Friday, 26 December 2008

Solo Race for the Galaxy

Race for the Galaxy is my current most played game, at 240+ plays. It is also Michelle's most played game, at 230+ plays. When its expansion The Gathering Storm came out, there was no doubt that I would buy it. The expansion comes with some new cards, the objectives (which most veteran players seem to prefer to play without), and the solo variant.

In the solo variant, you play the two-player advanced game against a robot. You get a bunch of unusual components for this variant, and quite a bit of unfamiliar rules for how the robot works. The robot's actions are driven by dice. Some die roll results directly dictate what the robot will do. Some results will make the robot mimic your actions. Some will make the robot take a preferred action, which depends on the robot's starting world. So basically you get to play against 9 different robots, depending on what the start world of the robot is.

Rules for the robot are rather convoluted at first. They are not like the rules for the human player at all. You still play as normal, but when it is the robot's turn to carry out actions, you do them for the robot in a very different way. There are parallels to what a human player does, but to me it was like learning a new game. The robot has a concept of credit, which is roughly equivalent to how many cards a player has in hand. The robot does have a "hand" of cards, but it uses it to search for the appropriate card (dev, 6-cost dev, regular world, or military world) to build, not for paying for the build. The robot has a concept of economy size, which is roughly equivalent to how many points a human player can earn by doing Consume. The difficulty of the solo game is adjusted by using the 6-cost development tiles (not cards), which is yet another new thing. The robot has 1 to 3 of these, each worth 6 or 9 points, depending on the difficulty level you want to play at. These tiles can be built by the robot quite easily (in my opinion) during the game, during the Develop phase (of course).

So, as you play, you also manage the robot actions, and you race against the robot. The game ends in the same way as human-vs-human games - victory point chips exhausted, or 12 cards built.

Having played about 10 solo games now, my first impression was the robot is rather dumb. There is sometimes no logic to the robot's "choices". Such "choices" are just random die roll results after all. Sometimes it feels that the robot is not making sense at all, "picking" an action that does not benefit it at all. Also it seems just very stupid to have 6-cost development cards worth 0 points. Why bother building it? Then I started to see that the robot is not really meant to be intelligent. "Robot" does not necessarily equal "Artificial Intelligence". The robot was designed to mimic a human player, or rather to mimic what a human player can achieve in a typical game of Race for the Galaxy. I say this because the robot is following very different rules to achieve it. Playing against the robot feels very different from playing against a human. It is more about managing risk and calculating probabilities (i.e. outcome of the die rolls) rather than guessing your opponent's intentions and guessing what his/her long-term strategy is.

To beat the robot you need to understand how it works, and you need to understand how to exploit its limitations, or at least know how to minimise the chances of indirectly helping it. Knowing that sometimes it "copies" your actions, you need to think twice before selecting Consume x 2 if the robot's economy size is 5 (i.e. it would score 10VP if it also does Consume x 2). The robot's main ways of scoring are the 6-cost development tiles and the victory points. I find that usually it doesn't score much from worlds or normal developments. The 6-cost dev tiles (as opposed to dev cards) are a mechanism to adjust the difficulty level. The robot most of the time will be able to play all of its 6-cost dev tiles, unless you are really really quick in ending the game. So when you decide to play with 1 6VP tile (easy), 2 6VP tiles (medium) or 3 9VP tiles (hard), you are more or less already deciding that the robot will at least have 6VP / 12VP / 27VP.

You need to watch the robot's economy size closely. To me, it's kind of a countdown timer. The robot's economy will grow, sometimes quickly sometimes not, and the bigger it grows, the more dangerous it will be. You cannot allow it to do Consume too many times, especially when its economy size is big. It can earn lots of VPs this way, and also end the game very quickly by exhausting the VPs.

In contrast, the robot playing cards into its tableau is a mostly random affair. When doing Develop or Settle, the robot just develops or settles the first card of the right type as you flip through its draw deck. The robot doesn't choose which card to play. Most of the time the robot is slower to play cards into the tableau, and normally it doesn't score lot from these cards played, especially the 6-cost development cards. But of course sometimes when the robot is lucky it will keep drawing and settling the big worlds, or it will develop a 6-cost development card which jives with its tableau well.

Correction, 28 Dec 2008. 6-cost dev cards played by the robot count as 6VP or 9VP, depending on the difficulty level, and do not score according to the special text on the cards. Thanks crushedguava for pointing this out.

One more difference between robot-vs-human and human-vs-human is you get to see many more cards, because you help the robot flip through its deck, and you can take the opportunity to see what cards have appeared. This can be useful, but I'm not quite at the level of expertise yet to make good use of this information or to do card counting.

One tiny detail that I recently discovered - the hand icons on the robot playmat (left) are different from the hand icons on the player action cards (right). Actually you don't need to use the player action cards when you play solo. The robot hand has sharp edges, and one thumb and three fingers. The human hand has rounded edges, and one thumb and four fingers.

A close-up for better comparison.

I am now playing against the medium difficulty robot. I win slightly less than half of the games. It gives me enough challenge. I still don't feel I have fully figured out all the tricks to beat it, but I'm enjoying the exploration. My view of the solo game is it is a worthwhile exercise if you set your expectations right. It won't replace human-vs-human games. It is something very different. I still prefer human-vs-human games, but the solo game is good enough that I'd pull it out now and then to challenge myself.

My summary: The robot is dumb, but it can still beat you.


crushedguava said...


I've been playing RftG solo these few days as well.

Just to correct you on something (at least, I think I'm correct), when the Robot flips and finds a 6-cost development Card (not the tiles) from the deck, its value is not 0, rather, its value is either 6 or 9 depending on the difficulty you're playing at, just like the tiles. This is stated in page 10 of the rulebook.

I hope you have fun (and frustration) trying to beat the robot. =)

crushedguava said...

Just to make myself clear, I think the 6 cost development cards are ONLY either 6 or 9 points for the robot, regardless of what it has in its tableau. You don't actually score them the way you normally do with a human player.

Hiew Chok Sien 邱卓成 said...

Thanks for the correction. Wow, then the robot will be even tougher to beat. I may need to go back to Easy level. :-D

crushedguava said...

Haha yeah! Its to provide us with a challenge so we don't get bored with it. I absolutely love how they personalise each robot according to the start world, so its like you're playing against a different robot everytime.

Hiew Chok Sien 邱卓成 said...

I played again with the right rules, and got absolutely crushed by the robot, 65pts vs 39pts. It had 2 6-cost dev tiles, and 2 6-cost dev cards. It also had 21pts from VP chips and 20pts from regular cards. So it would have beaten me soundly even without the 2 6-cost dev cards.

I checked back previous games played, and to my surprise this rule correction doesn't change the winner much. The robot didn't build 6-cost dev cards much. And when it did, and I scored too little for it, it had won those games anyway. This is one time that I'm happy to be obsessive-compulsive - I record all my games played and scores, and for RFTG, even the breakdown of the scores. So I can go back to re-check the results.

Having played a few more games, I feel that the different robots actually do not play too differently from one another. They still do behave slightly differently, but some feel quite similar.

crushedguava said...

Hmm. I find that to beat the robot, you'd have to absolutely do what your own home world does best, instead of doing hybrid strategies. For example, if you were New Sparta, you'd have to make sure you get down multiple big military worlds plus at least 2 of the 6cost devs that help you, or you have no chance. Doing a production strat with New Sparta because you have good production cards in hand (which you would do when playing with another human given the proper situation) will get you thrashed. But maybe I'm not good enough at that yet.

I find your comments on the robots behaving similarly quite interesting. I've been playing using New Sparta against the difficult robot with different homeworlds quite a bit (getting trashed by Ryan from Mage Cafe was an eye-opener for me), and I find that its not too difficult to beat the Alpha Centauri robot and the ELC robot (both multiple times), but playing against the Old Earth robot and the Epsilon Eridani robot was an absolute nightmare.

Even against AC and ELC, I found that I had to play differently (regardless of what cards I had) to beat them. From my limited experience, against AC, its better to start of with Settle/Trade into better cards and a good 6dev, while with ELC, its better to Explore/Explore (even if I have the correct cards for a Settle/Trade), just because of the way the robot is programmed to respond.

Also, there are some scripted events where it'd be much better for you to do a consumex2/produce rather than a settle/consumex2 even if the latter option is what you'd do in a normal game because if the robots decides to copy you, it would copy the produce rather than the consumex2.

I'd be interested in knowing your thoughts about the robot from your plays. Didn't expect to actually enjoy playing the robot as much as I do now haha.

Hiew Chok Sien 邱卓成 said...

I have not played very many games against the robots. Maybe I'll appreciate their differences more after I play more.

I find that I tend to play to maximise what I can do with my hand, rather than to minimise what I may allow the robot to do. This is one area I need to be more careful about. Thanks for sharing this strategy consideration.

Regarding having to play to your start world's strength to have any chance of winning, I'd be quite disappointed if that's the case. If that is the case with the hard difficulty then I can still accept it, but if this is necessary even when playing on medium difficulty, then we might as well be playing the first 2 rounds, and then decide to concede defeat if we don't get the right cards to execute a strategy that jives with our start world.

So far I tend to balance both start world and start hand when I decide what to pursue, and I find the win rate is not too bad against the medium difficulty robot (although less than 50%). I like to have some flexibility. I may not ever "graduate" to hard difficulty level.