Designer Diary: Roll for the Galaxy

Picture things back in the summer of 2009: Race for the Galaxy, designed by Tom Lehmann with a lot of advice and development assistance from me, is (around) the ninth-ranked game on BGG (before the site split the game rankings into different categories). Development of The Brink of War, the third expansion, is mostly done; Tom and I are having some back-and-forth with the graphic design team, but there isn't much active playtesting happening in RftG-land.

At the 2009 Gathering of Friends, Lucas Hedgren had brought a prototype, a proposed dice version of Race for the Galaxy. Tom and I were asked to evaluate it, and we both agreed that while it was a reasonable game in its own right, it didn't quite capture the feel of Race. This started us both thinking about what would do this. As Tom was busy with other projects, I created a prototype first, in time for the 2010 Gathering of Friends. Graciously, Tom let me pitch "Dice for the Galaxy" to Rio Grande Games and it was accepted!

A name change, hundreds of revisions, and five years later, here we are. Roll for the Galaxy is being released soon. It's a space-themed civilization-building strategy game for 2 to 5 players, with more than 100 custom dice, 55 double-sided tiles, 18 faction and home world tiles, dice cups, player screens, empire mats, and more. And, of course, its gameplay which is, in my humble opinion, the best part of the game.

The First Part Is Always the Origin Story...

I'm often asked what the inspiration behind Roll is. It goes back to the progenitors of Race for the Galaxy. Race is based on an unpublished CCG Duel for the Stars and Tom's "Puerto Rico card game" prototype. That prototype, in turn, was inspired by the board game Puerto Rico (which also inspired San Juan). But there was one thing both Puerto Rico and Tom's prototype had which was in neither San Juan nor Race: "colonists", that is, tokens representing the populace of your space empire, without which its technologies could not function.

It occurred to me that if we put dice in a Race game, then they would be an interesting way of representing populace, given that people don't always want to do what the government — that's you, the player — wants them to do. Governments have a lot of control over their subjects, but it's indirect and slow; you can motivate some of your populace at any time, but certainly, in a modern democracy, it just isn't true that everyone works for the government.

Employment ebbs and flows based on economy and trade, so the core conceit of Roll is that you have a bunch of dice, representing workers, who will do some work for you (but it might not be what you want them to do, especially if employment is high). That work is all about expanding your civilization and your economy. With the money you earn, you then hire more workers from your general populace, the cycle repeats, and players get to roll more and more dice as the game progresses, which is just plain fun.

...Modified for Dramatic Effect

Okay, I have a confession to make. That stuff about the inspiration behind Roll being drawn from history and how dice are the perfect way to represent galactic population? It's all false. It came later. If Tom were the lead designer on Roll, it might be believable. Tom is a regular role-player and uses story as part of game design. He doesn't create a single card in Race without having some idea of its backstory. Me? I'm much more biased towards game mechanisms and tactical interaction.

What actually inspired the game is that I set the following goals:

• The game should have the "look and feel" of Race. Players should be building up galactic empires and player interaction should be indirect, not "attack-y".

• The game should use dice differently from many other games.

• The game should feel simpler and more streamlined than Race.

I thought the second point would be tricky. Dice are often used in games in two ways:

(1) Dice are rolled to resolve a player decision. For example, I decide to attack you, so I roll some "attack dice" (and maybe you roll some "defense dice") and then we look at the dice to see what happens.

(2) A player tries to roll some combination of dice faces. Typically, you get some number of re-rolls, subject to certain rules. Often, there's a "press your luck" element in which you can accept a mediocre result or re-roll, hoping for a better result, but risking a worse result.

I wanted something different. There's a classic post by Allen Doum about "resolution luck" vs. "situation luck" that goes into this in detail (so I won't), but I wanted a game with "situation luck". Instead of the player making choices and then the dice deciding what the outcome is, I wanted the reverse: The dice make the choices, then the player decides what the outcome of those choices are. I also didn't want players re-rolling dice as re-rolls slow down play, as players angst over whether or not to re-roll.

I also had been thinking about economic engine-building. I had just played Alan Moon and Aaron Weissblum's New England and I noticed that games often ended in close scores, even though players were usually choosing quite different strategies. It made me wonder whether there was an underlying mathematical explanation for this — some basic formula that turns resources and opportunities into victory points — and that the good players were squeezing synergies from the game mechanisms to give them the edge needed to win. Race also has some of this, but it is a bit harder to see because there is so much flavor added by the story elements.

Based on these concepts, I came up with the die-rolling and phase mechanisms used in the published game. Players roll their dice once a round. They can then slightly modify what they rolled by using any die to choose a phase, guaranteeing that this phase will occur. (Certain powers also allow players to shift dice to different phases.) When players are done assigning their dice, the player screens are lifted, the chosen phases occur, and all dice assigned to unchosen phases are returned unused to cups until the next round. Roll is very much a game about "making the best lemonade" from what you roll, while also correctly predicting which phases your opponents will choose.

The economic formula is very buried now and I'm not even sure it's accurate for the published game, so I'll let the players see whether they can figure it out for themselves.

That Tile's Mother Wouldn't Even Recognize It

Here are some tiles from my very first prototype:

Here are some tiles from the published game:

Below are some differences between that first prototype and the published version, as well as some details as to how they changed.

In the first prototype, world tiles and development tiles were separate:

• World tiles were in a bag; it cost you one explore die to get a new one to work on. When you settled a world, you received dice, then flipped over the tile so that you could start producing and consuming on it.

• Development tiles had to be built in an order. It cost you multiple explore dice to flip over the tile, then you had to use develop dice to get it into your tableau to be worth points. It was always my intention to add powers to the development tiles, but in the early prototype I made them just worth extra points as I wasn't sure whether my economic model worked and I didn't want to add the complexity of interacting powers right away.

• The big number represented VP value, not cost. The costs were always in dice and represented by the gray dotted outlines. The idea was that you put dice on top of those outlines until the tile was full, then you triggered the icon in the lower-right. (For worlds, the VP value was the same as the cost.)

Most of this changed in August 2009. Tom thought the game had interesting aspects but, without development powers, was boring. Having distinct behavior for developments vs. worlds was a good idea, but it involved a lot of complexity. Once I had a better handle on what sort of development powers I wanted, I decided on a new rule, which is that cost and VP value were always the same. This allowed me to remove all those shaded grey squares and fit the behavior of each development/world on just one face of the tile.

At that point, why not make the game simpler by having all the tiles in the bag? As players could get frustrated if they drew all worlds or all developments, why not save some game material and make each tile double-sided, with a world on one side and a development on the other? So that's exactly what we did.

Nobody Ever Complains That "The Game Started Just When It Was Getting Good"

Early on, players began with one start world, as in Race, but unlike Race that start world wasn't even built! After a lot of playtesting in early 2010, it was clear that the game was taking much longer than Race. While some of this was unfamiliarity, Tom convinced me that the real problem was that the game took too long to get to the interesting bits, with about 15 minutes each game of players getting their engines started. While I was skeptical, Tom convinced me to start players with some existing stuff in their empires.

I thought that giving players multiple starting powers would make the game less accessible. I also worried that creating diverse starting powers would result in some empires being too strong or too weak. In late March 2010 I created the "faction tiles" and the number of possible starts increased from 5 to 81. In retrospect, Tom was totally correct; new players generally don't have a problem with the starting powers — it turns out that newer players can just ignore them and still have a good learning experience — and game length is much shorter now.

How to Paste on a Theme Like You Mean It

I hadn't yet come up with the conceit that dice represented workers; instead dice were just abstract dice. Dice that you didn't roll were in an area called the "Recycle Bin". The next big change happened after the pivotal 2010 GoF.

In this prototype, all powers moved dice around. Phrases like "choose two dice from the Recycle Bin and put them in your cup" abounded. I didn't realize it at the time, but this was a big source of slowdown. Players would think that which dice they chose was an important decision (it often wasn't) and would take a lot of time doing it.

After a long discussion and partial brainstorming session with Jesse McGatha, he came up with an idea I hadn't considered: taking these actions and combining them into a concept called "money" (galactic credits). Now, a player simply buys back (recruits) all his spent dice at once. If he can't afford them all, then he has to choose (once a round) which dice to recruit. Otherwise, he just adds them all to his cup (and decrements his credit total accordingly).

Adding this mechanism went against the general "simplify" guideline of game design. Normally, one shouldn't add more rules to a complex design, especially one that introduces a new layer of abstraction. Tom was even more skeptical than I was. But when my first money-laden prototype showed up in September 2010, we were both convinced that it worked. (I suppose in an eerie way that this isn't too different from the actual invention of money.)

Big Worlds Don't Fly

One remaining balance issue was that building lots of costly worlds was too hard (unless you had just the right development powers). Building a high-cost world tends to tie up a bunch of dice over several rounds (as it is hard to get enough Settle faces in one roll). In Race, high-cost worlds provide either useful powers or lots of VPs; in Roll, where worlds provide dice but no powers and VPs are equal to cost, it was simply more efficient to build several lower-cost worlds for the same VPs and more dice (even though one needed to explore more to acquire these tiles).

In November 2010, Tom took Roll to BGG.CON, verified this problem, and came up with a solution on his flight back, namely that high-cost worlds produce rebates of 1-3 credits when completed. Now, building high-cost worlds still ties up lots of dice, but their overall cost is reduced relative to the VPs they provide.

Here's an example comparing the autumn 2010 prototype to the published version:

There are some other differences worth noting. Having the credit rebate allowed us to do more interesting things with the location of the die gained from settling a world (here it goes into the cup instead of becoming a good on the world), which allows for finer differences between worlds and more flavor. Due to inaccuracies in die-cutting, the final artwork now bleeds off the tile edges; if my original black borders had been kept, they wouldn't always be equally spaced (as shown here).

Dice Are Evolving! It's Super Effective!

My first prototype had colored dice, but the dice faces didn't vary that much. I like symmetry, so initially there were six dice colors. All dice had at least one each of the five phase faces, so only the sixth face varied between dice. Here's a die face chart from March 2010:

By this point, I had added Military dice to represent the Military penchant for expansion while avoiding the details of Race's military rules.

The asterisk face was weaker than in the published game; it always matched the phase that the player chose. I was reluctant to make the distribution much more skewed. One issue is that a "phase strip" behind the shield didn't exist yet. Players had to actually change the die to the desired face using powers. I thought players would get frustrated if the dice they wanted to change didn't have the face they wanted to change it to.

Tom argued that variety in dice was one of the most interesting parts of this game and that the game would work better if the dice colors were more biased and had more "personality". In early 2011, these factors came together for the last major design change in Roll: that a strip would be used to organize dice behind player screens, that wild faces could be assigned to any phase (instead of matching the phase that a player chose), and that re-assigned dice would simply be shifted into the desired column under the phase strip (instead of being turned to the new face).

With these changes, taking greater advantage of other players' expected actions was possible via wild faces and the dice could be differentiated a lot more. Here is the final dice chart:

And You Thought Designing the Game Was the Hard Part

By July 2011, I had been actively playtesting Roll for two years, bringing it to every gaming event I attended. At this point, changes to the game were few and slow in coming. Once a game reaches a certain level of maturity, it is hard to get any solid game-balancing feedback from new players. It's possible that a certain tile combo might be too strong or an individual tile costs too much, but you can figure this out only if you've played the game hundreds of times, and many of those players were starting to succumb to groupthink.

Once the game mechanisms were stable, it was time to test the written rules. That's when a lot of the playtesting switched to blind playtesting. I would present a group of new players with our written rules and observe what they had trouble with. This process is much slower as it gets harder to find players who enjoy playing prototypes, are willing to read rules (instead of having a game explained to them), and haven't already played your prototype.

Jay Tummelson of Rio Grande Games had emphasized to us not to worry about publishing constraints or the number of the dice in the game, but instead to just "get the game right". After turning in the game, much of 2012 and 2013 was spent by RGG figuring out how to produce a game with so many custom dice in it, while I worked with the artists to create new artwork for the completely new tiles and adapt existing artwork for tiles shared between Race and Roll. The year 2014 was spent finalizing the rules, proofing everything, and working out the final production details.

It's been a long five years getting this game to market and I hope that the extra time spent polishing the product shows. In any case, I hope you enjoy playing Roll for the Galaxy.

Wei-Hwa Huang