블록을 돌릴 수 없다는 점
이 게임에는 회전이 없다. 모양을 판에 맞추는 것이 아니라 판을 모양에 맞춰 미리 비워 두어야 한다. 세로로 긴 다섯 칸짜리가 나왔는데 세로로 다섯 칸 빈 곳이 없으면 그 블록은 놓을 데가 없다.
세 개가 한 묶음이라는 것
세 개를 다 놓아야 다음 묶음이 오므로, 첫 블록을 놓을 때 이미 세 번째 블록이 들어갈 자리를 생각해야 한다. 판은 점수가 아니라 놓을 자리가 없을 때 끝난다.
직접 돌려 본 숫자: 가장자리 팁은 맞았다
이 페이지의 뽑기 규칙과 점수 계산을 그대로 두고, 놓는 방식만 다른 프로그램에게 판을 맡겼다.
- 아무 데나 놓기: 평균 85점, 블록 20.3개 (1,000판)
- 줄을 지울 수 있으면 지우고 나머지는 아무 데나: 106점 (1,000판)
- 줄 지우기 먼저, 그다음 가운데부터: 216점 (1,000판)
- 줄 지우기 먼저, 그다음 가장자리부터: 334점, 블록 53.6개 (1,000판)
- 놓은 뒤에도 받을 수 있는 모양 수 지키기: 390점 (120판)
줄 지우기만 챙기면 아무 데나 놓는 것과 차이가 작다. 판 대부분은 지울 줄이 없는 수이고, 점수는 그때 어디에 놓느냐에서 갈린다. 가장 좋았던 방식은 놓은 뒤에도 스물한 가지 모양 가운데 몇 가지가 판에 들어갈 수 있는지 세어 그 수를 지키는 것이었다. 빈칸이 몇 개냐보다 어떤 모양으로 남았느냐가 중요하다는 뜻이다.
어떤 블록이 자주 나오나
한 개를 뽑을 때 세 칸짜리가 31.2%, 네 칸 26.0%, 다섯 칸 24.9%, 3×3은 2.3%다. 한 묶음에 3×3이 끼는 것은 열네 묶음에 한 번꼴이지만, 다섯 칸 일자 막대는 네 묶음에 한 번꼴(22.9%)이다. 가로와 세로로 다섯 칸 빈 줄을 하나씩 남겨 두면 판이 자주 산다.
점수와 오늘의 도전
놓은 칸 수만큼 점수가 붙고, 줄을 지우면 지운 줄 수의 제곱에 10을 곱한 만큼(한 줄 10, 두 줄 40, 세 줄 90) 더 붙는다. 여러 줄 지우기가 연달아 이어지면 10점, 20점씩 더 얹힌다. 그래도 실험에서 점수를 가른 것은 보너스보다 버틴 길이였다. 오늘의 도전은 모두 같은 블록 순서를 받으므로 점수 차이가 곧 순서를 쓴 솜씨의 차이다.
더 읽을거리
2048 모서리 전략은 어디까지 통할까: 한쪽으로 몰아 두는 전략을 다른 퍼즐에서 돌려 본 결과.
There is no rotation here
Pieces cannot be turned. Anyone who grew up nudging falling shapes into place stumbles on that once. You are not fitting a shape into the board; you are keeping the board shaped so the next piece fits. If a five cell vertical bar arrives and nowhere on the grid has five free cells stacked vertically, that piece simply has no home. Laying it down sideways is not an option.
Pieces come three at a time
The tray hands you three shapes, and the next three only appear once all three are placed. That means the first placement already has to account for the third. Will the other two still have somewhere to go after this move? Skip that question and you will place two comfortably and then jam on the last one. Remember that a round ends when nothing can be placed, not when a score runs out, and the arithmetic becomes obvious.
Manage the shape of the empty space
How many cells remain matters less than people expect. Twenty singles scattered across the grid can host nothing, while nine cells gathered in a square can host almost anything. Strong play therefore chooses the move that keeps the board able to accept any shape, rather than the move that scores most right now. Clearing a line is less a goal in itself than a way of restoring that flexibility.
Clearing more than one line at once pays a bonus
Placing a piece so that a row and a column complete together earns a bonus, which creates a real choice: you can see a line ready to clear and still hold off for one more placement. The board keeps filling while you wait, though, and losing your placement space while fishing for a bonus is the most common way these rounds end.
The daily board hands out the same piece order
A new puzzle opens at midnight and everyone playing that day receives the same sequence of pieces. With luck removed from the comparison, a difference in score is a difference in how the sequence was used. One attempt per day, and playing on consecutive days builds a streak.
We ran the numbers: the edge tip holds up
Using this page's own piece draw and scoring, we handed the board to five simple programs that differ only in where they place pieces. None of them knows the next tray; each scans every legal spot and picks by a single rule.
- Anywhere at random (1,000 games): average 85 points, 20.3 pieces placed, 0.9 lines cleared
- Clear a line if possible, otherwise random (1,000 games): 106 points, 22.8 pieces, 2.4 lines
- Clear first, then fill from the centre (1,000 games): 216 points, 37.0 pieces, 7.2 lines
- Clear first, then fill from the edges (1,000 games): 334 points, 53.6 pieces, 13.1 lines
- Keep as many shapes placeable as possible (120 games): 390 points, 60.9 pieces, 15.6 lines
Grabbing clears alone barely beats random play (106 versus 85). Most moves have no line to clear, and the gap opens on those moves. Filling from the edges scored about four times random play and about one and a half times filling from the centre. The best rule counted how many of the twenty one shapes could still fit after each placement and chose the spot that lost the fewest, which is the empty space management above turned into a number.
How often each piece appears
There are twenty one shapes, weighted so the 3×3 square and five cell pieces show up less. Per draw: three cells 31.2%, four cells 26.0%, five cells 24.9%, two cells 10.4%, one cell 5.2%, and the 3×3 square 2.3%. A tray of three holds a 3×3 only 6.9% of the time, about once in fourteen trays, but holds a five cell bar (either direction) 22.9% of the time, about once in four. With no rotation, keeping one free row of five and one free column of five saves the board more often than you would think.
How scoring works
Placing a piece scores its cell count. Clearing lines adds ten times the square of the number of lines: 10 for one, 40 for two, 90 for three. When multi line clears come on back to back placements, each adds ten times the length of the run (10 for the first, 20 for the second), and a single clear or no clear resets the run. Even so, what separated scores most in the experiment was survival rather than bonuses: the edge rule placed 2.6 times as many pieces as random play.
Further reading
How far the 2048 corner strategy goes: the same keep it to one side idea, played out in another puzzle.
