The Settlers of Catan is something of a phenomenon, when it first came out it took the gaming world by storm. Everyone was playing it, it won all the awards and people just couldn't get enough of it. Lately, the game seems to have lost some of its appeal, however. Personally I think it's a case of a game simply being overplayed but many mention that they've lost interest because the game is "too random". Its pretty easy to see where this idea comes from, after all the game does use dice as its driving mechanism. The player that receives the lion's share of the rolls is more likely to win but does that mean that the game is "too" random? Certainly there's a random element and there's been plenty of anecdotal evidence of a game that was won purely by a series of lucky die rolls. In spite of this I'm not sure that I'd want to classify the game as luck based. Some have suggested that any skill in the game is in the initial placement of settlements, the game is one-third skill (the initial placements) and two-thirds luck (the rest of the game). I'm not sure that I agree totally with this although my opinion may have something to do with my record:
I've played the game 21 times at my weekly gaming session (from which the results are recorded) and I've managed to win a total of 16 times. These were all 4-player games. I suppose its partly ego that leads one to the conclusion that success in a game is attributable more to your skill and intelligence than anything as unseemly as luck.
Still, there's something to this argument. You would expect, in a game of pure skill, that the better (i.e. more "skillful") player would win most, if not all, of the time. If this is so then it seems a natural conclusion that a games "skill" level could be gauged by the regularity with which certain players win. Unfortunately, this "test" has a couple of problems, primarily the fact that you have two unknown (non-quantifiable really) variables: the degree to which the game is "skill-based" and the actual skill of the players. Consider world championship Chess, usually played over a series of 20 or so games with the final result often in the neighbourhood of 11 to 9. Compare this with a "Roll the Dice" contest which will usually result in a similar final score. I think it safe to say that Chess is much more skill based than "Roll the Dice" so these results don't really tell us much in and of themselves. At the very least they don't confirm what we already suspect, which is that Chess is a more skillful game. Even with this misgiving I still think that it's a useful test, a rough indication of a games level of skill, even if it is a rather loose one. In any event why do certain games feel more random to some players than others? Maybe it would help if we had a look at some of the different random elements in games.
The most obvious element is
the use of physically unpredictable devices such as dice. This would
also include such things as spinners or exotic items such as the
devil's pot from In Teufel's Kuche. (If you haven't seen it
its a small pot belly type stove with a button on top. Players
alternately press the button until the devil lurking inside pops out,
very cute!) Basically this includes anything that has a physical
mechanism that's not predetermined.
Two related, and very commonly used randomizing elements, are
cards and tiles. A little distinction is necessary here. What I'm
thinking of is a closed, known set of items and these can take many
forms. In Bridge you know exactly what cards are in the deck.
In Acquire you know that there is a single tile for each
space on the board. In Settlers you know that there are four
sheep, wood and wheat hexes, three brick and ore hexes and one
desert. The random factor in each of these is in how the cards are
distributed, the tiles drawn or the hexes placed. I should also note
that these items are not necessarily random influences. The commodity
cards in
Settlers aren't, they're simply an
accounting tool. The tiles in 1830 aren't a random element as
they're chosen rather than blindly drawn. (I should also note that
cards can be considered as a "dice-type" randomizer in certain
circumstances. If you are drawing and then replacing a card from a
fixed, constant deck then its much closer in function to dice than if
you are drawing and discarding or keeping a card. The specific
example I'm thinking of is Hase und Igel where the Hare cards
are constant in number; you draw one, follow its instructions and
then replace it in the deck. In fact, the original English version
uses dice to perform the same task, albeit slightly modified.)
Another random element is the seating/turn order. In a two-player game this is usually simply a case of who has the first move, normally it's an advantage to go first. In the case of games that have unequal sides (such as wargames like Hannibal: Rome vs. Carthage) its also a matter of who plays which side. In multi-player games its a little more complicated. The order in which you are sitting can have a great impact on your strategy in a game. In Illuminati its generally better to ally yourself with someone sitting next to you as you will have back to back turns. In Show Manager it's very important that you try to stage a different show than the person to your right otherwise they'll be hiring the actors that you want. This is in addition to any advantage of going first. (Note also that its not always an advantage to go first, some insist that going last in Settlers is better as you get to place two settlements at once. Also the first two players in Durch die Wuste only get to place a single camel instead of two.)
The final element (that I can come up with at least) is the players themselves. What I'm thinking of is the mechanism (if there is one) that causes a player to perform a certain action. It would be quite easy to argue that there is no randomness in this at all and I don't necessarily disagree with that line of thought. In any case such a debate is not likely to lead to any useful conclusions and is best left to another forum. I simply include its mention for completeness and that some of my opponents' moves sure as hell seem random to me!
I suppose that each one of these elements is going to bother every player to a different degree. A lot of this is going to have to do with how a game "feels" to us. There are going to be people that have a problem with any game that includes dice. Equally, there will be players for which cards are too great a random factor. I believe that it's possible for any of these elements to be modified to an acceptable level. (I don't think you can actually modify my final element, the players themselves, but we'll ignore that.)
The unpredictability of dice or cards can be modified in many
ways. In Settlers the dice are added together which gives a
bell curve distribution. The game is less random because there is
greater predictability in the commodity rolls than if it was a single
die that was rolled. Any particular result is still random of course
but you will expect more 6's to be rolled than 10's over the course
of the game. This is another important concept: as you increase the
number of die rolls you will get closer to the expected average. I
often hear someone make the comment that they don't like a particular
game because "It's too random, there's too many dice rolls." This
might at first seem to be a logical argument but it ignores the law
of probability that as you increase the instances of a random event
you are more likely to get an expected average. 
To illustrate this, consider a dice contest
between Danny and Roger: A single six-sided die is rolled. On a 1 or
2, Danny wins otherwise Roger wins. If the "game" is a single throw
of the die Danny will expect to win 1 out of 3 times. However, if the
game is the best out of three rolls Danny can only expect to win 7
out of 27 times which is slightly less than 1 out of 3. This is a
result of Rogers advantage in the game. As we increased the number of
rolls the random effect became less of a factor and the advantage
became more significant. The point is that the more dice rolls in the
game the less pronounced their individual effect. There are, as
always, a couple of points to note about this specific example. First
is that each roll had exactly the same effect on the outcome as all
others. In a real game certain rolls are more important than others,
for example, in Titan your movement roll is far more
important than any attack roll. Also, the order of the rolls was not
significant whereas in a real game the effects of earlier dice rolls
often affect later ones. e.g. In Titan if you roll lots of hits at
the start of a battle your opponent will have fewer attacks to make
on subsequent turns and so, in general, it is better to roll well
earlier. So, one way that you could decrease the luck in a game that
uses dice, such as Settlers, would be to increase the number
of individual rolls in the game. This is probably going to be
somewhat difficult to actually implement in a game outside of the
design process. Another, more easily added method would be to
actually remove the dice from the game and there are a couple of ways
this could be accomplished. One particular method I've heard proposed
for Settlers involves a set of 36 cards with each card showing
a number from 2 to 12. The total number of each card listing a
particular result would match a standard distribution so that you
would have one card showing a "2" and six cards showing a "7". These
cards would then be used in place of a die roll to determine which
commodities appear. There are even multiple methods by which you
could employ them; one is to have a single common shuffled deck from
which you drew a card. Another would be for each player to have their
own decks and they would choose which card to play each turn. I'm not
so sure that the final results would be all that much different in a
game like this but it would eliminate the rare games in which a wild
and unusual set of rolls determined the outcome.
In the case of cards, their unpredictability can be modified as well. First consider the case of a game where players draw from a face down deck. A common method of reducing this random effect is to have several face up cards beside the stack. A player now has the choice of choosing one of the face up cards or drawing blind from the deck. Alan Moon often employs this so-called "drafting" method in his games and it works very well. A particular bonus of this method is that it's very easy to tailor it to a particular level, if you want less predictability, have fewer face up cards, if you want more, increase the number of face-up cards. Simple. Another way of reducing the random effect of drawn cards (at the design level anyway) is to reduce the variety of cards. In Euphrat & Tigris there are only four types of tiles so while you can never be sure exactly what you're going to get you have a pretty good idea.
What about the randomness in turn order involving games where it's advantageous to go first? In these cases a bidding mechanism can be introduced where players compete for this right. In the case of two player games you could play a pair of games with the cumulative score determining the winner. Unfortunately there's not much that can be done for the "randomness" of other players' strategies. As there are many that would argue that this isn't really a random factor anyway its probably best to take the practical approach and resist any attempt to "fix" it.
As a practical exercise let's have a look at some real games, specifically Reiner Knizia's so called "tile laying trilogy", Euphrat & Tigris, Durch Die Wuste and Samurai. What random elements are there?
Euphrat & Tigris: The only random elements are the tile draw and the seating order. Implementing an above-described "drafting" tile draw would reduce much of the luck in the present system.
Durch die Wuste: The only random elements are the seating order and the initial setup. (I'm assuming a four-player game where each player has all five caravans in play. In a five player game, where each player only has four of the five caravans, it depends on how you remove each players "extra" caravan. Obviously if you determine it randomly that's another random element.) As a particular layout does not necessarily favor one player over another I'd suggest that there's very little luck involved at all. There's recently been some discussion on the Internet about how to conduct this setup and a couple of people have suggested that the board be setup as follows: Each player, in order, chooses an oasis or waterhole token and places it on the board. Placement proceeds clockwise around the table until the board is complete. I really doubt that all this extra effort is worth it in the end but each to his own. If the turn order was seen as a problem this could probably be solved by implementing a bidding system for who goes first. There are a couple of ways that you could increase the luck in the game: Instead of choosing which camels to place you secretly draw from one of five concealed bags. The game ends when one bag is empty. Instead you could have 6 hidden camels behind a screen, you may play any of them and you replenish your "hand" to 6 from a common stock. Game ends when 12 oasis tokens have been awarded.
Samurai: Random elements are the initial setup,
random draw of tiles and seating order. As with Durch die
Wuste you could solve the initial setup "problem" with a similar
placement mechanism. I don't think there's a 
bidding
system that would work for determining the first player though. The
only thing I can come up with would be something along the lines of
bidding how few of your initial 5 tiles you get to choose. I doubt
that this would work all that well in practice. The random draw is
interesting to me especially when compared to Euphrat &
Tigris. On the one hand Samurai seems less random due to
the fact that everyone will get the same distribution of tiles. On
the other hand there are more varieties of tiles, which makes it less
likely that you'll draw the one you want at the right time. Which is
more random? The biggest complaint I've heard about Euphrat &
Tigris was the random draw of tiles and I can understand this
somewhat, it's very frustrating if you never draw those damn green
tiles! There are a couple of tactics that can be used to overcome
this however and I personally feel that this is what separates the
good players from the bad. I'm not sure if Knizia designed the tile
draw in Samurai as a direct response to this but it does seem
that people have fewer problems with it.
All three of these games exhibit some degree of luck. I think its safe to say that Durch die Wuste has the least of the three but I'm not sure how I'd rank the other two. In Samurai I often have to speculate whether or not Al has his 4 Buddha available. In Euphrat & Tigris the equivalent is speculating if he has 3 red tiles. For this situation Samurai "feels" more random to me but perhaps I'm in the minority?
There are also many games where a certain degree of randomness is "built in". The best, fictitious example I can think of is Chess-Dice. (Which I first read of in a post to rec.games.board by David desJardins. I'm not sure if its his original idea or not.) The game is played in two parts. First a regular game of Chess is conducted and then each player rolls a single die. If there was a winner in the Chess game, that is, the game was not a draw, then that player adds 1 to his roll. Highest roll wins. I think everyone would agree that the game is pretty random even though it also rewards skillful play. In fact, there is just as much depth and skill inherent in Chess-Dice as there is in regular Chess. The big difference though is that the rewards for skillful play are not as pronounced in Chess-Dice as they are in Chess. This is fairly obvious in this case but I'm only using it to make a point. The point being that some games have the appearance of requiring great skill but the game does not actually reward that skill as much as others. There have been a couple of situations where random events have totally ruined a game for me. The most obvious ones have been games that have a "switch positions" rule. Twice I've been the victim of this, the first in a PBEM game of Cosmic Encounter, the second in Flying Carpet. I'd argue that these rules make the games as ridiculous as that of Chess-Dice, a single random event overwhelming any skill that the game requires.
So, back to the original question (was it the original question?), why do some games feel more random to some players than to others? It could be a simple case of closely matched opponents. Most people play with the same group of people and its quite likely that they are of approximately the same level of skill. This is going to have the effect of emphasizing any luck inherent in the game. It could also be the small number of times a game has actually been played by an individuals. Even the 21 games that I've played of Settlers are probably not statistically significant to make any claims about my ability in the game. Also, if you've only played a game two or three times any aberrant events are more likely to cloud your opinion. So, once again I've rambled on about a subject but have I actually stated anything useful? I suppose that's for others to decide but I think the most important point I've made is that it's usually possible to fine-tune a game to your particular luck/skill ratio. As always I'd be interested in hearing any thoughts on this subject.
- Greg Aleknevicus
