Roadblock is a series of articles where I interview other designers, developpers, and others involved in the industry, to do a deep dive into a specific issue they’ve dealt with in a project. The goal is to add concrete examples to the mass of game design advice out there.
JV: Today I’m sharing with you folks a discussion with Chris Anderson, designer of the Tempus series of solo games and host of The Board Game Workshop podcast. Today, Chris will tell us about an issue he ran into in designing the latest Tempus game, Tempus Infinitum.
Can you give us a bit of background on what the game is like, so readers better understand?
CA: All of the Tempus games are similar. They work like a roll and write, but instead of rolling dice you write down the day and time you start playing and use those numbers to populate the map with resources and obstacles, and also to determine your actions for the game.
Tempus Infinitum specifically has unique maps generated based on a player’s email address.
JV: That random seeding through the date and time is interesting to me. Why not simply go with cards or dice?
CA: The original game design idea that eventually spawned the Tempus system was to make a diceless roll and write. I wanted a game that required minimal materials to play, in this case just a single sheet per player and a pen. The Tempus idea came up as an option for solo play but worked so well I abandoned the multiplayer idea.
Now, the Tempus system is what makes the games stand out from other roll and write games.
JV: The number distribution in dates and times is skewed: no numbers over 3, 2, and 5 for the first digit of date, hour, minute, respectively, for example. How did you consider that in your designs?
CA: That’s the best part of the system. It’s like designing around weighted dice and gives a lot of opportunity that uniformly random dice don’t have. The numbers affect two things in the game: the setup of the map and the player’s actions.
The map is a 10 by 10 square grid. Since we only use day, hour, and minute we have 6 numbers to assign. Each of those is assigned to one column and one row. During setup, each of those columns and rows has an item that will be drawn on the map in the space dictated by the number. So if the first digit of the day is 2, the item in that column will be drawn in the second box down. Zero is always at the end.
Since, like you said, the first digits in day, hour, and minute are limited, this creates empty spaces that won’t get drawn in items. These empty spaces are what make the unique maps.
With 10 columns and 10 rows but only 6 numbers, there are also completely empty ones. The digits always go in order, but that still gives 210 possible arrangements for columns or rows. When you combine the possible columns with the possible rows you have 44,100 possible map layouts.
Each of these maps has a different arrangement of empty spaces. In previous Tempus games with only a single map I would choose the layout I wanted and add static items to the empty spaces. They could be obstacles, enemies, resources, or goals. With Tempus Infinitum, though, we are potentially using all 44,100 layouts. So the static spaces and the column and row write in items are arranged by each map’s random seed from a pool of 37 items with some restrictions. The arrangement of these items combined with the different map layouts is what gives us millions of possible layouts. Not infinite, but Tempus Very Large Number doesn’t sound as nice.
The other thing the numbers control is the player’s actions. Again, the unique aspect allows for some control. Each of the 4 actions is assigned a number range. Make a Road 1-2, Dig a Lake 3-4, Build a building 5-8, and use a building 9 & 0. Based on the statistical occurrence of each number in all possible day/times, each action is approximately equal in potential occurrence. But the first digit can only be 0, 1, 2, or 3 and 3 is pretty rare. So almost two thirds of the time your first action will be building a road. Building a road is the most critical action early on, because it’s the foundation to performing other actions. And again with the first digits for hours and minutes we have a restriction on potential actions that skews towards lower numbers and the more important actions of building a road and digging a lake.
So it’s specifically the missing numbers that create the setup and makes each minute a unique puzzle to solve.
JV: Very cool! So getting back to the whole point of this interview: What is the roadblock you ran into? How did you identify it?
CA: The issue with Tempus Infinitum is the variety of possible maps. In previous Tempus games I had a lot of control of how maps were laid out and could balance difficulty by adjusting the layout. Since Tempus Infinitum has to work with millions of potential layouts it’s much harder to control difficulty.
JV: Why is variability of setup particularly important to the game?
CA: Variability of setup is the unique twist that sets Tempus Imperium apart from the other Tempus games. The system has been refined through previous games, but offering every player a unique map makes it much more personal.
JV: Could you expand on that? Why is that worth the extra balancing issues?
CA: I’ve been working on Tempus games for years now and what continues to surprise me is just how much design space they have. Tempus Quest is 13 unique maps in a campaign, where what you do in previous episodes helps you in the future. It was designing those different maps that made me realize how many possibilities there are for the setup, even if you keep the exact same rules.
Tempus Infinitum is my attempt to use all of that potential in one game. It’s definitely an ambitious design challenge. And, at best, each uniquely generated map is no better than an individually designed map.
But, I think the ability to offer players a unique map that is theirs alone, that they can try and master and share with their friends, adds a level of community and metagame that solo games don’t often have.
JV: You coded a tool to help you for the setup: what did you use for that? What do you see as the pros and cons of the tool?
CA: I’m only an amateur when it comes to coding. I use Processing (https://processing.org) to create the individual maps. It’s based on Java and the main benefit is that I know how to use it. I’m pretty sure it’s not the most efficient method for what I’m doing though.
JV: How do you imagine it working in a final product, or will it just always keep that computer generated aspect?
CA: Currently I have to manually run the program to generate the maps. Ideally the tool would be integrated into a website so players can instantly get their unique map.
JV: So this will always stay a PnP? What pushed you in that direction, versus trying to publish a physical copy with 100 different setups in a block of paper?
CA: Yes, the plan is for it to always be a PnP. Since the main goal is offering each player an individual, unique map, it’s the easiest way to get those to them. When testing is done, I’ll set it up as a “pay what you want”. So if anyone would like to support the work, they can. But the biggest reward for me is having people play my game. So I much rather see a picture of a completed game on Twitter than make a few cents selling a pad of paper.
But, with millions of possible maps, most of which will never be generated for individual players, if people would find some benefit from having a printed pad, I’m not opposed to the idea.
JV: Had you ever encountered a similar problem before? Why was this one different?
CA: A lot of my game designs use combinatorics to generate a variety of states, but this is the first to use that in the design instead of just in the play.
JV: And how do they differ?
CA: When combinatorics is in the play, for example having a card game that works with multi card combos, an individual play will work out a certain way, but the next play will be different. So there can be some forgiveness in edge cases that are less than ideal.
When combinatorics is in the design, the generated game is all there is for that player. So a less than ideal edge case will always be less than ideal for the player that got it.
JV: Can you talk about the process of solving your roadblock? What worked? What didn’t?
CA: I’m still in the process of solving the issue of balance.
In early versions I had a lot of variability which led to a huge variance in difficulty between maps. Some players would have 5 enemies to deal with and others would have 10. As I’ve gotten playtester feedback, I’ve restricted some of the variability so that maps have a similar difficulty. The trouble is getting the difficulty balanced while still having a huge variety of unique maps.
JV: And how do you contain the difficulty behind the scenes? Do you define more closely how much of each thing will appear, or do you go with a point system to evaluate how “positive” a setup is? (or anything else of course)
CA: In the first version, the item in each space was chosen independently, similar to a die roll, so my only control was adjusting the statistical chance of an item being chosen. This caused the huge variance. Now all the items are selected from the same list of 37 items, similar to a deck of cards, so only the arrangement differs between maps.
I’ve also added restrictions on where certain items can appear. Currently the restrictions I have in place limit only certain items to being in the write in spots or static spots. When it places items on the map it will always place an enemy after a farm. This increases the likelihood that enemies will be next to farms to start, increasing the difficulty. But still has potential to leave a space between them depending on the map layout.
A player request was to have a target score scale so players could tell how well they did. So a potential addition is to have a unique scale for each map based on its difficulty. So on an easy map, 150 is a great score. But on a hard map, 60 is great. This would be a great way to make all maps comparable. But it requires being able to accurately judge the difficulty of each setup automatically during creation.
JV: It feels to me like a variable setup is particularly important in a solo game, where you can’t rely on player interaction to make games play differently. What do you think of that statement?
CA: I think it is important for replayability. A solo game with a single setup is fine for a single play, like a puzzle or choose your own adventure. And could potentially get a few more plays if you want to fix mistakes and get a better score. But the interest to play drops off sharply. Variable setup makes each setup its own puzzle to solve while still rewarding skills and strategies learned from previous plays.
JV: Where else do you want to take the Tempus series?
CA: Once I get the map generation figured out for Tempus Infinitum, I’d like to make new ages for it. Currently it’s set in a vaguely medieval age with castles and farms. I can see making slight adjustments to actions and setting, like I did in Tempus Quest, and getting millions of new maps to play.
But my greatest hope for the Tempus system is that other designers will use it. So far I’ve heard of one designer trying to make a game with it, but I’d love to see what others can do.
JV: Is there a game that you think does this variable setup particularly well? One that you could point designers to so they can learn from it?
CA: Friday by Friedemann Friese. It’s a solo game where you work your way through a shuffled deck of benefits and threats. So each game will be a different order but the things you learn from each play about how to deal with threats will inform your future plays.
JV: Well thanks a bunch Chris! I’ve been thinking about trying and making a solo game during the confinement, and I’m sure I’m not alone in that situation: surely this discussion will be helpful to many a designer!
If you are interested in designing your own Tempus game, you can submit it to the Tempus Design Contest by May 4, 2020. https://www.venntikgames.com/contests/tempus-design-contest
You can listen to The Board Game Workshop and get more info on its annual design contest and design day at www.theboardgameworkshop.com
You can contact Chris on Twitter at www.twitter.com/BlueCubeBGs