Archive for Jonathan Chaffer
Grand Rapids Area Boardgamers
I keep a journal handy with me at all times. One section is full of mechanisms I like, which I want to fiddle with and which sometimes result in full-fledged games. Another is full of "feelings" I want designs to evoke. This part is esoteric but occasionally very useful. A third is a collection of puns and strange phrases that come to mind. Some of those become game titles, or card names, or just spark other ideas.
This is the story of a stolen mechanism, a warm and friendly feeling, and a horrible pun.
The "design seed" for Filler came from an ironic source. One of my favorite Friedemann Friese games — right up there with Foppen and Fiji — is Copycat, his homage to seminal modern board games that overtly co-opts elements from Dominion, Agricola, and Through the Ages in particular. These mechanisms interlock in puzzling ways, and I'm always happy thinking through the ramifications.
One of the most intriguing facets of Copycat is a mechanism that I don't think was lifted directly from any other game — please correct me if I'm mistaken! — which is the way turn order is determined. Each round starts with a bid, of sorts, for turn order, with players setting aside one of the cards from their hand. The more powerful the card, the earlier the player's turn.
However, these cards can't be used for their normal effect, so this becomes a fraught decision. How important is it to go early in the round? Other games have methods of bidding for turn order, of course (Fresco springs first to mind), but this card-based system seemed particularly interesting to me.
Bidding for turn order in Copycat
From a mechanical perspective, Filler began with the thought experiment of how to capture this decision point in a much smaller game. I gravitate toward "light but tight" games, with a few rules and a lot of emergent behaviors, like No Thanks or Linko. The obvious approach was to pare back the mechanism to a deck-building game, but I felt this territory was too well-trodden for this twist to be interesting enough. After a few false starts, it became clear that the right fit was not deck-building, but hand-building. By making this shift, players formed a stronger relationship to their cards since they saw them much more often. This meant the decision to use a card for its early start time was a bit harder to make as you otherwise would be planning on using the resources generated by the card.
In addition to the turn order mechanism from Copycat, in Filler you'll see bits and pieces harvested from a number of hand-building games, such as Concordia and Lewis & Clark, as well as the card flow of something like Mission: Red Planet, but stripped down to the basics.
Oh, and lest we forget, Century: Spice Road, which had not yet been announced when I started working on this title. It's of course another hand-building game that fits in a short time frame, and it's excellent. Filler occupies a somewhat different niche, though, since rather than focus on a Bazaar-like resource trading mechanism, it goes all-in on the hand-building itself, which means it is shorter and in a smaller package, with more of the decisions around the turn order mechanism than on creating card combos.
Keen eyes will also uncover a mathematical detail lifted from a classic Knizia title. Can you find it?
On Your Marks, Get Set, Bake!
A mechanism does not a game make, however. A great theme can do a lot to increase enjoyment by drawing in players, providing metaphors for easier teaching, and inspiring beautiful artwork to savor.
As I was staring at this fledgling deck-building or maybe hand-building game, I wasn't happy with the standard process of cards granting currency that was then spent on other cards. It was important to the design that cards be situationally good, rather than universally good; having some cards grant three coins rather than two coins was just not interesting enough.
Earlier I mentioned the puns, themes, and feelings I collect for later use. The solution presented itself from that list of themes: "a filler game about filling pastries". (Even better, this twenty-minute filler game called Filler would go on to be published by the creator of the podcast 20 Minutes of Filler, Jason Kotarski of Green Couch Games.) Now I knew what would make the "currency ladder" more interesting: The currencies would be ingredients in a recipe, and purchasing a card would require specific ingredients, not just "enough".
A delicious dobostorte to make Mary Berry jealous
As a funny bit of synchronicity, this shift in direction happened a couple of months before my household became obsessed with The Great British Bake Off. Most people who know me assumed the inspiration came from there, and while that's not the case, TGBBO certainly shaped the later development direction of the game! I was so taken with the gentle, warm, loving feeling of this show that I knew that's what the game needed to evoke —not just competition, but good-spirited, friendly competition. I think Claire Donaldson's art style captures this warmth perfectly and makes the whole game experience feel like "home"...or at least like a friendly tent in an English country garden!
In a meeting of our local design group, GRUBS, my compatriot Gray asked whether I had thought about solo rules for Filler. To be honest, I hadn't. The core of the game is the jockeying for start times, and this didn't intuitively make any sense to me in a solo context.
This did prompt me to start thinking about it, though. I imagined something along the lines of Sagrada's solo mode, in which the player's choices would determine their own score but also influence the score to beat. In my first draft of this mode, you would choose a card from your hand and compare the time on it to cards in the middle of the table. Cards with a time earlier than the one you chose would be hoovered up into a pile, and the points there would be compared with yours. This didn't work for a long litany of reasons.
Before our next meeting, Gray had been busy and came up with a system for playing against a virtual opponent. He worked out the basic interfaces with other opponents, patterning this after the Automa system. (See episode 154 of the Ludology podcast for more on this concept.) Since the main interface here is bidding on start times, this was the primary thing to model. We would reveal cards from the deck each turn to determine the other chef's start time, and that chef would claim cards automatically based on their start times.
Sample recipe cards
I liked the basic idea but was worried gameplay would get too stale with the randomness of the opponent, or worse, that a player's choices would get too obvious. The next step was inspired by the fantastic solo mode in Mint Works in which each play is against a different AI opponent. I set us the challenge of coming up with six different AIs, each of which would be tied to one of the chefs in the game. That way, every player could have a number of experiences in the solo game that would feel different from one another, while sharing the basic rules.
Within a couple of days, Gray had a slew of personalities to test. A few dozen playtests later, and we scrapped some of these, tweaked others, and came up with new ones, landing on a set of fiendish rival chefs to battle.
The Tyranny of Time
The most difficult aspect of Filler to balance has been the central mechanism itself: the turn-order bidding. Since this is the source of tension in the game, it needs to lead to a meaningful choice, so turn order needs to be a powerful thing. At the same time, since early cards stay in a player's deck once obtained, it needs to be balanced against the potential of a runaway leader.
A starting time and ingredients card, ready to go
The costing of cards has changed a lot over the course of development. I mentioned earlier that first drafts used a single currency for card purchases, which was uninteresting. The multiple currencies helped with this, but added an interesting wrinkle. There was suddenly the possibility of a "dead end" wherein players couldn't purchase a card because they simply did not have the right ingredients and couldn't get them.
Addressing this led to two outcomes. First, starting player hands were adjusted so that every player has exactly one of each of the five ingredients available to them. (Early drafts included a wild among these, but that was too powerful a card with which to start the game.) Second, the costs of cards now never repeat an ingredient, so that some player will be able to fill the recipe with their starting hand. Third, the number of ingredients required was capped at three (and that is in extreme cases) for the same reason.
These requirements set boundaries on ways to address the balance of early cards. I now knew the most I could make any card cost was one common and two rare ingredients, so the earliest card got this cost and requirements scaled down from there. However, this wasn't enough to counteract the power of early turn order. The solution was to further increase the cost, but how?
The answer here was adding "any ingredient" as a potential requirement. The flexibility of this allows recipes that include it to have four requirements rather than three. While it's generally easier to come up with a few general ingredients than one specific one, the important part is tempo. The cost to obtain a card with an early time near the start of the game should almost always mean that a restock turn follows, allowing other players time to react and counteract that move. Doing this, along with putting more points on later cards, solved the problem as statistics began to bear out.
Ready to play...
But even though I saw from the numbers that the issue was mostly resolved, there was still a difficulty with perception. Although I knew it wasn't always an optimal play to pick early cards, I saw that players felt like it was and began to believe that the earliest card would win the day. To balance this perception further, the advanced cards introduced an "arrive early" mechanism. These cards let you adjust your time card by an hour, so the player with the earliest card in their deck is no longer guaranteed to go first when they play it. Doing this requires you to play two cards instead of one for time, so there is a tempo tradeoff here as well.
(Side note about Arrive Early: When this was introduced, I wanted to preserve the uniqueness of times so that tiebreakers would not be needed for turn order. The solution for this was to adjust the times so that all cards with even hours have even minutes and odd have odd, so subtracting an hour from a card will never leave you with a time that collides with another card.)
After all this balancing work, it was a real relief to read this in Jonathan Liu's review of Filler:
I did have some concerns after my initial experience that the player who grabbed the earliest starting time would have an unfair advantage, but after repeated plays that hasn't proven to be the case. Going after the cards with early start times can help you in certain rounds and may even allow you to fill more recipes, but victory depends more on which recipes you choose rather than just sheer quantity. There have been some games where a player managed to get an early start time and dominated the game, and there have been others where the player with the earliest start time was in last place by the end.
I hope this design work makes your experience feel just as varied!
Grand Rapids Area Boardgamers
Some things are far more obvious in retrospect. Such is the design of Stroop, a speedy perception card game with simple rules.
I can't remember exactly when I first saw a Stroop test. I feel like it was the kind of thing I would have encountered in a GAMES Magazine issue pilfered from my mom's bedside stand. I do recall being delighted by the idea and making flashcards with markers and note cards to test myself and my friends. Throughout the following years I saw it referenced time and again, most notably in the briefly-popular Brain Age game for the Nintendo DS.
The Stroop test is simple. A subject is presented with a series of words, each printed in a different color. The subject is then asked to quickly speak aloud the names of the colors in which the words are printed, and they are timed during this task.
Next, the subject repeats this task, but this time the words are the names of colors, instead of being random words.
This causes the subject to stumble and take longer to complete the task. The experiment demonstrates the Stroop effect, named after psychologist John Ridley Stroop, and shows that the interference between different systems in the brain — in this case, language and color recognition — can slow down both systems.
From Experiment to Game
Flash forward to 2013 when I was in the midst of brainstorming ideas for new tabletop games to develop, and I randomly stumbled across the Wikipedia page for the Stroop effect. This led to the immediate question: Could this be a game?
Now Brain Age had used the Stroop effect in its simplest and most obvious form. It was not really a game, but rather an activity at which you could improve; you were timed on how quickly you responded to the colors and given a score that you tried to improve upon the next time.
The clear way to transform this into a card game was to do exactly the same, but with multiple players. I would print the names of colors onto cards, with ink colors that didn't match. Then players would run through the deck like flashcards, saying the colors out loud and being timed on their effort.
Even before physically prototyping this, it was instantly an unsatisfying implementation. For one thing, a speed contest such as this is usually uninteresting. It is a solo experience that people happen to compare their efforts on, which is something I can enjoy at times but rarely gravitate toward.
But the bigger problem is that the Stroop test can be defeated. Once a subject knows what they are being asked to do, they can use techniques, like squinting, that make the words harder to read, which makes saying the names of the colors much easier. I certainly didn't want players to be able to circumvent the challenge in this way, or worse, to have to make rules against squinting!
Chain, Chain, Chain
The key to cracking this problem was, as is usually the case in design, to come up with the correct incentives for the behaviors I wanted. Need players to read the text and not just squint at blurry colors? Don't make a rule telling them to do read it. Instead, force them to use the text for something.
What purpose could reading the color name serve, then? The clear choice was to link the name of this color up to the color of another word. This forms a nice chain of words, each of which describes the next one.
For the first Stroop prototype, I lifted wholesale the rules of 7 Ate 9, a speed game involving simple arithmetic. Players would race to get rid of their cards by playing onto a central pile, and legal plays consisted of any card that was described by the center one.
In broad strokes, this worked as a game, but it had some issues. The biggest one was the number of potential legal plays on a given card; with eight colors, as in my first prototype, one in eight cards are legal to play. This turned out to be far too small. Iteration revealed that anything smaller than about one in five cards being legal made the game grind to a halt. Shrinking the color space this much, though, made the deck homogeneous and uninteresting.
The solution to this was to introduce additional axes for card descriptors. Aside from color, what else could be used to describe these words? The original Stroop psychology experiments included some other ideas, such as the position of words, but these did not tend to lend themselves to card designs. Instead, I experimented with typography and decided I could easily distinguish the case of a word and could outline it or not.
This was an improvement, but one more axis was needed to flesh out the deck. Some brainstorming surfaced the idea of counting the letters in the words themselves. To make this work, I needed to finesse my color choices, and settled on the following word list:
RED • BLUE • GREEN • YELLOW
BIG • LITTLE
HOLLOW • SOLID
THREE • FOUR • FIVE • SIX
The final list had the very nice property that each word has 3, 4, 5, or 6 letters, and there are three words of each of those lengths. This meant that, at minimum, one in four cards were legal plays on a given center card.
The Round 2 Head Trip
The 2013 prototype was workable, but the game came into its own in the run-up to Protospiel Michigan in 2014. As my testers began getting very fast with the existing rules, I began looking for ways to provide variants and new challenges. The winner was to reverse the legal play rule: Instead of playing a card that is described by the card in the middle, players now had to read the words on the cards in their hands, and play one that describes the middle card.
The fun of the game is in players getting confused, and how better to confuse people than to switch up the rules midstream? The variant became codified as round two: After the first round is over, scores are recorded and players began anew, with the altered rule for legal plays.
The rules were then simplified to reduce the need for a scoring mechanism. Instead of keeping score, I realized that performance in round one could be used to handicap round two. After round one, players keep their unplayed cards, and the played cards are redistributed evenly, so the better a player performs in round one, the fewer cards they have to get rid of in round two. This neatly determines an overall winner without the need for scorekeeping.
The possible combinations of attributes could yield a total of 192 cards: 12 words x 4 colors x 2 sizes x 2 patterns. This deck was clearly overkill, so for my working prototype I used half of these combinations, chosen so that exactly half of the cards were big, exactly half solid, exactly one-fourth red, and so on.
I went to some lengths to retain this balance throughout development. When green letters turned out to be difficult to distinguish from blue and yellow in some lighting (and as I endeavored to serve colorblind players as well as possible), I moved to black letters mostly because "black" and "green" both have five letters.
My insistence on a balanced subset of cards turned out to be a bit superstitious; once the card distribution was defined, it could be altered a bit from perfect symmetry without anyone noticing. The final deck has 65 cards, enough for a four-player game, and is slightly uneven without an effect on gameplay.
One improvement in the composition was removing as many "self-describers" as possible. It turned out that players had a reduced challenge in dealing with cards that happened to describe themselves, e.g., a blue card that reads "blue". The final deck has no cards of this type, with the notable exception of the word "four" which inherently describes itself. Now the "run of fours" that can happen in a game just gives a bit of fun texture to the proceedings.
Experiments Along Further Axes
The twelve-word list is enough for most players for quite some time, but I also put some effort into ideas for further expansion to keep the game fresh for as long as possible. Heather Newton gets credit for the seed of the idea for the expansion included in the game box, which features cards with backwards text:
Some other experiments have proved less successful, but fun nonetheless. Never will a typographer squirm so much as if you show them the following card:
And now Stroop is in print! The journey isn't over for me, though, as I'm actively working on variant rules for less stressful games and figuring out what it means to translate this game into a foreign language when word lengths are such an integral aspect of play.
I hope you'll enjoy this tiny brain-twister of a game!