Popularity of e4

I mostly play e4 myself. However when I look at the stats on game explorer it shows that the next 4 most common moves are 3-4 percent better. It seems like the most logical move. You put a pawn in the center and open 2 diagonals. The numbers don't favor it.

Maybe Game Explorer is a limited database. Other databases show 1. e4 as preferred. Here a list of all 20 opening moves from another database, ranked by their popularity, with the predominant outcome shown:

1. e4 {King's Pawn Opening. #1 pop/20. w 38%. }
1. d4 {Queen's Pawn Opening. #2 pop/20. w 38%.}
1. Nf3 {Zukertort Opening. #3 pop/20. w 37%.}
1. c4 {English Opening. #4 pop/20. w 38%.}
1. g3 {Benko's Opening. #5 pop/20. w 38%.}
1. f4 {Bird's Opening. #6 pop/20. b 40%.}
1. b3 {Nimzovich-Larsen Attack. #7 pop/20. w 36%.}
1. Nc3 {Dunst Opening. #8 pop/20. w 38%.}
1. b4 {Polish Opening. #9 pop/20. b 41%.}
1. e3 {Van't Kruijs Opening. #10 pop/20. b 46%.}
1. d3 {Mieses Opening. #11 pop/20. b 41%.}
1. a3 {Anderssen's Opening. #12 pop/20. b 39%.}
1. g4 {Grob's Attack. #13 pop/20. b 45%.}
1. c3 {Saragossa Opening. #14 pop/20. b 40%.}
1. a4 {Ware Opening. #15 pop/20. n 45%.}
1. h3 {Clemenz Opening. #16 pop/20. b 44%. }
1. h4 {Anti-Borg Opening. #17 pop/20. }
1. Nh3 {Amar Opening. #18 pop/20. b 44%.}
1. f3 {Gedult's Opening. #19 pop/20. b 44%.}
1. Na3 {Durkin's Attack. #20 pop/20. w 42%.}

Chess.com also shows similar wins. However the losses are 30 percent for e4. Others are below 30.

The March 2017 issue of Chess lists the top twenty openings compiled from a list of 1795 January games where both players were rated over 2400 Elo. One can not take position on this list too seriously because it is greatly influenced by how the openings are grouped. For example, all the Retis are grouped together, while English is separated into 1...c5, 1...e5, etc. Nevertheless, for what it is worth, some of the list entries are: 107 Retis, 89 Caro-Kanns, 87 King's Indians, 72 Najdorf Sicilians, 71 Nimzo-Indians, 65 Slavs, 46 1...c5 Englishes, 44 Classical Gruenfelds, 42 Queen's Indians, 40 1...Nf6 Englishes, 39 Taimanov Sicilians, 37 declined Queen's Gambits, 36 Kan Sicilians, 31 1...e6 Englishes, 31 Italians, 31 Semi-Slavs, and 28 Catalans

Good point. Those darned transpositions. Why couldn't chess just stick to a simple tree structure instead of a directed graph structure with cycles?

A few percentage points are meaningless, and the character of the game isn't established until at least a few moves later.

Those stats seem somewhat spurious even if they are correct.  I don't think many people will look at those and decide 1. e3, 1. g4 or 1. a4 is the best way to win a game.

People who play such ridiculous openings (though I don't necessarily mind g4, I like a wing gambit, though usually on the Q-side) are usually so low rated that their opponents are very poor also so it's a total toss-up who wins and draws are nowhere near as common as between much better players playing the more sensible openings.

Larsen used to play 1. b3 quite often when he was the most successful tournament chess player in the world. The stats above are, to be frank, useless. What matters is the net score, the difference between the number of white wins and black wins.