I was interested in a real-time, luck-pressing game and thought the atmosphere of pushing your luck in a mine to get the most treasure possible before it collapsed was a strong one. This is not the first time I have worked on a real-time game; my first was a collaborative design, Twitch, in the 1990s. While I really enjoy the game play of Twitch, I do have trouble getting people to play because it is so intensely speed oriented. Players who stick with the game can develop strategies that could counter faster players to some degree, but in the end Twitch is a game about reflexes and fast playing.
With Greedy Greedy Goblins, I wanted to see something different; I wanted the game to be less about reflex and more about making game decisions. To this end the core game was designed as follows: Players look at face-down tiles and decide which cave those tiles go in. At any time, a player can lock down a cave as their own by putting one of their pawns there.
At first, the pressing of luck was almost entirely about another player beating you to a cave you have played a lot of treasure on. The real breakthrough element in the design was the dynamite, which made the cave much more valuable with one or two pieces, but penalized a player if there were three or more. (You can be and should be greedy-greedy but not greedy-greedy-greedy.) With the amount of dynamite I put in the game — a lot — you can be sure that if there are a lot of tiles on a cave, it is a dangerous place to send your goblins.
This core gameplay appealed to me because a player who chooses to play slowly and be observant can often make better decisions about which cave to take ownership of than a player who just plays quickly at the expense of strategy or a better global awareness. Some of the best players I have seen make just enough moves to throw monkey wrenches into other people's plans while trying to read the best positions to take based on other people's play.
It was important that during the play portion of the game there were not many options that opened because of what you drew; the core play needed to be (1) look at a tile, (2) put it somewhere. If this were a turn-based game, I would likely have a cup of coffee tile which would make the person who played it put two tiles without looking at them face down in a single cave. In this game, though, I had to restrain myself from these sorts of mechanisms as there is plenty of decision-making simply in deciding where to put something; being forced to process more was just asking for players to inadvertently cheat or lose track of the simple flow of tiles that makes the game fun. One exception to that was the torch – which allows a player to see a tile someone else has played – but even this I made optional since so many players forgot to use the power in the heat of the moment.
Using cards to modify the game is one of my staple design tools. I didn't want the cards to be "played fast" for the same reason I didn't want to add tiles more complex than the torch; the focus of the game play had to remain look at a tile, choose where it goes. For this reason, none of the cards are played quickly; instead they are played when the tile-laying is finished. Even so, they still bring variety to the tile play since, for example, a card which makes rubies more valuable will encourage you try to do something clever with all the rubies you run across.
My favorite moment from playtesting was when my opponent realized I was beating them because I was correctly guessing where they were building their biggest treasure trove and grabbing it before they did. Anticipating that they were going to try the same to me, I put a lot of tiles in one cave and when they hesitated about taking it, fearing a trap, I faked trying to get my goblin there — allowing him to beat me to the dynamite warehouse that the cave had become.