Data Base moves versus Book moves

I have noticed that opening moves in data bases are usually referred to as "book moves ". However I have reason to question just which is more reliable in giving me a good opening move ? The book, or the database ? The reason for my scepticism is that in my database after : 1 e4 e5 2 Nf3 Nc6 3 Bb5 a6 one of the best known, and most popular openings in chess has appeared and now my database gives 8 possible moves for white !! One of these moves is that absolutely ridiculous move  4 Bxa6 ??  ( no this is not a typing error and the bishop takes the pawn at a6 !  Whats even more incredible is that there are 2 games with this blunder from white and yet white wins both games !!? After checking the games I note that none of the players involved were rated apparently which makes one ask : what criteria is used for games that go into databases ?  For this, and other reasons I still trust books more than the databases and perhaps you should too ?  To make this even more ridiculous/humorous in one of the 2 games after 4 Bxa6 ?? black replied 4... Nf6 ?? 

Great question and post Reb! The quality of databases vary as many just want to have as many games as possible and this includes all sorts of Internet games (even 1-0) which a high quality database will sort out the ridiculous. Interpreting the data is key to "reading" the statistics and I use both books supplemented by my files. I am guessing pre-moves explain your humorous B:a6

Databases also give the average rating of the player who played it and the average performance rating  (well, obviously not all databases).  The safest bet is to choose moves that have been played in a large number of games.  Once the database moves get down to just a few games unless you know your research it's best to start thinking on your own at that point in my opinion.  Sometimes books don't have all the good moves like MCO, even if they would (or should) have the most solid replies.

I am not an expert in this, but it is my understanding that not every opening that has entered a database can be called as a book move. If an opening has an ECO number, it needs to be in the Encyclopedia of Chess Openings. Thats what ECO stands for. So I can't just play a random opening, put it in a database and call it a book move, just because it is in the database now.

So perhaps we should mark book moves as BM and database moves as DM or DBM ?  Are there databases available that the games in them meet a high criteria for inclusion ? I would like to have one in which there are only games involving titled players on both sides of the board, for example. There could also be an "elite DB " which would only include games where both players are GMs of a certain rating... say 2650 or higher or even 2700 ?  Perhaps such databases already exist and this old country boy just isnt aware of them ?

I believe there are all kind of different databases like databases only includes World Championships, databases only includes CC chess etc... I do not know if they are seperated into different ratings, but you can filter the players rating if you are using one of those database softwares like chessbase. So you can tell computer to search for only games with certain lines played by players in certain rating range. 

Also you can test the book move versus database move thing in chess.com database. If you go to game explorer and make 1.b4 it says polish opening, but if black reply this move with f6 it is not considered as a book move anymore. Although there is one game in the database with 1. b4 f6 2. Bb2 line, the database does not classifies it as a book move after b4. (It's a random line I picked) .  

Reb, what database are you using? If you have chessbase you can filter out all games of players below x level. Also openingmaster makes a database of games that are 2600+ only.

That being said for correspondence chess, I think there is a great opportunity to steer your opponenets into know bad lines where the database favors a move but it was refuted in analysis but never played. Example, I recently played a game that quoted Ivanchuk - Polgar hoogeveen 2009 but at move 17. My opponent uncorked a "rybka" improvement that completely changed the evaluation of the game. Ignoring  that my opponent probably cheated, the point is that there are all sorts of pitfalls in otb games that can be used to snare unsuspecting database followers.  

Well, the inclusion of non book moves in a datavase is a great tool for mediocre players like me.

Books do not normally include bad moves, so it does not normally include their refutations. Databases do.

But, as you have pointed out, it should be used with some care. Historical results are not enough.

I thought databases were useful in exploring theory deviations a lot more.

 When somebody leaves book, I don't assume they were wrong  and certainly don't consider it a novelty (as rare as it might be at the amateur level). I would query the  Master-and-above databases and  use the statistics to form an opinion and decide that from a practical chances standpoint, would that deviation be playable for the kinds of games I normally played OTB (G/30, G/45).

Even the worst looking moves that theory/GMs claim busted can create complications and demand accurate play by opponents (lines that Tim McGrew calls high in "caltrop coefficients") and I think from a practical standpoint,  if I see enough Masters weasel an advantage out of, there may be good chances of playing it at the club level with short time controls.

I always considered database moves and book moves more or less the same thing.  Theory constantly changes based on the newest games, and there isn't really some grand 'book' (except maybe ECO) that dictates opening lines.  They are just moves that are considered best by practice.  A database is far more useful than an opening book for seeing practical results of different lines imo.

 Well, in the Ruy after 3...a6 not a single one of my opening books gives 4 Bxa6 ?? , but my database does. ( Chess Assistant )  Its pretty clear that books are more trustworthy.

4. Bxa6!!

In one move White avoids 400 years of theory, and gets great chances at an early lunch.

As far as "book" goes, it's worth noting that if you reach a position that eco calls += but you don't know how to play it.. might as well be =+. So what I do is as I play games I gradually build my own book. Maybe in ten years I will have something semi-comprehensive.

Clearly book moves and database moves are not the same thing. Practical games don't always follow the book, and often deliberately go "out of the book." They will still of course be in the database.

Personally I don't find databases very useful. For example, you find that a certain line is popular but doesn't score well. Is it that the line is not good or that it is used by weak players ? Or, a line that is not popular scores pretty well. Is it that the line is good, or that because it is not well known it surprises some opponents ? If a line is both popular and scores well then you can probably proceed with a certain amount of confidence, but you still don't really understand the reasons that this is so, and cannot without studying the resulting middlegames and endgames.

That said, it is better than nothing for those of us who don't have a large chess library at home.

Trust but verify is a good motto for database use.

If you use rating filters you will not game any games pre 1960 and only a few in the 60's. Also many databases do not have accurate ratings--the key is where you get your database and high quality you want it to be --for instance Internet games at 1-0 I think add very little value yet many multi-million game databases have them. I also place a premium on annotated games in databases as that reminds me of books.

 In the KGA you have the " Bishop's gambit " which is 3 Bc4  and the  "knights gambit " which is   3 Nf3

Just a check, from some reason this thread is not appearing in my Posted Topics.

 twic, openingmaster, chessbase.. any of those are reasonable. filter as needed. If you are looking for opening theory, I dont think pre 1960 is going to be terribly usefull at the initial level. Once you are leaning towards an idea then you might want to see old master games that might have started the line.