Thanks to Temple Gates Games for providing a review copy of the Race for the Galaxy app and all current expansions.

Tom Lehmann’s Race for the Galaxy has made an incredible impact on the board game world since its 2007 release by Rio Grande Games, spawning five current expansions (with a sixth, Xeno Counter Strike, on the way) and inspiring long-term play in a way few other games of its age have. There’s a fascinating history to the game, especially in its production at the same time as San Juan and in the crossover between them (Shannon Appelcine details that here), and it made some important strides other games have since built off (simultaneous role selection, cards as currency),  but its enduring impact has been because the game is so good.

This has led to the recent iOS/Android/Steam Race for the Galaxy app by Theresa Duringer and Temple Gates Games. The app was launched last year and currently offers the first three expansions (The Gathering Storm, Rebel Vs. Imperium, The Brink of War) as in-app purchases ($3.99 each; the base game is $6.99). It also includes the New Worlds promo pack. The game is not just a faithful interpretation of what makes the physical game great, it’s one that even improves on it on some levels. This is an app that’s jumped right to the top of my current favorites, and it’s one highly recommended both for those who already love Race and for those looking to see what all the fuss is about.

Some personal context here; Race is quite likely the physical game I’ve played the most, hundreds of times at this point, and its previous digital implementation (a computer version by Keldon Jones, who will be mentioned more below) is definitely the digital board game I’ve played the most (probably close to 500 games at this point). I love physical Race for how it produces fascinating decisions in a short time frame, how it shines in a wide variety of multiple player counts, how it plays out so differently each time, and how it rewards both exploration and experience. The computer version kept all of that with some improvements, and now the app version (which I’ve played about 100 times over the last couple of months) has improved things even further. I’m a big fan of board game apps in general (I have about 40 installed across my devices), and this is one of the best I have.

How does this game work? Well, the app includes an excellent tutorial, which is one of its big steps forward and something that’s very important for a game like Race that’s often seen as tough to teach. The tutorial takes you through several training missions, introducing you to the various concepts of the game and to its iconography. And something that’s very useful is that you can always tap to expand particular elements, which will usually help clear things up. Plus, there are in-app rules explanations (accessible both in-game and outside of a particular game), which are particularly helpful for detailing the changes in any specific expansion.

For those who are new to Race, the base idea is that each round has five possible phases that each player can do if they’re selected. Each player will select one phase per round, or two if playing with the advanced two-player variant (which is an option in the app). The phases are explore (draw some cards, then discard some), develop (play cards with diamond symbols into your tableau, settle (play cards with circle symbols into your tableau), consume (turn in goods for cards or victory points) and produce (put goods on your production world). There’s also a bonus trade action that can happen before consume and involves turning in goods for cards, but only happens for the player or players that select it.

A clever idea in Race is that each card can represent one of three things. Each card in your hand is a possible development or settlement for your tableau, but you’ll have to use other cards to pay its costs. And goods, when you get them (either by settling a windfall world, which comes with a good already on it but doesn’t naturally produce, or by settling a production world and then later producing on it), are also the same cards, but flipped face down, placed on the world in question, and never usable as anything else. Cards as goods doesn’t really matter in the app, but it’s a nice way to avoid adding another component in the physical game.

But the base idea that everything is either what it says on the card (development or settlement) or fuel to place other developments or settlements is a crucial one, and it brings in a lot of the game’s tough decisions. Many of the most powerful cards (especially the six-cost developments, which usually reward you with differing numbers of victory points based on what else you’ve built) aren’t easy to play, so you’ll have to sacrifice lots of other cards to get them out. Choosing what to hang on to and what to discard to build other things is a vital element here, and not an easy decision to make, especially as you don’t know what cards you’ll be getting next and what synergies they’ll have what you’ve already played.

The game is largely an engine-builder, with everything you play helping you in some way by giving you more cards or making it easier to play other things. At the start, you can’t do all that much, but as the game progresses, you’ll be drawing more cards, producing more goods, and playing higher-cost worlds and settlements. There are two main paths to take, military (which requires a bunch of specific developments and worlds to get going, but then lets you settle military worlds without discarding cards) and civilian (which often involves a lot of producing and consuming goods), and you start each game with at least the theoretical option of either; you’re dealt a military starting world and a civilian starting world, plus six other cards, then have to choose one of the worlds and four of the cards to keep. This opening decision provides a useful look at what path might make sense to explore, but it’s possible to shift course later on if you draw other cards, or even to do a little of both paths.

The game then moves through its cycle of rounds. Each round involves phase selections at the beginning and discarding down to 10 cards in hand at the end. This continues until one player hits at least 12 cards in their tableau (there’s one particular card that lets you go to 14 before triggering the end of the game, but only one), or until the victory chip supply is depleted (the pool size is 12 times the number of players). Scores are then tallied based on the victory point value of each card in each player’s tableaus, the number of victory point chips each player has, plus potential bonuses for goals or prestige if playing with expansions.

Something that’s awesome about the app is the sheer speed of play. The physical game doesn’t take too long (the box lists 30-60 minutes, and games with new players can be on the upper end of that range, but games with experienced players can take place in as little as 15 or 20 minutes, especially with the advanced 2-player variant), but the app improves on that. I just played a match in four minutes (base game only, vs. one AI, 2 player advanced variant), and an average match against the AI  probably takes me between five and 10 minutes. Moreover, any slowdown is probably more about how long it takes you to think rather than any delay from the app; the AI makes its decisions very quickly, and the animation speed is quite fast. And games with other players (online play is available) are also quite fast, and the app even lets you play multiple games at once (online and/or against the AI) and jump quickly to whichever one it’s your turn in.

A big part of that time improvement over the physical game comes from the automation of setup and of dealing out cards in each phase. This game comes with a hefty deck, especially when you incorporate the expansions, and it can take a little time to get it all ready. The app speeds that up significantly, and it makes sure the rules are all followed and everyone gets what they need in each phase. But there are improvements even beyond that; it’s much easier to inspect your opponent’s tableau digitally (and thus figure out what phases they may select, a big part of the game), you can also see what level of points they currently have, and the app presents how many victory points each six-cost variable VP development will earn you based on the present state of your tableau (which is a huge step up from the computer version), eliminating mental math that can slow the game down. This is also key when it comes to final scoring, which can take a little while with the physical version, but is incredibly fast in the app.

Where the app is also very useful is in expansion sorting. Each of the three expansions available as in-app purchases have their own cool features that tweak the game, in addition to a new array of cards, and after you buy them, you can select to play with any or all of them. On the tabletop, that would be an incredibly cumbersome process; yes, the cards are marked based on what expansion they come from, so you could theoretically pull expansions out after you add them in, but that would take quite a while. It generally makes more sense just to play with all the expansions you have (at least, expansions #1-3, which work together and are the ones available in the app; expansions #4 and #5 are supposed to be played with just the base game) than to try and sort them out. But in the app, after you’ve played the most complex version with all the expansion elements, it’s easy as anything to go back to just the base game for a change of pace. And that adds further replayability here; base Race is quite different than the Race at the end of the first expansion cycle, and both are fun experiences.

Also, a huge selling point of this app is the AI. Race has a long history of solo play, with the first expansion in 2008 including official solo rules, and that led to Keldon Jones creating the first version of an AI for it in 2009 as part of a research project into neural networks. He’s improved and refined that AI over the years, and it’s included in this app, now with various difficulty settings (which is good for those who don’t like how brutally challenging the hardest-level AI can be). There’s a good description of just how this AI works and what makes it distinctive in this Venture Beat piece on the app from December:

Race for the Galaxy has been one of the top-selling digital card games 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 does not require human input to generate training data, which makes it extremely 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 change between predictions from one turn to the next is diminished. Through this method, over 30,000 training games, it has 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. This makes the game replayable over time.

Duringer tapped an artificial intelligence engine developed by AI pioneer Keldon Jones. Jones created the AI for Race for the Galaxy as a research project on neural networks. He released it as open source code, and Duringer used the A.I. as part of her game in a partnership with Jones and Lehmann. Jones’ method was based on one created by Gerald Tesauro, who created TD Gammon, which is based on backgammon.

The AI here really is great, making smart decisions incredibly quickly, and the easy/medium/hard settings for it are excellent. Even the lower levels can provide a challenge, but they’re not insurmountable even for beginners, while the hard level remains quite the opponent even for seasoned veterans. And that speed should be praised again, and it’s a big part of what keeps me coming back to this app, especially for solo play. Online play can be interesting, with options for real time (30 minute max) and longer games (1 or 4 week maxes), and there are usually a fair amount of people there, but the solo game really shines for me; it’s an opponent that’s always around, that’s quite customizable, that you’re never waiting on, and that can finish a game in 10 minutes or less. That’s tough to top. Oh, and you can play against multiple AIs at once, allowing you to experience the differences in a larger-player-count game without actually having to find more players.

So why keep coming back to Race after all these plays? For me, it’s the card combos. Each game has its differences based on what cards you draw when, and there are tons of interesting combinations to explore. And that’s especially true when you add in expansions, which bring in goals (The Gathering Storm), takeovers (Rebel Vs. Imperium) and prestige (The Brink of War) in addition to hordes of new cards. But the base game alone has a ton in it, and even drawing the same cards you’ve held in previous games often doesn’t feel identical, as the order you draw them in matters, the decisions you make with them matter, and what your opponent does matters. The physical game’s held my interest for over a decade now, and the app’s provided me with another great way to experience it, one I intend to keep exploring for years to come. And hey, if anyone’s looking to play…let me know.

The Race for the Galaxy app is available on iOS and Android for $6.99, with expansions available as further in-app purchases for $4.99 each. Screenshots seen here are from the game page on the Apple and Google app stores.