Cleaning Up Deadwood
A Post-Mortem on the Deadwood Re-Design, 2010-2011
James Ernest / April 24, 2011
This year, I made some dramatic improvements to a board game called Deadwood. I thought you might like to read about it.
Deadwood was originally published in 1999. It’s a board game about bit actors on a Western backlot. Players take turns acting, wandering around the backlot, and trying to improve their status.
Deadwood was a product of the “golden age of Cheapass Games,” a period of about three years when I released roughly one game every month. Some of them were pretty good. Deadwood was popular, earning an Origins Award nomination and a spot in the GAMES 100. And I made a couple of expansions in the following years.
Before I played Deadwood in August of 2010, I planned to reprint it more or less as it was. But I had learned a lot in ten years, and I didn’t like what I saw. The game as it was originally published had a little too much “crazy.” So I performed a tune-up, making lots of rules tweaks, applying some game theory, and doing a lot of testing. It got better.
I’m going to talk about a few changes between the first edition to the second. This won’t teach you the whole game, so if you haven’t played the new Deadwood I suggest you do.
Deliver on the Promise.
Players’ biggest complaint with the original Deadwood was that, of the two basic strategies, the boring one was dominant. The choices were “stay a low-value actor and make money a dollar at a time” and “pay money to upgrade, and then make more money because of your higher rank.”
Given the premise of the game, players naturally expect that upgrading is a good thing to do. If your group decides that it’s bad, the game falls apart. This didn’t happen all the time, but in games where players decided that upgrading was stupid, the mechanics failed to deliver on the promise of the story.
I originally figured the scoring would work like a blackjack tournament: if you bet very little, you might get ahead, but you will probably lose to a player who bets a lot. That is, the players on the safe train usually lose to some lucky player on the crazy train. Thus, the crazy train is the place to be.
If everyone believes this and gets on the crazy train, the game works. If all players play conservatively, it doesn’t. And for many groups, it just didn’t work.
In expansions we tried to moderate this, adding some rules that gave advantage to higher-ranked players. But in the new edition, I needed to rebalance the core mechanic. This started with a statement of the goal: “Players should feel good about upgrading.” I took this to mean that the strategy should revolve around when to upgrade, not whether to do it. However, we also agreed that it should not be necessary to reach rank 6 in order to win.
We made a lot of little changes to make this work. We changed the rate at which credits were awarded, changed the prices of upgrading, and changed the final score to include money, credit, and rank. (It originally cared only about money.) These, along with some other changes listed below, made it much less plausible to win the game as a 1.
Rank Never Drops.
One of the original problems with being high-ranked was that it might not last. There were several ways to lose rank. We invented these because we thought they counterbalanced the money-making power of high ranks, and because they seemed to fit the flavor. You could be shown up by a bit player, walk off a role, take a role lower than yours, or even be stuck on the last shot of the day, and all of these caused you to lose a level. Some of these penalties were within your power to avoid, but some really weren’t.
Unfortunately, on analysis, being a 6 wasn’t actually all that much better than being a 1. During the tune-up I built a table of the expected dollar value of roles, something I’d never done in 1999. I found that upgrading to a higher rank was barely worth the cost, even ignoring all the ways you could slide backwards.
And regardless of “flavor,” there is nothing more frustrating that having your rank taken away from you. Even if it were fair and funny, players felt terrible when it happened. So they learned to stay rank 1, from which there was nowhere to fall, and take advantage of the benefits of bottom-feeding.
With the goal of making it more attractive to upgrade, we re-factored the upgrade prices to reflect the actual value of the high ranks, and removed all the ways to lose rank. A side-benefit of this set of changes was that players could now start the game at rank 1. They used to start as rank 2, so they would have something to lose. And starting as rank 1 makes a lot more sense.
Because I was changing printing formats for the new edition, I needed to redesign the board. The original Deadwood backlot had four blank spaces. These cost players extra turns as they walked back and forth to the Casting Office. The purpose of those spaces was to make it interesting to spend your movement time wisely. But as permanently blank spaces, they represented more frustration than challenge. And every player hit them just about exactly the same number of times, so their effects basically canceled out.
In the process of rearranging the board, I took out the blank spaces, and realized something important: they were still there. After every scene wraps, that set becomes a blank space. So getting around on a half-finished board is already challenging, and there is no need to add more blank spaces.
The new board has none of the blank spaces and much better connectivity, and the result is that the game takes fewer turns. As an added bonus, I made the board sections modular, so players can experiment with creating their own crazy layouts.
Not that it’s relevant to the original Deadwood or the remake, but the best example of “waiting sucks” comes from the “Deadwood: On Location” expansion. That game contained a set of six separate boards that were connected by a “bus” mechanic. To ride the bus, you roll three dice. If a die matches one of the three values on the bus stop where you are standing, you can take the bus to that set. If not, you wait another turn. This turns out to be a highly realistic representation of waiting for the bus, but a terrible game mechanic.
Who Starts the Day?
In the original game, players re-seated themselves at the start of each day. The new turn order is based on rank, then credits, then money, or basically “how well they are doing.” The ostensible purpose of this shuffle is to give an advantage to players who have fallen behind, and to “keep things interesting.”
This turned out to be a bad idea for a whole bunch of reasons, not the least of which is that some players have chairs they really like.
Problem #1: We could not decisively prove that going first is actually an advantage. When a player strikes out into unknown territory, he has the first chance to get into a new scene, but he also commits himself to a path sight unseen. Worse yet, he can’t even jump on the high-numbered roles that those behind him are qualified for. So really, if you go first on day 2, you’re just flipping scene cards for players who go after you.
Problem #2: “Catchup features” are terrible. A rule that gives an arbitrary boost to players who are apparently losing is not a solution to the fundamental problem, which is whatever makes you need a catchup feature in the first place.
If players fall behind through poor strategy, then they should play better next time. If players fall behind through random chance, then your game is simply not good, and no post-hoc catchup mechanic will make it better. So even if going first were better than going last, which it isn’t, and even if low rank were worse than high rank, which in the original game it wasn’t, giving an advantage to that player glosses over the core problem that led to that player’s low rank in the first place.
Problem #3: Irrelevant, given the new rules overall, but letting low-ranked players go first on the next day is one more thing that makes it more attractive to keep your rank low.
When we changed this rule, we had a pretty humorous conversation about exactly who should go first on the next day, resulting in the conclusion that it should be the player who has been idle for the longest. Which is to say, the next player in sequence. As obvious as that sounds, when new players get whisked back to the Trailers at the end of day 1, they still inevitably ask “who goes next?” It’s tempting to come up with something clever for that. In this case the design challenge was to avoid being arbitrarily clever.
Too Much Crazy
In its original form, Deadwood was a highly volatile game. This means that results of certain actions could vary wildly based on randomizing factors. The champion of this effect was the 6-Million dollar movie. If you were working on one of these, you had to roll a 6 to advance it. Anything else meant no progress. If the scene had three shots, you might accomplish this in three turns (highly unlikely) or it might take you, oh, basically forever. The average number of rolls in this situation is 18. And even in Deadwood, that is a lot of turns.
In the original, we let players walk off roles, so they wouldn’t feel permanently stuck. But that cost you a level of rank, and players just preferred to avoid those high-volatility roles altogether. We assumed, in time, that a scene would finish more quickly because more people would join and work together. But if no one wants to jump on the big movie, this doesn’t happen.
Before re-making Deadwood, I worked on a game called Lords of Vegas. During that process I learned that players need a good way to manage their volatility. In early versions of Lords of Vegas, players had two basic choices when they owned a vacant lot: they could build a casino, which pays only when a certain color comes up, or they could build nothing, which pays nothing. “Not building” was not really a choice. They had to play the odds, pick a color, and hope that color came up. This meant that they had no viable low-volatility choice.
Upon identifying this problem, we altered the game economy to allow undeveloped properties to pay a little bit of money every turn. Now there is an actual strategic choice: to keep the guaranteed but small income, or to take a chance on a random income stream with a higher average value. This single change greatly improved the game. And it informed the solution in Deadwood.
One could argue that jumping on a 6 million dollar movie is the player’s opt-in to high volatility, and that staying off the movie is the low-volatility strategy. But we didn’t want that to be the decision, because we wanted people to enjoy working on these high-budget films. Instead, we created a new low-volatility path called “rehearsing.”
Rehearsal adds a +1 chip to your future rolls for that job. You skip your turn to do this, so you earn nothing, in exchange for better odds later. If you save up enough +1s, you can’t fail. If a scene has more than one shot, rehearsal usually clears the set faster than rolling. However, this is counterbalanced by the fact that other players can come along and take advantage of your hard work, which means that even with the time advantage it’s not a clearly preferred move.
Rehearsing allows players to take big roles and feel confident that they will not be stuck there forever. It gives them a sense of control in a situation where, previously, the only reasonable choice was to sit tight, turn after turn, trying to roll a 6.
Cleaner is Better
When designing games, it’s always easy to come up with dirty little solutions to all of your mechanical problems. We came up with a lot of them during this re-write, and it was quite a challenge to clean them all up again. Here is an example.
One big debate was over the question of whether higher-ranked actors can take lower-ranked jobs. In the original game, the answer was basically “no.” You were permitted to take a lower job, but it permanently lowered your rank. This was a huge sacrifice and was rarely worth it, except in the endgame where all bets were off because final rank had no value. Basically it sent level-6 players on a tedious search for work, because they wanted only 6’s. And it contributed to why being a 6 was terrible.
When we proposed letting players take jobs of any rank beneath them, with no penalty, the first concern was that lower-ranked actors would be shut out of jobs, when malicious higher-ranked actors took them out of spite. The theoretical time when this happens is during the last few shots of the day. A very mean player who is a “4” takes a job that is beneath him, leaving the 4-spot job as the only available work, which the lower-ranked players can’t take. This happens mostly on off-card roles, if it happens at all, and we basically eradicated this problem by paying a higher bonus to the bigger job.
In the meantime we offered up all kinds of ugly half-measures. We didn’t want a 6 to be forced to look for a 6, because those are hard to find, but we also didn’t like downgrading people just for taking work that was beneath them. We proposed that taking roles beneath you was legal only if there were no roles of your rank on the set. Then, only if qualified players were not around to work that role. Then, a “3” could oust a “4” from a “3” role if the “4” was sitting on it. Then we even flirted with being able to take roles above your rank, if no roles of your rank were available.
It became clear that none of these rules was worth the trouble to remember it. The game itself needed to be tuned to the point where it was simply okay for a high-ranking actor to take any role beneath him. And that’s just a matter of correctly pricing the upgrade.
It’s easy to get caught up in the “necessity” of your intricate solutions and forget that your game is being played by people, not machines. The game designer’s job is to do the math so that the players don’t have to! Don’t expect your players to put up with fiddly rules just because you’re lazy.
Beginning, Middle, and End
Upgrade costs at the Casting Office changed from a combination of dollars and credits to a choice between dollars or credits. There is a linear pricing structure in credits, and a slightly curved pricing structure in dollars. This is actually a pretty major change, and it helps distinguish the beginning, middle, and end of the game.
Because different strategies earn these two currencies at different rates, it made sense to offer players a choice between the two income types: work on cards, and earn more credits, Work off cards, and earn more money. Which are you going for right now? When upgrades had mixed prices, this choice was much less interesting.
Strategy games should have distinct phases of beginning, middle, and end. In the beginning, players jockey for resources and position. In the middle, they run their engines in a steady state. At the end, they cash out, converting their resources into score. Every game is different, of course, but many games suffer from too much middle and not enough (or none) of the other two. If your game is in a steady state for very long, it has too much middle.
In the early game, the “or” rather than “and” in the pricing scheme reinforces the distinction between taking roles on the card (earning credits faster) and roles off the card (earning money faster). It also gives players a meaningful choice in the mid-game, deciding which resource to spend now, and which to save for later. Finally, the 5-point incremental pricing in credits let a late-blooming player rise from a 1 to any rank by spending, effectively, zero points. This price and value relationship tailors the pricing scheme to make sense through the entire game.
“Dice Games Are Like That.”
One of my testers sent me a note about the game, and it got me thinking about my notions on games “like this one.” He had pretty much the same advice for me as he did ten years ago when we were working on the original. Except this time around, I didn’t buy it.
His premise was this: Deadwood is a game with a lot of dice rolling. Bad rolls can screw you, but that’s okay, because it’s a game like that. There’s no avoiding bad rolls in a game like this. I used to agree with that. But now I don’t. Lots of randomness doesn’t necessarily lead to a game with poor control.
Randomness comes in three forms in board games: cosmetic, biased, and functional. (I just made that up.) Cosmetic randomness just changes the scenery, like changing the background from red to blue. It has no bearing on strategy or on who wins. Biased randomness gives clear advantage to a single player, like rolling dice to advance your piece in a racing game.
Functional randomness changes something about the game in such a way that players must respond, but does not automatically favor a single player. It changes strategy, but not who wins. Functional randomness mixes up strategic decisions and makes games re-playable, and this is the kind of randomness you need if you want your game to be fun.
Deadwood clearly has all three of these. But it’s not a foregone conclusion that games with “lots of dice rolling” can’t be strategic. It depends entirely on what you let those dice affect. The new rules give more opportunity for players to react to the randomness and turn it to their advantage.
It’s a Wrap!
I have a lovely notebook filled with session logs, design notes, and rules changes. The final set of tests were for the graphics only, in which I discovered that it would be a good idea if the cards were a different color from the board (front and back) and that sliding a single shot counter along a track is marginally worse than picking three shot counters off the board.
While redesigning Deadwood I was also redesigning the format and business model of Cheapass Games, and I’m pleased to think of it as my new flagship title. If you like it, I hope you’ll play it, tell your friends about it, and send me a donation so I can keep making games like it.