Me and my meeples
I began to play Carcassonne around 2007; I discovered a site where you could play Settlers of Catan online, and you can guess the rest of the story. I didn't own my last copy until last year, so I originally had to extrapolate the rules by observing what tiles gave what points. At that point Carcassonne was one of a number of games I played casually on the internet, but all of those games were abandoned in 2009 or so to free up time for a short-lived Diplomacy obsession. (A story for another day.)
Then last year, one of my partner's friends introduced her to Carcassonne, and it quickly became her favorite game. We play it... a lot. I wouldn't exactly say we're experienced. We both still make pointless blunders. But I've played enough to have some sense of what works and what doesn't.
Carcassonne is, by now, a very old game -- archaeologists in Provence are still working hard to figure out how old. But in case you've been in a cave (hiding from the Inquisition?) for the last decade, a summary.
The game starts with a set of face down tiles. Each player has seven wooden men ("meeples") in his or her color. On your turn, you place a tile, lay it next to an existing tile; next you may place a meeple on a vacant feature; finally, you score any features which your tile completed, and return any meeples on them to their owners. Tiles are square, and portray the medieval French countryside. The tiles are some mix of green ("farmland") and brown ("cities"); each side of a tile will either be completely green or completely brown. Roads run through farmland, and a green tile-edge can either have a road or no road. The edges of a newly placed tile must match the edges of all adjacent tiles. One terrain feature, the monastery, is in the middle of the tile and does not cross the tile-edge.
Features are scored when they are completed, or at the end of the game. A road is completed when it forms a complete loop, or has two dead ends. (Roads can dead end into monasteries, cities, or other roads.) A monastery is completed when it is surrounded by eight tiles (four adjacent, and four at its corners). A city is completed when all brown tile edges in the city are touching another brown tile edge. Farms are never completed and are only scored at the end of the game. Completing a feature is the only way to retrieve a placed meeple (making it eligible to be placed on another vacant feature). Meeples can only be placed on vacant features, but a tile placement may connect two (or more) features; if both features were occupied, then more than one meeple may end up on one feature, and in this case only the player(s) with the most meeples on the feature can score points for it.
Multiple expansions are available which add new features and rules; but I don't particularly enjoy them and will not mention them.
I have played many games with four or five players (all online), many two player games, and a smaller number with three players. In my opinion, Carcassonne works best as a two player game. First, the play experience is better. In the two player game each player can only place tiles with an eye to helping himself or hurting his opponent, and since both players are equally interested in both goals, every move is of equal interest to both players. Both players are attentive, and the game moves quickly and stays absorbing. With three or more players, other player's moves are not as important, so it's harder to focus on them, so each player needs to study the board longer on his turn, so it's harder to focus... and so on. Second, the game doesn't balance particularly well. There is a collective action problem that makes taking down the leader relatively unrewarding; tiles and meeples devoted to hurting other players' positions help the by-standers more than the aggressive player, if there was a good alternative use for them. The aggression issue is particularly pronounced in the struggle for features. In a two player game, a jointly-occupied feature is a dead zone until one player decides it's worth slipping yet another meeple onto it, which leaves the board in a state of tension. In a three player game, both occupiers are generally happy to cooperate to finish the feature off quickly, which is rather boring.
A final point: the more players, the higher the variance on the expected tile-draws. This gives the game a punchier, luckier feel (which some people may prefer); you will get fewer valuable tiles, and whether those tiles are very good for you or just somewhat good matters much more. With fewer tile-placements per player, solid play is less likely to overcome random events.
That last thought on five-player Carcassonne seems like a good pivot for more thoughts on the role of luck in Carcassonne. This is an irreducibly lucky game. The order of the last two tiles may lead to a thirty-point swing in the score. In the last month or so, my partner has drawn six out of six monasteries twice (equivalent to an extra 27-54 points). This will happen in 1 in 16 games. With luck playing such a large role, I don't think I will ever come to take Carcassonne seriously as a strategy game. Spending extra time plotting and analyzing the board brings a relatively low dividend in terms of games won. I may eventually try playing Carcassone in sets and/or using a doubling cube; these are the standard ways to stiffen the backbone of a game where too much hinges on luck.
One good thing about Carcassonne is that cardsharping (tile-sharping?) is not an important skill in the game. When I started playing I assumed that the process of getting better would include gradually memorizing the tile deck and tile-counting. I don't find cardsharping particularly fun. But in fact, the categories of tiles you are looking for are relatively simple and if you want to know the odds to make a decision, it can be done very easily by glancing at the board and consulting the chart of tiles that comes with the game.
Strategy and tactics
The strategy of Carcassonne is much more interesting than I had anticipated when I started playing. There is a conflict between efficiency and flexibility. You want to change the net score by as much as possible per tile placed. Maximizing the points you get from a tile requires you to place a meeple, or to already have lots of occupied, unfinished features so that you can connect the new tile to an occupied feature. But once your meeples are placed you can't control when you get them back, so you start to run into dynamic flexibility problems. If you have no meeples, you lose the ability to pick up free points on features that you can occupy and complete in the same turn, and you also have to pass up profitable features like monasteries. But once you only have one meeple, can you really afford to place it on a monastery and be left with no meeples? So perhaps it would be better to have two meeples. But with only two meeples, you have to start passing up a number of interesting scoring opportunities, because that leaves you with only one meeple. Three meeples, then? But how efficiently can you use your tiles with only four occupied features...?
This is the main tension in the strategy of the game. There are lots of smaller points that I'm gradually discovering (partly through the metagame: see below). There are endless variations on the ways to strand your opponents' meeples, either permanently or temporarily. The timing for completing features is subtle; when will it be large enough for your opponent to block/invade? How and when to react to an opponent's farm also has wrinkle upon wrinkle, given all the interactions between farms, cities, monasteries, and roads. (For example, your apparently worthless farm is a safe place to leave vacant cities.)
People use "metagaming" in different ways. Usually it refers to preparation for the game that happens before you are interacting strategically with your opponent: deck-construction in collectible card games, build orders in real-time strategy games, opening research in chess, and so on. Here the Carcassonne metagame is just the way my partner and I adapt our play style from game to game to react to the way the other is developing as a player. Just one example: when we started, I thought my partner over-emphasized fighting for features, so I would just let her waste two meeples and several tile-plays on a feature and rack up points elsewhere. She realized what I was doing, so she tempered her aggression slightly and began to really stomp me. Then I had to change my peaceful style in order to deal with her (now quite clever) invasion of my features. We play the game way too much, enough that I'm a bit sick of it, but I don't think it can get too boring precisely because the metagame gives this little addiction of ours an interesting narrative. It's similar to the way each of us puts pressure on the other's tennis style, if that comparison makes any sense to you. It keeps our games fresh.
Carcassonne is fun! If you've only played it a handful of times, it's a good go-to casual game. I like to play as black. Any questions?