In the post by Antoine Fourrière that George linked to, Antoine talks about the legitimacy of pieces, an issue I have not addressed on this page. There are two issues here. One is that some pieces seem strange and awkward. He complains in particular about Gold and Silver Generals, saying that they 'seem to have originated out of the blue from the brain of a drunk goblin.' The other issue is about including some pieces without including other pieces that seem equally legitimate, such as including the Cannon but not the Vao.

One of my principles is related to the first idea here: 'Favor pieces whose moves are easy to visualize, such as Knights and line riders.' Perhaps the Gold and Silver Generals seem less legitimate to Antoine because their moves are harder to visualize than a Rook or Bishop. Having adapted these two pieces to a hexagonal board, I understand these two pieces in a way that reveals them to be legitimate pieces. I understand the Gold General to be an enhanced Wazir and the Silver General to be an enhanced Ferz. Each is enhanced in the same way by gaining the ability to move in any forward direction. So, the Gold General can move one space orthogonally or one space forward, and the Silver General can move one space diagonally or one space forward. The enhancement to these pieces makes them more useful for attack than for defense, which follows my principle 'Make offense stronger than defense.'

The other idea is one I have followed in some of my games, and Antoine mentions one of them, Eurasian Chess, as an example. I think Eurasian Chess is an excellent game, and it benefits from including Vaos with Cannons. But I don't think as much of its predecessor, Yang Qi, which tried to improve on Xiang Qi by, among other things, including Vaos with Cannons. At the time I made Yang Qi, I didn't fully understand what made Xiang Qi a good game. Also, I have used Cannons in Grand Cavalier Chess without also using Vaos, and I haven't missed them. So I don't think it is always essential to include both the orthogonal and diagonal versions of a piece when you include one. Still, I can understand the intuition behind this idea. It makes a game easier to learn if the pieces come in orthogonal/diagonal pairs. I think this is one of the features that makes Gross Chess easier to learn despite its addition of six new pieces. It also makes a game appear more symmetrical and hence more beautiful. But despite having Rooks but no Bishops, Cannons but no Vaos, Ferzes but no Wazirs, and Elephants but no Dabbabahs, Xiang Qi remains a good game. But this could be because, as Antoine explains, 'Chinese Chess features an interesting opposition between (mainly) orthogonal attackers and diagonal defenders.' He then complains that 'Shako feels strange with its orthogonal Cannons and diagonal (Firz+Alfil)s known as Elephants but not the corresponding Vaos and (Wazir+Dabbabah)s.' I came to this site during the CV design contest that Shako was entered in, and in its favor I will say that it was my favorite game entered into that contest. But I will agree that Eurasian Chess is a better game for pairing Cannons with Vaos instead of Elephants. So, I won't say that pairing pieces in this way is a requirement of good game design, but I will agree that it can improve a game.