The best first move for White

There is no "best" first move for white and relying on statistics compiled from databases that also include very questionable games is a sure way to be led astray. 

hicetnunc, those are difficult questions to answer, and I will not even try. I've only looked at the data for the first White move, so that's all I can say anything about at this moment. 

NM Reb, I agree that the quality of the games in the database does influence the results. Still, they do tell us something. Like, do play 1. c4, don't play 1. f4. 

It would be interesting to look at the numbers for a million games played between chess engines rated above 3,000, where White is forced to play each and every of the 20 possible opening moves 50,000 times. 

It is highly unlikely that some first moved will not turn out to score better than others.

cigol, if I applied your reasoning, I wouldn't have a bunch of awesome miniatures with 1. f4.  I wouldn't have discovered many neat positions that are dynamically equal - with 1. f4.  I would not have discovered the "complex" that results from 1. f4.

I used to have a teacher on playchess.com, Viktor Gavrikov.  He called the opening a "defensive" setup.  I have to disagree with him - 1. f4 is no more defensive than 1. b3 to me - and he is way better than I am.  But that being said, that is also saying that 1. c4 is defensive.

Both 1. f4 and 1. c4 serve similar purposes - controlling a key square.  For 1. c4, White often seeks to put pressure on d5.  In 1. f4?  Well, it's e5 in this instance.  

Let's look at a couple of diagrams...

Better yet, let's get back to basics - just look at 1. c4 e5 2. g3.

#1 King Safety - with 1. f4, the king is slightly exposed, at the cost of gaining more central influence on e5 by move 1.  This also brings us to the tempo saved in many of these systems - for instance, in the KID, Black has to move his knight to f6 to begin his development, then move his knight again to push the f-pawn, then often the knight comes back to f6!  Yet, White has eliminated his tempo issues with 1. f4.  

Don't like From's Gambit from the White side? Take up the King's Gambit, which is still an awesome opening - I don't care what Fischer said.  He played 1. e4 e5 2. f4 after he said it.  People say a lot of things.

Still don't like my king exposed?  The a6-f1 diagonal is too open?  The simple e3 secures White's position (remember, in the highly-respected Sicilian, White is already at e4 from move 1, so Qb6 ideas are more effective.  

Do some study on 1. b3, and you will find that the later f4 is a very important part of the complex in many instances.  Ask Bagirov what he thinks about using f4 in the NLA.  

So the issue is, is 1. f4 too early?  Depends on what you are trying to achieve with your setup.  For me, not at all.  It creates the exact scenario I am looking for - a move that doesn't necessarily open the board up by move 1, with central control, opportunity for kingside pawnstorms.

Let me give you a typical idea of what I see when I play 1. f4.

This isn't an exact game, just a real quick rough idea of something akin to what I typically see.  What is bad with White?  Please tell me.  Fine, maybe you don't like g4?  There are more positional routes.  But these type of positions - I like them.  The center is closed, and I can hone in on Black's king.

I could keep going, no sense in taking up all the room here with banter.  But don't play 1. f4?  Because of the stats?  How about actually looking at the board and seeing for yourself what works for you?  1. e4 is best according to Bobby Fischer - but not for me. 

There is no best first move for White, because White is not the same person in all games. Playing chess is dependant on skills, and skills are personal features. There is no objective reasoning that may prove a single first move to be universally overwhelming, because it would mean chess is a solved game, which is currently not the case.

Almost any first move [except 1. g4] is equally good for White as with best play the game ends in a draw.

BirdBrain, this is all very interesting, but it doesn't change the numbers. 

Glex, you are right, chess is not (yet) a solved game. That, however, doesn't mean all moves are equally good. Life isn't a solved game either, still it's better (if one wants to stay alone) to cross the street after the truck passed, rather than walk out in front of it. Physics isn't a solved game either, still we can apply the theory of gravity to send people in orbit. 

ponz111, what do you base that on? A gut feeling, or do you have some literature to back up your claim? 

AnthonyCG, according to my calculations, 1. g4 does indeed score poorly, namely 0.73. It's the 2. lowers score. Slightly better is 1. e3 (0.79), and worst is 1. h3 (0.54). However, less than 500 games (of the 1.3 million in the database) were played with each of these moves, so those figures mean very little.

cigoL, if you base your chess on the numbers, remember that most GM's don't say 1. c4 is the best move.  You are about the same rating as me, and you are preaching that 1. c4 is the best move.  Not that 1. c4 is not one of the best moves.  Most strong players say that 1. e4, 1. d4, 1. c4 and 1. Nf3 are all strong responses.  Also, if you only go by numbers, I see that 1. d4, according to www.365chess.com, has a higher win percentage and lower loss percentage, so according to the numbers, 1. d4 is better than 1. c4.  But then again, I don't want to only base my chess on the numbers - it must be based on understanding.  

1. f4, according to strong GM's, is a normal move.  It isn't the best move, but not the worst either.  My point is not to say that 1. f4 is better to, or equal to, 1. c4.  It is a different opening, probably slightly worse honestly than 1. c4.  But there is less theory, so I get the opportunity to play less explored ideas against my opponents.  I just think it is shallow to tell people not to play 1. f4, as it is totally a reasonable response (and it also gives you an awesome response against 1. d4, the Dutch Defense).

BirdBrain, I'm not preaching anything. I'm simply telling what the data in the Game Explorer says. My rating is irrelevant, since I'm talking about statistics, not chess. Anyway, seeing that 1. d4 scores better that 1. c4 at 365Chess tells one thing: my conclusion about 1. c4 being the best first move was too bold. So, thanks to you, I'll retract that. But do notice that both 1. c4 and 1. d4 scores way better than 1. f4 on 365Chess as well. More than 30 % better.

cigoL, it may be true that 1. c4 and 1. d4 score better than 1. f4.  This I know.  I know that 1. c4 and 1. d4 offer better prospects in many ways to 1. f4, but this doesn't mean that 1. f4 doesn't offer perks of its own.  Here is the thing - I get more wins with 1. f4 than with 1. d4 and 1. c4.  I know 1. f4 better.  And - 1. f4, regardless of the percentages, is still a decent move.  

It is not important to only lean on statistics.  Many chess players fall into this trap - they lean on the "best" move, instead of playing a move that they understand.  Also, the stats for 1. f4 on 365chess.com are largely biased.  I will admit that in most cases, White loses more than he wins in those stats.  However, I find that at my level, I do well with 1. f4.  I have never leaned solely on stats, but I seek to understand the position.

Where we agree is that 1. c4 is practically a better move than 1. f4.  Where I diagree with you is telling people not to play 1. f4 based on the stats.  This shows me that you know absolutely nothing about 1. f4.  

Please, without telling me stats, give me reasons why I should not play 1. f4.  I would like to entertain a friendly discussion about this, since I am very passionate about that opening.  I helped start a very strong group on this website, Bird's Opening Lovers, which focuses around 1. f4.  The group also has many members in the 1. b3 NLA group, so they basically play often into the 1. f4/b3 - 1. b3/f4 complex, the Bird-Larsen Attack.  

My only thing is if you are going to tell people not to play 1. f4, at least give a reason besides stats.  I'll give you one reason to be aware of when you play 1. f4 - king safety in the opening.  However, I have played 1. f4 enough to have a pretty good idea of how to handle keeping the king safe, while developing within the ideas of 1. f4.  

2. Less Central Control in comparison to 1. d4 or 1. e4 - Well, this is true, but also in the case of 1. c4 e5, then White cannot immediately achieve the c4-d4 pawn center.  However, that may play right into what he wants.  The same goes for 1. f4, in the case of 1. f4 d5.  White prepares the e4 push (just like in many King's Indian setups) and can often emerge out of the opening with at least a dynamically equal game, with winning chances.  

Okay, I offered two platforms to discuss "weaknesses" in the opening - why don't you discuss these areas, or at least offer some other weaknesses of 1. f4 that we can see how terrible of an idea it really is?  ;-)

BirdBrain, I like that you want a friendly discussion. However, I don't think I have much to say about this, at this point in time (being rather new to chess). Ask me again in 3 years or so. ;) 

How your score with 1. f4 at your level is irrelevant to me. I can play 1. f4 against my girlfriend - starting without my Queen - and still win. This doesn't "prove" that 1. f4 is a good first move (or that starting without the Queen is favorable). But, honestly, I'm not interested in what works at our level, if it doesn't work at GM level. Why? Because, playing something that works at our level, and getting to know it well, will not help us much at the GM level (if we ever reach it), if what we have spend our time on is giving our opponent a slight edge at the higher levels. 

So, let me finish my part of this discussion by saying: do play 1. f4 all you want, then I'll stick to what I think is worthwhile :)

cigoL, I do find it funny that you have "the best move for White", and you are willing to tell people "don't play 1. f4", but then you have no real good reason for why you say that, other than the stats.  Please, do me a favor - don't tell somebody something is bad, unless you have good reason for it.  At least say "GMs don't play it - however, as many say, it may not be the best move, but it is a normal move."  According to stats, 1...f5 is the 4th most popular response to 1. d4, and it has the highest win percentage of the top 4 (1...d5, 1...Nf6, 1...e6, 1...f5). It also has the lowest percentage of draws of the four - this means that it is the most bloodthirsty attempt for a win of the 4.  

Needless to say, I see you are not interested in deeper discussion of why 1. f4 is so bad, so I will leave it at that.  You should play what works for you - that is why I play 1. f4.  In spite of even GMs giving me reasons why they prefer not to play it, it still works for me - and quite well.  Maybe you should pick up Taylor's book on Bird's Opening some time for a quick read.  He says that many GMs don't understand the Bird - that was why he was able to claim a few GM scalps!

ChessDB has a feature in the tree window that computes the reliability, the p-value, that one move is better than another in that position relative to that database.  Not surprisingly, if the database doesn't have a ton of games for each move, then the difference in score is meaningless.  Also true if the db has a lot of 'low-quality' games ie non-GMs.

In my experience the only time I see a really stupid opening like Na3 or the latvian is if it's a much higher rated player who's screwing around.  My only win online against a master is when he played the latvian gambit.  3.Nxe5 turned out pretty good.

In any case 1...d6 is not a better response to 1.d4 than 1...Nf6 and 1.Na3 is way worse than 1.e4, and 1.f4 is much worse than 1.e4 (though not as bad as Na3).  If you look at the database you'll probably see that e4 has tens of thousands of games while Na3 has maybe a few hundred.

Bobby, I disagree that 1. f4 is much worse than 1. e4.  It may not be as good objectively, but in no way is it much worse.  At the best, it gives Black a more free development, because typically White has to spend a few moves to get two pawns in the center.  But for the most part, 1. f4 players can easily achieve the type of game they seek.  

The only game a player should seek is a winning game. The only criteria for which move to play should be: is this the move that gives me the best chances of winning? If yes, then play this move. If no, then don't. Ones choice should be based on this judgment alone, and shouldn't be influences by any kind of "pet bias".

cigoL, that also goes for what type of player you are.  Some people are terrible at 1. e4, and some are terrible with 1. d4.  So, "pet bias" is extremely important, when you are finding what type of player you are.  For me, tactics is not my strongest point - I prefer a more closed structure.  This is why I don't thrive in 1. e4.  I could play 1. d4, but I like to play on the kingside early.  So 1. f4 does give me excellent chances of winning, per my style.  

I wouldn't call it a "pet bias", if it's actually your best weapon, in accordance with your style of play. Naturally, ones openings have to fit ones style.

The truck example you gave, although interesting, doesn't have much connection with the case. It doesn't represent the situation correctly. I believe, if we have to make a parallel with chess language, that Black won't be able to trample White with his very first replay, and White needn't fear that because it is him that has the initiative actually. If someone needs to be afraid of reacting inaproppriately, that's Black.

Your physics example shows that increasing knowledge leads to creating and using new opportunities. If we have to apply it to chess, it will actually support the thesis that there is no ultimate first move, but rather a variety of opportuities that are discovered with studying a certain matter. This can also be proven with chess opening theory development that has progressed significantly since its beginning, and will progress much further, as other areas of human interest.

To find out what's the best first move for White, such a move needs to exist. The term "best" is an evaluation, and as all evaluations it is dependant on criteria. Without having criteria to evaluate a first move, we can't and won't find out which one is the best, even if it exists. I don't mean having opening principles, because they are not comparative. I mean a system that will result in an objective evaluation that can be compared with other first moves' evaluations. Some people like to rely on statistics, but as far as chess is a game between individuals, statistics can't determine the outcome of the game. The only exeption can be when one uses his own performance statistics with certain openings against a certain opponent. But it's hard to claim other people perform the same way (since we assumed player-specific statistics are used).

So, until there is an objective (not player-dependant) evaluation method for White's possible first moves, chess players will have to rely on their own skills, knowledge, experience and preferences to find out which one is best for them. And as long as those factors remain player specific, so will remain White's first move choices, and inductively all chess games' move sequences.

Glex, I appreciate your view. But I do not fully agree. Let me explain why. 

imagine a chess engine playing 99 % correct chess, playing against another chess engine playing 90 % (or 98 %) correct chess. The "99 % engine" would win the majority of the games, if they played, let's say 1,000,000 games. I'm sure you agree. Now, why is that? Well, it's simple, really. It's because the "99 % engine" more often chooses the best possible move. The difference between the 2 engines might be small, but essentially it's this slight difference, this accumulated positive imbalance that in the end leads to a win for the better engine. So far, so good.

Now, imagine that our two engines play 2,000,000 games against each other. 1,000,000 games playing White, and 1,000,000 games playing Black. Also, imagine that both engines were forced to play each and every of the 20 possible first move 50,000 times each (for a total of 1,000,000 games). Now, do you seriously think that the "99 % engine" would have the same win rate with each of the 20 opening moves? I strongly doubt it! Which first move the "99 % engine" scores the most with, I do not dare to guess. However, I would be very surprised, if some first moved didn't score better than others. Wouldn't you agree?

I agree with the first part. But it doesn't include any any reasonings about the first move.

Why the second experiment, although testing different openings, is not informative about move quality: because of the subjective move evaluation criteria. Flaws need to have been implemented in the second computer's play, and if we're searching for the best move, we should use the same move choice criteria for both sides (as I wrote, it shouldn't be player-dependant).

Anyway, even if the computers are of the same strength, a similar experiment will be informative only about their own style of play, and no further. You can't claim a first move to be optimal only because someone specific (be it a computer, a grandmaster, a chess theoretician or investigator, or simply someone who you admire) has better success with it than with his other opening moves. That's because as a soccer match is played 90 minutes, a chess game is played for all the moves until the end. All moves are important, and also there must be internal logical connection between them, something that is implemented by personal playing style.

If dullards knew half as much as they think they know , they wouldnt be dullards .