BrettSpiel is a blog about board game design, written by game designer Brett J. Gilbert.

3 Comments

Carcassonne for iPhone: Charmingly Similar, Curiously Different

There are two things to say about the Carcassonne app:

  1. For people who have never played Carcassonne it’s a great introduction. Elegant, engaging and completely faithful to the charm of the original, with a great interface that makes learning and playing the game a real joy.
  2. For people familiar with the Spiel-des-Jahres-winning original it’s just as elegant, engaging and charming, but it can be an oddly alien experience.

I heartily recommend the app, and for people in the first group it’s one of the best implementations of a Eurogame on the platform, and has the advantage over Small World, another excellent app which I have already reviewed, that it runs equally well on the iPhone, iPod touch and iPad. But as a true fan of the cardboard edition of Carcassonne, I have my reservations.

Image: Brains and Beauty: The app very clearly and very smartly displays the choices for placing your follower after you place each tile. The whole interface is a great, great success.

A great feature of the app is its variety of built-in AI (articifical intelligence) players, which means that you can play against one or more opponents even when nobody else is around. Carcassonne is at its best as a two-player game, and almost all of my games on the app have been two-player.

What struck me first about the stronger AI players was how adept they were. Creating effective AIs for anything other than the simplest games is a sophisticated task, and though Carcassonne is a relatively simple game, it isn’t necessarily easy to play well, nor are all the tactics and strategies obvious or short-term.

But as good as the AIs are, they are essentially one-trick ponies. The basic AIs really are only suitable opponents for the true novice, and the more advanced AIs have, for me, become too predictable too quickly to provide genuinely satisfying opposition for the long-term.

Image: To Capture the Count: With three farmers already on the board, and three followers trapped in unfinishable features, the Count only has one follower out of seven still available to score additional features for the remainder of the game — and he has to get that follower back first!

The advanced AIs are all incredibly competitive, which is to say that no effort on my part to build a road or city goes unchallenged. Competition is good and fine and necessary, but the AIs can do it so blindly, and in a way that can, more often than not, diminish their own chances of winning more than mine. The only form of defense against this playing style is attack — to play the AIs at their own game — which only becomes an easier task the more you play each AI, since their behaviour is just too predictable.

My dissatisfaction, then, is that this dynamic creates a sort of ‘anti-Carcassonne’, where the only objective is to tear down and trap your opponent; all the while reducing their score rather than building up your own. Many 2-player games are ‘zero sum’ — where a point taken from your opponent is as good as a point in your favour — but the ‘better’ AIs know nothing else.

Image: Scoring Made Simple: The graphic breakdown of each player’s scores is an excellent feature. The differently coloured chevrons represent the relative number of points scored from different features: beige = roads, orange = cities, red = cloisters, green = farms.

And there are other factors that add to this sense that the game is unsettlingly ‘different’ in this incarnation. For one, the AIs are not terribly effective or efficient farmers, often going for the farms too early and then neither protecting nor competing for them. Farming is the trickiest bit for new players to learn, so its not surprising that the AIs have trouble; and I would expect the AIs to get better at this in future updates to the app, so this isn’t a dealbreaker.

Another thing that contributes to the altered dynamic is the fact that the number and configuration of the remaining tiles is always available information (you can tap the little ‘tile stack’ icon on the large blue banner to see the numbers). It took me a while to cotton on to this, and also to realise that the AI players were obviously using this information to their advantage.

As a Carcassonne old-timer, I have a feel for the balance of tile configurations in the game, but I’m hardly Derren Brown. Knowing — rather than just guessing and hoping — what tiles remain radically changes the game, and significantly enables the ‘anti-Carcassonne’ gameplay style by making blocking moves much more attractive and definitive.

Image: Final Analysis: The app keeps a record of the statistics for every player, including the AI players. And what’s surprising to me about my own scoresheet is how similar the average proportions are for cloisters, roads and farms.

But, to repeat what I said at the beginning, this is a top-notch implementation of a great Eurogame, and one that I can, for all its faults, still heartily recommend. The original Carcassonne is one of my all-time favourites, and so perhaps my criticism is borne more of a personal disappointment than a completely objective analysis.

For me, the app somehow manages to be, at one and the same time, both an entirely faithful and an essentially inaccurate experience, and I am reminded of that great quote by Richard Avedon, the famous American photographer:

All photographs are accurate. None of the them is the truth.

Richard Avedon

Cardboard, I’d both wager and hope, isn’t going anywhere anytime soon.

Good article, Brett. Competent AI in boardgame apps makes them much more interesting. However, Carc has *good* AI! Especially compared to say Catan or even Scrabble, both of which seem to play deliberately sub-optimally (the Catan AI is especially poor with choice of starting positions, which is unforgivable; the Scrabble AI is barely adequate even on "Hard" and its endgame play is laughable, given that it has perfect word knowledge).
The criticism of the Carc AI as a one-trick pony is a little unfair, since you can make it a lot tougher for yourself by playing multiple opponents. The game in its 2-player incarnation is necessarily zero-sum, but that's not true in the 3+ mode. Playing say vs. Maid and Countess is tough, because the Maid usually loses but cleans up on the farms, because you simply don't have enough meeples to compete with both, and if you hammer away at the Countess the Maid can sometimes beat you both. I just ignore the tile breakdown, and it hardly matters that the AI is aware of it. I'd be interested to hear if you find it just as easy (the games with multiple AI opponents are quicker, so no excuse there!)...

Good article. I've been an occasional "cardboard" player, but the app has been a go-to game for me. I find it one of the better replayable games for my iphone.

Like makendo, I find playing with more than 2 AIs at mixed levels a much more compelling experience. The "count" has definitely taught me a more aggressive style of play than I every experience on the "cardboard" version.

As for differences in game play, I like that the app shows what plays are possible, as the small screen makes it hard to discern. I don't allow myself to peek at the remaining tiles as it isn't analogous to regular game play, even though I suspect the AIs do.

However, what does bug me is that the app will tell me when placing a tile, what spaces will no longer be able to be placed. That is impossible in the cardboard version and adds a definite element to the game.

Being able to know what tiles remain does change the game a lot but I find I actually like it a lot. There is much more tension in knowing a certain tile is coming up rather than just hoping there is still one in the stack. I'm sure some people can track this and have that ibformation in their head at the table, but I am just not at that level of play and really love this iPhone version. Convinced me to finally buy the real game.

Post a Comment

Older post / Newer post