of the board supporting each other (or next to each other supported by pawns). Other times I'd try to to fork his king and rook, but soon realized this was a cheap pipe dream that wasn't really smart to chase after, but something to keep an eye out for while developing pieces. Often I would look to trade to make the game easier (less pieces, less confusion). Other times, I'd get a real kick out of games where it was like 10 swords all drawn on 10 different necks in a circle (like in the movies). As I was saying, tonight I just finished playing some chess for the first time in a really long time against an otherwise smart enough opponent who was prone to some unnecessary blunders and didn't plan turns ahead; but made some pretty cunning plays from time to time (especially when he was on the back foot). Second game I was pretty even but I fell behind when I foolishly allowed him to get my rook with his queen. But I focused on position and was was a bit clever in getting pressure on his king with my queen. He failed to remember the importance of his bishop as a support in keeping pressure on my king with his queen when I tried (successfully) to divert his attention by putting his king in check with my queen with what largely was an empty and pointless check (all he had to do was move his all-important bishop back to position after blocking the check-- since I had to pull my queen back to a center file in order to defend against imminent mate). I flipped the tables on him when I managed to snooker his rook in the corner with my queen, mirroring my side of the board, when he tried to get creative with his bishop instead of just bringing it back again. I continued to take my time with my turns and slowly outplayed him as I slowly ensnared his queen that was still in the corner from taking my rook early on. And then it finally happened; I encroached on his queen and he had no escape. Once he saw he could no longer save his queen he immediately went on a "hail marry" offensive, not even bothering trying to take the best piece he could with his queen. I scoffed at his pathetic efforts, laughed and derided him. I just couldn't control myself, I had superior board, an impenetrable wall of knights and pawns protecting my side of the board--a veritable fortress, and so I mostly ignored his developing knight and took his queen. Two turns later that developing knight forks my king and my queen! I was royally forked! I was still up on material, but rather flustered from the very surprising fork (I thought he was aiming for my king, not my queen), and I then made some late game blunders and he surprisingly came back to win it. Third game: Very early on, my opponent got his black pawn one square from Queenage. I wasn't too worried about it though, given the state of the board. I made some more bad moves and he didn't capitalize on them for some reason so instead of falling even further behind I took out one of his knights, and several turns later had all his minor pieces and he only had one of mine. He definitely had some sloppy moves, though he did gain some positioning and nabbed one of my rooks. He then pinned my ever important bishop to my queen (my mistake): my bishop was holding his pawn-to-queen square. I was threatening checkmate with a knight in position to support queen-to-king smother mate. But suddenly his pawn was threatening to not only dash my checkmate dreams but totally turn the game around. If I took out his newly queened pawn with my bishop, his queen (threatened by my pawn) would take out my queen, putting me in check! So I had to sacrifice my rook to take out the new queen, which meant by bishop would continue to protect what would now be my rook and his queen would not be able to put me in check. He put me in check with his rook (unsuppported: a sacrifice to get my king one square forward so he could put me in check with his queen) but my bishop was simply able to block the check and he then didn't see any other option so pulled his queen back, but wasn't able to block the smother-mate threat that was never neutralized. Anyway, pretty amateurish games, but definitely some flourishes of creative play to keep things interesting. Edited April 10, 2018 by Hellbent 3 Share this post Link to post
kb1 Posted April 11, 2018 (edited) All good advice, especially Mr. Graham Burgess (Grazza), wow! (General notes...I'm no expert) Chess is a huge balancing act. Defense is on equal ground with offense. Trading similar pieces often neither gains, nor loses much (on it's own). You want to balance the ability to attack with the ability to defend, with a slight pull towards keeping your pieces, all other things being equal. However, depending on the situation, sacrificing a piece, even the queen, and even if you don't get a piece in return, can be ultimately beneficial (sometimes). The only piece that really matters universally is your King vs. your opponents. (What I mean is that, causing checkmate is only second to preventing checkmate!) You may threaten an opponent's piece to provoke an attack, or to prevent an attack, by promising retaliation. Again, "balance". General goals: Prevent your opponent from making effective moves. You do this by avoiding attack, and/or by threating a worse retaliation, or by forcing your opponent to use all pieces in a defensive way. Knights have more mobility near the center. Pawns are usually more effective by being on different files, and often when staggered, forming "arrows" or chains towards the opponent's side. The effectiveness comes from their ability to defend each other. Be very careful with pawns - many people don't give them the respect they deserve. Don't forget to castle, when it makes sense to. Pinning can be very useful. That's when a single piece threatens more than one piece, especially important pieces. Balance! Counter threats with threats of your own. Attack when it makes sense to. Defend when that makes sense. Do both if possible. But, be sure to have a plan based on a balanced approach. That means: If your strategy is to make small, inconsequential moves, your opponent's best move is probably to advance his/her pieces to their most powerful agressive and defensive positions, as quickly as possible, preventing you from ever doing anything but small moves against a fast, massive force! Dish out the worst possible punishment, while suffering the least worst fate. Chess engines look at literally millions of moves, and look many moves ahead, with one goal in mind: Choose the move that causes it to suffer the least worst fate. That's the best advice I can give. Good luck! Edited April 11, 2018 by kb1 : Edited for clarity, and, wow: props to Grazza - what impressive chess knowledge, and book career! Awesome! 1 Share this post Link to post
Memfis Posted April 12, 2018 (edited) I think I've improved quite a bit since I started this thread, partially thanks to these nice responses of course. Today I reached my highest rating ever on lichess (1843 rapid) after playing 8 games and somehow losing only 1 (and drawing 1). Though I feel like usually I win because my opponent does some mistake and I manage to exploit well enough, not because I myself come up with something particularly clever. Maybe that's okay for now, but clearly eventually I'll have to study theory in more detail. I often see people doing theoretically bad stuff like moving the same piece too much but I'm rarely confident that I know how to punish them for that. And I don't really have any games I can proudly show because computer analysis always finds some pretty poor decisions. Well, maybe this one was kind of cool, if anything just because of the sexy final position: black is completely backed into a corner while all of my pieces are coordinated, leaving the king with nowhere to run. It was fun to play so aggressively, something I do kinda rarely. Btw I'm terribly afraid of King's Gambit and Fried Liver Attack. :D So you'll rarely see me playing open games as black. Edited April 12, 2018 by Memfis 1 Share this post Link to post
Grazza Posted April 13, 2018 (edited) I wouldn't be too worried about computer analysis tearing your games apart. Even good grandmaster games can look pretty crappy when scattered with the computer's question marks. The best engines are something like 3500 strength, with average GM a bit over 2500, and good club player about 2000. Also, bear in mind that a lot of moves that computer analysis criticizes aren't really errors at all, but just a more human way to handle a position (e.g., a safe simple win rather than a forced mate). In the game you cite, perhaps the main thing to note is the forced win with 10 Nd5! that you missed. You really should pounce on opportunities like that; tactical 'awareness' comes into play here: Black has allowed the guts to be ripped out of his position, so you should be looking for a knockout, in a way that you wouldn't if he had played more sensibly. (As soon as he allowed the powerful move 7 e5, you should be smelling blood.) I do understand though; you see that 10 Nb5 is no killer blow and somehow forget to think about putting the knight on d5. I also noticed that in a game you played a few hours ago you allowed a snap mate in a favourable 2N vs 2N ending. Awareness again: two knights control a lot of squares, so you need to give a little extra thought to king safety. I guess this sounds like being wise after the event, but I saw those last few moves live, and spotted the mate possibility before you walked into it. I appreciate that you probably only had a few seconds left, but it does pay to be a little paranoid about king safety. There are some good active lines for Black against the King's Gambit (such as 2...d5 3 exd5 c6). By the Fried Liver, I presume you mean just 4 Ng5, since you should never come close to the actual Fried Liver, which is 4...d5 5 exd5 Nxd5? 6 Nxf7?!. Black gets plenty of active play in the standard lines with 4...d5 5 exd5 Na5, etc. In both lines mentioned, Black gets lots of activity, and you're unlikely to to lose many games because of White's extra pawn, as long as you play positively. Bear in mind the value of tempi. The initial move makes a difference of about half a pawn (the initial position can be assessed as roughly +.25), and an additional tempo is worth about a quarter of a pawn. You're somewhere near the level where openings start to matter quite a lot, so if in general you like playing the black side of 1 e4 e5, it makes sense to find solutions to individual lines you dislike, rather than fleeing to a different opening entirely, where you are also bound to find variations you dislike sooner or later. 1 Share this post Link to post
kb1 Posted April 13, 2018 (edited) @GrazzaNice evaluation! A question for you: Many people refer to the stages of a game as "Opening", "Middle Game", and "End Game". With your experience and expertise, what's the best way to definitively determine that you've transitioned from midgame to endgame? An interesting fact about computer chess: What's the number one biggest flaw with computer chess? Unless the computer can foresee a definite checkmate, stalemate, or draw, the computer must produce an evaluation (a score) of how good/how bad each move is, and that process is only an educated guess. This guess is based on things like the number and type of pieces still on the board, and the absolute and relative placement of each piece. For example, a queen is worth more than a rook. Many engines also calculate bonuses or penalties for specific setups. In many situations, these bonuses or penalties are appropriate. But, occasionally, these adjustments to the score make no sense. The computer must be able to generate this score very quickly, as it is used to evaluate many thousands, or millions of moves. To speed up evaluation, these specialty adjustments are often oversimplified, like "reward knights sitting in one of the middle 4 squares", with no regards to the rest of the board. I think the hope is that specialty penalties will help balance an inappropriate specialty reward. Often, these simplistic rules are at the very core of why you'll occasionally see silly moves, especially near endgame. Adjusting these rules and their point values is very tricky, and it can make the difference between a good engine and a great engine. 1 Share this post Link to post
Grazza Posted April 13, 2018 Obviously a fair degree of simplification is needed, but the most critical difference is: Middlegame: king safety is paramount Endgame: king activity is more important (though be very careful when queens and rooks are about!) and the primary struggle revolves around promoting pawns Computer evaluation functions are extremely sophisticated nowadays: multi-faceted with a lot of tuning and testing put into them. Check out the Stockfish logs if you want to see further information. But generally speaking their evaluations are pretty reliable, even in position-types where until relatively recently they tended to be clueless. 1 Share this post Link to post
kb1 Posted April 13, 2018 (edited) 1 hour ago, Grazza said: Obviously a fair degree of simplification is needed, but the most critical difference is: Middlegame: king safety is paramount Endgame: king activity is more important (though be very careful when queens and rooks are about!) and the primary struggle revolves around promoting pawns Computer evaluation functions are extremely sophisticated nowadays: multi-faceted with a lot of tuning and testing put into them. Check out the Stockfish logs if you want to see further information. But generally speaking their evaluations are pretty reliable, even in position-types where until relatively recently they tended to be clueless. I just downloaded the logs for further study. From a quick look, yes, it does look a lot more sophisticated than I'm used to seeing. I guess better evaluation is worth the extra eval time. But, in what I saw about hash tables in the logs, they're got a handle on speed as well. I still find it funny, though, how the computer must evaluate hundreds of thousands of positions, whereas a person doesn't evaluate more than a handful, yet us humans give it a run for it's money (and I get crushed...) It may look like thought, but it's really a glorified search at this point. Google is getting into some scary stuff, though. We'll have to wait and see. EDIT: And, I had to grab a copy of the source. Just what I need - another project... Edited April 13, 2018 by kb1 0 Share this post Link to post
Grazza Posted April 13, 2018 It's a balance between depth of search and complexity of evaluation function, for sure. Some engines have simpler evaluation functions in the interest of greater search depth and vice versa. And as you'll see in places in the Stockfish logs, in some cases they have axed or simplified a part of the evaluation function because current search depths do a better job of sorting out the issue in question. One of the main purposes of the evaluation function is to determine which branches of the search tree should be either extended or axed - without that aspect working well, engines would play far more weakly. On my current desktop machine, Stockfish typically analyses about 40 million positions per second - so it's easy to see that this depth might be more important than some fuzzy evaluation that isn't relevant to pruning the tree. Hash tables are pretty important, so you should adjust their size depending on the type of analysis you're doing. For general-purpose interactive analysis I might set them to just 8-16 GB, but for more in-depth "leave it on overnight or while I'm away on a ski trip" type of work, I'll give it the full ~100 GB. A full set of 6-man Syzygy tablebases should also be used, ideally on an SSD. Yes, it is amazing how well humans do with a minimal "search depth". I suppose we have a highly advanced "evaluation function", but the days when human vs computer was in any way entertaining are long gone (though it is still fun to beat the crap out of the lousy programs you find on in-flight entertainment systems). The DeepMind approach is interesting, basically seeking to develop a highly advanced AI-based evaluation function. It is hard to assess empirically how effective it is, since the testing so far has been done with it having a huge processing advantage over its more traditional computer opponent. The fact that it runs on fundamentally different hardware makes a fair comparison difficult, even if they were attempting to make one. It's worth noting the perhaps surprising fact that in the rise in strength of chess engines over the last 20 years, software improvements have been a greater factor than increases in processing power. That is, e.g., 2018 software at 2004 speeds will beat 2004 software at 2018 speeds. (At least I assume so, but the last time I recall a test of that type being made, it wasn't even close. And yes, my choice of 2004 is not random, as Fruit's implementation of LMR came in 2005, changing everything.) If you are seriously interested in chess programming, then this is a good source of info. Just to illustrate what I said about computer analysis being taken with a pinch of salt, I have attached a handful of my own games with automated analysis (zipped PGN file, readable by most chess software). As you can see, it throws around a lot of question marks - some justified, some much less so. Moves that reverse the assessment (or change it from "winning" to "not winning") are important to note and learn from - ones that change it from "completely lost" to "even more completely lost" less so. miscgame.zip 2 Share this post Link to post
Memfis Posted April 13, 2018 That knight mate was very educational! It's true that I was short on time, but more importantly I'm just so used to the king having a lot of space in the endgame that I didn't even entertain the possibility of getting completely surrounded. Actually, such endgames are perhaps some of the trickiest for me because knights move in a strange manner that I find hard to visualize, they always threaten forks, and the way they travel between distant squares can be confusing (the knight's tour problem is a testament to all this). I found this diagram which shows a clear pattern, so I wonder if that's what good players rely on, maybe unconsciously. I'm always impressed by good tactical awareness. There are many videos of grandmasters playing and commenting their games, and often they'll do a couple of powerful looking moves and then say "okay, there should be a mate/win somewhere", even though they haven't calculated much yet. And indeed, soon enough it materializes. It seems so satisfying, like they are in complete control because they have such a deep understanding of common patterns. Makes the game look really logical and enjoyable. It would be great to get closer to their level one day. I think it's true that I should learn to fight some of these "scary" openings instead of completely discarding a big portion of my possible repertoire. I'll definitely look into these suggested lines, thanks. :) 11 hours ago, Grazza said: It's worth noting the perhaps surprising fact that in the rise in strength of chess engines over the last 20 years, software improvements have been a greater factor than increases in processing power. That is, e.g., 2018 software at 2004 speeds will beat 2004 software at 2018 speeds. Wow, that is surprising indeed. I always assumed that it was mostly about the calculation speed. But this certainly explains why chess engine ratings can differ so much. 1 Share this post Link to post