We had this project's strongest AI (gen1420, adopted 2026-11-20; the same AI as "Strongest" on the play page) play 48 games against itself and recorded every move. For comparison we used "an AI that just picks the move scoring the most points this turn" (the greedy AI below). The strong AI chose the same move as the greedy AI only 28% of the time. The other 72% of moves are full of tips a beginner can copy.
Strong AI vs itself: 48 games
Greedy AI vs itself: 200 games
Average score per player: 95 (greedy AI: 79)
Reading the diagrams:First player (red)Second player (blue)Tile the AI just placed× square no tile can fill
MeeplesHow to use your 7
You only have 7 meeples. This is where the strong AI differed most.
Don't use them all up early
It's tempting to place a meeple whenever you can, but the strong AI declines to place one more often than the greedy AI and uses them more slowly. At the end of the opening (48 tiles left) it still has 2.0 in hand on average. One spare meeple lets you seize big chances, such as joining an opponent's city or claiming a large field. From the middle game on, even the strong AI gets down to about one in hand (0.9 on average with 24 tiles left), so think "don't rush to place them all" rather than "hoard them".
Strong AI36%
Greedy AI9%
Share of turns where it chose not to place a meeple despite having one
Average meeples in hand at the start of a turn. With 48 tiles left, the strong AI has 2.0 and the greedy AI 0.5.
From a real game (opening, 69 tiles left)Blue placed a tile but kept its 7 meeples in hand. The greedy AI would have placed the same tile and put a meeple on the road. A meeple on a feature with no clear end won't come back for a while.
Only claim roads that will end soon
Roads are cheap at 1 point per tile, and they never finish if the ends aren't fixed. The strong AI mostly puts meeples on roads when extending an existing road, and those meeples come back quickly. Only 7% of its road meeples started a new road.
Strong AI7.2turns
Greedy AI14.8turns
Average time until a road meeple returns (93% of the strong AI's road meeples were placed on existing roads)
Average number of turns from placing a meeple until it returns (counting your own turns). Farmers stay until the end.
Start your own new cities and grow them
43% of the strong AI's meeples went into cities (greedy AI: 35%). And 62% of those were placed on the first tile of a new, unclaimed city. A completed city is worth a lot, 2 points per tile plus 2 per shield, and a city of your own is hard for others to interfere with.
Strong AI62%
Greedy AI30%
Share of city meeples placed in a new city (one not connected to anything)
Few farmers, in fields with many completed cities
Farmers (meeples in fields) stay until the end, but at 10.3 points each on average they earn more than any other placement. Even so, the strong AI places only 2.4 farmers per game on average, fewer than the greedy AI (3.3). It spreads them out, 0.8, 0.7 and 0.9 on average in the opening, middle and end game, and late on it puts returning meeples into fields that touch many completed cities.
With farmers, location matters more than count. The average gap in field points between winners and losers is +1, and it grows or shrinks with the AI version and the luck of the games (where the gap comes from). Don't place so many early that you run short for cities, but when you find a field where completed cities are likely to gather, claim it.
Average points per farmer10.3pts
Farmers that scored 06%
The strong AI's farmers (2.4 per player per game on average, on the board for 17 turns on average)
From a real game (end game, 0 tiles left)Red placed a meeple in a field (a farmer). This field touches 8 completed cities, so it is already worth 24 points. It also touches 6 unfinished cities, worth 3 more points each if completed.
TilesWhere to place them
The strong AI doesn't just extend its own features; it also gets involved with its opponent's. It placed a tile touching an opponent's feature on 14% of turns, versus 11% for the greedy AI.
Join your opponent's big city and share it
You can't place directly in a city your opponent occupies, but you can start your own small city next to it with a meeple and connect them later with a tile. With equal meeples, both players score in full. In the strong AI's games, 13% of all points came from such shared features.
From a real game (opening, 67 tiles left)Blue connected its own city to Red's, making a city of 4 tiles. Red and Blue have 1 meeple each, so both score in full when it is completed. It stops Red from keeping the city to themselves.
Make your opponent's city uncompletableAdvanced
If you leave a hole next to a city that none of the remaining tiles can fill, the city can never be completed. Instead of 2 points per tile when completed, it scores only 1 per tile at the end. The strong AI played this move 5.1 times as often as the greedy AI.
The strong AI plays this 0.78 times per player per game on average (75 times in 48 games). But you can only aim for it if you count the remaining tiles. As a beginner, focus on completing your own cities first; "when unsure where to place, consider spots that get in the way of your opponent's city" is enough.
Strong AI0.78times
Greedy AI0.15times
Times per player per game it made an opponent's city uncompletable
From a real game (middle game, 45 tiles left)After the tile Blue placed, no remaining tile can fill the square marked ×. Red's neighbouring city can never be completed.
Fill around your monastery yourself and go for 9
A monastery meeple scores 8.3 points on average, but 55% are never completed. When there are gaps around its monastery, the strong AI fills them with its own tiles.
From a real game (middle game, 30 tiles left)Red filled the last empty square around its own monastery, completing it and scoring 9 points with this move. The meeple comes back and can be used again next turn.
ScoringWhere the gap comes from
Winners pile up points in cities
Comparing winners and losers by source of points, winners also scored more in cities. From the winner's side, the gaps are cities +7, fields +1, monasteries +6 and roads +2. Apart from cities, the gaps change order or even flip with the AI version and the luck of the games, so wins come from many small gains rather than one decisive factor. Field points also come from completed cities, so cities are the foundation. When in doubt, prefer moves that complete your own cities and that claim fields touching completed cities.
Average points by source for the winner and loser in strong-AI games
Summary
Don't use up your meeples early. Having about 2 left at the end of the opening is about right.
Only claim roads that look like they'll end soon.
Start your own new cities and grow them. Winners score more in cities (field points also come from completed cities).
Keep farmers few. Late in the game, put spare meeples in fields that touch many completed cities.
Join your opponent's big cities and share them. Once you're comfortable, also look for moves that make them uncompletable.
Fill in around your monastery yourself to get the meeple back quickly.
Now put it into practice. Try the next-move quiz to see if you pick the same moves as the AI, and when you play the AI, use the hints and focus on one of these tips per game.
If this site helped you, you can support the author.☕Support on Ko-fi