August 10, 2016

Development Blog


Press Release: Brink of War



Race for the Galaxy Expansion Launches

One of this year’s top selling boardgame apps launches
Brink of War expansion


SAN MATEO, CA – DECEMBER 14 2017 – Temple Gates Games in collaboration with Rio Grande Games brings a new expansion to one of this year’s top selling digital boardgames, just in time for the holidays.

The Brink of War adds 36 new cards, 4 new start worlds, and 5 new goals. Prestige adds a new way to earn victory and gain the advantage against opponents. A new action, Search, lets players find particular types of cards to compliment their galactic strategies. This third expansion is now available on iOS, Android, and PC, priced at $3.99

As conflicts spread, galactic prestige becomes all-important, both to extract concessions and to fend off attackers. Meanwhile, the Alien departure point is located and the Uplift Overseers arise. Can you build the most prosperous and powerful space empire in a galaxy on the brink of total war?

Designed Tom Lehmann, with a neural network powered AI crafted by Keldon Jones, this expansion challenges players create the most prestigious empire.

Beyond the expansion, all Race for the Galaxy players will enjoy loads of new content in the latest Race update, including better timers, more language support, and a revamped tutorial, making this a great time to put the game in front of new players.


▪New prestige mechanic

▪ New in-game Search

▪ 4 new start worlds

▪ 36 new game cards

▪ 5 new goals



For more information on Race for the Galaxy, visit


Race for the Galaxy AI

From TD-Gammon to Race for the Galaxy
Temporal Difference Learning for Boardgame AI

What makes a game replayable over time?  It offers new challenges over and over again.  One way to do that is to include an AI opponent that is so skilled, even advanced players will continue to be challenged after hundreds of hours of play.  Race has been one of the top selling boardgames this year partly because of the neural network that powers its AI.  Race for the Galaxy uses a temporal difference neural network.  This knowledge-free system requires no human input to generate training data, which makes it efficient for a small team with limited resources. Instead, it learns by playing randomly, making predictions at the turn level on which player is winning, and updating the weights in its multilayer perceptron architecture such that the delta between predictions from one turn to the next is diminished. Using this method on over 30,000 training games, it’s learned the black box function that best represents the relationship between input (the state of the game) and output (prediction of who’s winning) for the neural network that drives our AI.  It’s a pretty incredible method, first used for boardgames by Gerald Tesauro who created TD Gammon and now applied to Race for the Galaxy by our AI expert, Keldon Jones.

Heuristic AI     At Temple Gates, we started making boardgame AIs with Ascension.  In this game, we used heuristic driven AI.  This means the AI favors particular moves that we deem superior.  For example it buys cards with a high VP to cost ratio or it trashes low value militias before other cards.  These heuristics are defined by human experts.  The AI moves are based on what we think are the best plays.  And it’s pretty good.  But an expert player will eventually win frequently against this AI.     

Keldon Client     As any devoted boardgame junkie would, sometimes I Google my favorite tabletop games to see if anyone has digitized them.  A few years ago I Googled my favorite game, Race, to see if a digital version existed and I found one!  It was an unlicensed project developed by Keldon Jones. I clocked hundreds of games in it, and wanted just one feature: a sound to alert the player that it’s their turn.  Fortunately it was open source, so I could start tinkering under the hood, and there was some neat stuff happening there.  This client had a world class AI.  I thought about all the other features I wanted and feature creep city, eventually I wanted it on my phone.  


Licensing    Back to Google, I looked up Race’s designer and got in touch with Tom Lehmann to see if that was something I could make happen.  Tom had one request, though.  He wanted to make sure Keldon could be involved in the project.  Apparently he had a pretty neat AI!  

TD Gammon     This AI was based on the original research for TD Gammon by Gerald Tesauro.  Yep, one of the most cutting edge game AIs ever created is based on a 5000 year old game: Backgammon.  Tesauro pioneered a neural network driven AI that ultimately changed conventional wisdom on how Backgammon should be played.  This AI successfully developed on its own unorthodox strategies that were eventually adopted at the highest tournament levels and are now widely accepted by players around the world.  For example slotting vs. splitting.  Traditionally, players whose opening roll includes a 1, use the 1 to move from position 6 to position 5.  This is a risky move with a high reward, if the opponent doesn’t bounce you out, they’ll have a much harder time escaping the 1 position later.   But TD Gammon, based on a neural network, found higher success rates over thousands of games by “splitting” rather than “slotting”, and moving one checker off of 24 to 23.


It doesn’t exactly matter why it’s a better strategy.  It simply has a higher win rate.  Both the predictions and the outcomes favored splitting.  


How did this AI discover the superiority of a strategy that humans hadn’t in 5000 years?  The AI starts with zero initial knowledge.  So unlike our heuristic driven AI in Ascension, it doesn’t come pre-populated with imperatives based on human expertise, in essence, it’s teacherless.  And that’s good, because human expertise can be fallible..

Formula     TD Gammon uses a neural network to determine its moves.  It yielded a formula for changing the NN weights every turn, to reduce the difference between the current and previous turn’s board position.  This is temporal difference learning.  


Untitled-1 is the amount to change a weight from it’s value on the previous turn.  Untitled-2 is a learning parameter.  Untitled-32 is the difference between the current and previous turn board evaluations.  Untitled-4 is the rate of decay in back propagating older estimates and Untitled-5 is how much changing the weight affects the output.  We use a similar formula in Race for the Galaxy.

Training Data?     Neural networks typically get better by adjusting their edge weights so that the computations result in the inputs and outputs better matching the training data. Training data usually comes from feeding expert data into the neural network.  


If you want a neural network to be able to identify photos of cats, first you feed it a bunch of photos that a human says are cats.  This is expensive.  It requires human labor.  Keldon’s AI does something different, and this is based on that teacherless system pioneered with TD gammon.   

Reinforcement Learning     So we integrated it.  We use what’s called a “knowledge free” system.  Where do we get our training data?  The neural network actually generates it by playing itself.  The AI is fed input, produces output, and receives a reward based on feedback signal.  The feedback is whether the move resulted in a winning game board, but how do you get feedback from turn to turn, when you don’t yet know if the game was won?  The reward at the end of the game is delayed.  The solution: A temporary credit or blame is assigned at each turn leading up to the final reward at game end.  This is the crux of the Temporal Difference method.

Temporal Credit Assignment     This style of reinforcement learning is based on temporally successive predictions.  What I mean by that is, turn n it trains turn n-1.  And this back propagates, with decay.  So if it’s making a prediction of player 3 winning on turn 2, but then on turn 3 it makes a prediction of player 5 winning, it trains itself that on that state on turn two, it should have skewed toward player 5 – it adjusts its weights.  At the end of game, rather than using prediction, it does use actual winner as training data, but It’s training every time step or turn it gets run.  Each opponent effectively pushes the NN.  30,000 games x 4 players x # turns is about a million Time Steps its trained on.  


Escape Bias     Using this knowledge free system frees us from relying on a human teacher.  The AI only needs to know the rules of the game.  This is interesting for two reasons.  1) It keeps our costs very low.  We can get a ton of training data with virtually no human expense.  2) It frees us from human bias.  This has been a big controversy for AI’s recently.  

Tactics Crystallize     What’s happening here is that the initial strategy is random.  During first thousand training games, tactics emerge such as Piggy Backing.  Keldon’s AI have developed Piggy Backing strategies, completely independent of human input.  


For example, here the AI chose Trade, even though it has no goods to trade.  I fell into the trap, chose settle, the AI can settle a world that comes with a good, effectively piggy backing my settle, and trade it.  The tactics employed can be nuanced and context sensitive, unlike a heuristic driven AI.  

Inputs     In our neural network, what are the inputs?  These are 800 nodes including a ton of game state data.  



AI Bifurcations     In fact these inputs don’t just feed into one neural network.  Keldon bifurcates the model.  One of the most important things is deciding how to architect your bifurcations.  We have twelve unique models of neural networks each trained for a different set of expansions and player count.  If you’re running a two player game, the AI is on a different network than a three player game.  We could always partition further, but there are diminishing returns and you’re burdened with complexity and size – which is bad.  The NNs are one quarter our download size, which is pretty nutso given the volume of art we ship with from all the game cards.  


And for each bifurcation there are actually two flavors of neural networks at work, each with it’s own main function.  So really we ship with 24 neural specialized neural networks.

Outputs     These functions determine the outputs.  The functions are ‘What move would player X make in state Y?’  We call that function Eval_Role.  The second function is given this board, score it: tell me how good it would be for me on a scale of -1 to 1.  We call this Eval_Board.  



Simulating Forward     On its turn the AI simulates through every possible move it could make, and it runs the function called Eval_Board on the results and chooses the best one. In fact, to move the simulation forward a step could involve one or more calls to Eval_Role to guess opponents role choices.  


Tuning Difficulty     The result of all this is that the AI is really hard.  Or at least it can be.   You might think a hard AI is not ideal for some people, maybe new players.  And you’re right!  It’s very difficult to make an AI harder, but you can nerf an AI to be easier pretty simply.  In our neural network outputs, we add noise to the score.  Increasing the noise will make the AI choose the 2nd or 3rd best options.  You can tune the amount of noise you add to find the right difficulty settings.  When determining the noise for a medium AI, you want it to win roughly 25% of games against a hard AI.  

TD NN Candidate Checklist     The main thing is, you can get an AI that can be very skilled, which means can satisfy and challenge players for hundreds of hours.  And that’s good!  But not all games are suitable for temporal difference learning AI.  I’ll leave you with a little checklist of the kind of games that will benefit from a temporal difference NN.  

  • Termination The game needs a definitive end, so that the final reward signal can propagate back to the previous turns.  Chess is a bad candidate because it can result in a stalemate dance between moves, and never generate this final reward signal.
  • High stochasticity A broad possibility space opened up by branching probabilities gives this type of NN space to discover novel strategies.  Lots of RNG is great, so deck shuffling and dice rolling work well.
  • Non-spatial Your NN is trying to learn a function that it doesn’t know yet.  A function is easier to learn when it’s smooth.  A small change to the input, should result in a small change to the output.  If you have a chess arrangement and make a small change to one piece’s position, that could result in a wild change to the probability of winning.  For this reason, games where relative arrangements are critical are bad candidates for TD NNs.


  • Fixed number inputs Your AI learns best when the inputs are fixed.  This is why we branch our NNs when we add a new expansion.  The AI is most efficacious when the incoming variables are the same every time it plays.
  • Multiple turns While it’s theoretically possible to make a TD NN to help with a co-op game, you wouldn’t want to choose a game like One Night Werewolf.  That’s because Werewolf resolves the entire game in one turn, so there could be no backward decaying reward for the nn to learn from.  Additionally, the acumen on a TD NN drops off toward the end of the game because it’s benefiting from fewer forward rewards.

Up Next?     We’re investigating adding a NN driven tree search to our Roll for the Galaxy AI for better performance, based on the recent publication about AlphaGo Zero that extends the NN technique to handle deep search trees.  Right now with Race we do a full 2 ply lookahead, but the branching factor of Roll seems higher and might be too much for phones.  We’ll see.


For games like Race, it’s often the case that simple mechanics produce large search spaces as well.  The most recent expansion for Race, Brink of War, occasionally results in game states where over 500,000 neural networks evaluations are needed to advance the game a single turn.  To overcome this, we currently heuristically discard low probability branches of the search tree.  In the future we could use the techniques from AlphaGo Zero to prune those evaluations and improve performance, especially on platforms like phones and tablets.

–Theresa Duringer

App-Store-Logo1-950x334 Google-Play-Store


Designer Diary: Race for the Galaxy: The Brink of War

The first two expansions for Race for the Galaxy – The Gathering Storm and Rebel vs Imperium – expanded the game by adding start worlds, new cards, and two new but optional mechanics: goals and takeovers.

The Brink of War (which requires both previous expansions) adds Galactic Prestige, which is woven throughout the entire expansion. Galactic Prestige represents the relative standing of each player’s empire and is gained by placing certain cards (with that symbol) or using various powers. With the appropriate powers, prestige can be spent to attack, enable certain powers to be used, or become cards or VPs. In addition, the Prestige Leader (the empire with the most prestige) receives a bonus each round, and any unspent prestige at game end is worth 1 VP apiece.

Thematically, I had the political brinkmanship before World War II in mind, where countries – by playing on old grievances – could use their international standing to both extract territorial concessions and to rally and unify their populace. The first card I designed was “Casus Belli”, which allows its owner – with previously gained prestige – to either attack any player (and, if successful, gain more prestige) or convert prestige into VPs. This second power creates a new strategy (whether takeovers are being used or not): garner lots of prestige, and then Consume:2x one prestige for a net gain of 5 VPs each round.

While 37 of the 48 TBOW game cards involve prestige, this is only ~20% of the combined deck. One challenge was making sure that players who drew only a few prestige cards didn’t feel hopelessly behind a player who got an early prestige lead. If the Prestige Leader bonus was too small, then vying for the prestige lead wouldn’t matter; if it was too large, then gaining prestige early on would dominate. Our solution was to vary the per-round Prestige Leader bonus: 1 VP, plus a card draw if the Leader earned a prestige on the previous round; otherwise (or if tied), just 1 VP (which is nice, but can be easily overcome by other game actions).

We also added a benefit for getting just a single prestige, namely being able to use the new “one-shot” Prestige Opportunity action card that every player starts with. By spending a prestige, a player can get a “super” action once per game (for example, turning Consume:2x into Consume:3x for one round). This action card also has another use, namely, Search, which doesn’t require a prestige, so players who don’t earn any prestige can still benefit from it.

Search: Looking for a Needle in a Draw Stack…

As the card deck gets larger and larger, while the overall variance remains the same (given that we maintain the proportions of worlds versus developments, various powers, etc.), the variation in the subset of cards that any given player draws increases. This can lead to player frustration, particularly if a player is pursuing a strategy that depends on a small number of cards.

Despite adding new explore powers in the expansions, the card variance was still too high, so we added two new mechanisms: draw then discard powers (in which a player draws two cards, then discards one card from hand) and search.

A player may search once per game, flipping cards from the deck to find a card that matches a selected category. There are nine possible search categories, so a player who needs just a bit more Military, for example, could search for a development granting +1 or +2 Military, while a player pursuing an Alien strategy could search for an Alien production or windfall world. When the player finds a matching card, they can either take it in hand or continue searching. If they continue, they must take the second matching card they find. The other flipped over cards go into the discard pile, so searching also increases the odds that the deck will reshuffle in games with just a few players.

The one-shot Prestige/Search, and your search choices

Takeovers: Our Dream of Safety Must Disappear…

The second expansion, Rebel vs Imperium, introduced takeovers, in which players could, under certain circumstances, conquer a military world in another player’s tableau. The Brink of Warextends this mechanic, portraying the descent of a galaxy further into warfare. With “Casus Belli”, a player with both prestige and a powerful Military can now potentially take over any military world, and if a player also discards the “Imperium Invasion Fleet”, even non-military worlds can be attacked. No empire is completely safe.

However, using the “Invasion Fleet” is expensive (though, if successful, prestige is gained), so aggressive players need to balance their potential gains against their costs. The Brink of War also introduces new defenses and incentives. The owner of the “Pan-Galactic Security Council” can, by spending a prestige, block one declared takeover attempt (against any empire) each round. A new 6-development, the “Universal Peace Institute”, rewards players who pursue peace by giving an endgame bonus for having negative Military. And, as before, takeovers are optional, so players who don’t enjoy this type of player interaction need not play with them.

Goals, Uplift, Aliens, Terraforming, and more…

Prestige and the tension of “guns vs butter” are reflected in the five new goals supplied in this expansion, including goals for most prestige, most consume powers, and the first to have two worlds and either a takeover power or negative Military. The “Uplift Code” was discovered in the previous expansion, so The Brink of War details the split between those who wish to breed and exploit the Uplift races and their victims, who rise up in revolt against this.

With the discovery of an “Alien Burial Site” and the “Alien Departure Point”, galactic interest in the long-lost Aliens reaches a new peak (or low point), with the “Alien Tourist Attraction”. Meanwhile, the Golden Age of Terraforming emerges, with “Terraforming Engineers” upgrading existing worlds and various cards with powers that allow players to use goods for discounts, increased Military, etc…

The Brink of War adds four new start worlds, five new goals, prestige markers and a Prestige Leader tile, six search/prestige opportunity action cards, and 44 new game cards to Race for the Galaxy. Enjoy!

Tom Lehmann


Roll for the Galaxy mobile


The popular dice game is getting a digital spin!

The award winning boardgame, Roll for the Galaxy, is making its way to phones, tablets and PC soon. Following the launch of Race for the Galaxy, Temple Gates Games in association with Rio Grande Games is excited to announce the upcoming digital release of the popular dice adaptation.  The game will be available on iOS, Android, and Steam.


RollForTheGalaxy s

Roll for the Galaxy is a dice game of building space empires for 2-5 players. Your dice represent your populace, whom you direct to develop new technologies, settle worlds, and ship goods. The player who best manages their workers and builds the most prosperous empire wins!

Designed by Wei-Hwa Huang and Tom Lehmann, this dice version of Race for the Galaxy takes players on a new journey.



Keldon Jones, the developer behind the Race for the Galaxy AI, is at it again with Roll for the Galaxy. This game will feature a new neural network AI that will challenge even the most advanced players.

Sign up for the newsletter to get info on how to join the upcoming beta.