Tetris

Most lost games of Tetris do not end on a brilliant miss. The stack gets jagged, the column you were saving closes, and one panic placement creates two more bad choices behind it. The 80/20 of an early game is the shape you leave behind, not how fast you can drop.

Give unequal attention to four visible things: a mostly flat top, one open column for the long piece, no covered holes, and the first mistake after a loss. If those are clear, the next piece has a home. If they are not, a faster drop only finishes the game sooner. Speed, skins, and T-spins can wait.

The long piece is not hiding

Modern guideline Tetris deals pieces in bags of seven: one of each shape, shuffled, then another bag. You can have a dry spell. You cannot have a conspiracy. The longest wait for a given piece is twelve drops (the guideline randomizer).

That fact does not cover the 1989 NES cartridge, which used a harsher randomizer and has no hold. In a guideline game, a late I-piece usually means the well was already compromised. The piece was coming. The stack was not ready for it.

Leave one column, and do not bury a cell

A well is one empty column kept open so a vertical I-piece can clear four lines. On a 10-column board, that reserved column is a tenth of the width, and it is the lane for the clear that actually changes the score. Most players keep it on the far left or far right because a side well is easier to protect than a gap in the middle. Pick one side and stay with it. If you have to choose between a slightly ugly placement and closing the well, take the ugly placement somewhere else.

S and Z pieces are where beginners bury cells. They cannot both sit flush against the same kind of edge. Set one against the wrong surface and you get a one-cell hole that no later piece fills cleanly. Surface bumps can be fixed. A covered hole is a debt: you may have to clear two or three lines before that cell is useful again.

Before you lock a piece, ask four short questions. Will this cover an empty cell? Will this close the well? Will this leave a notch only one shape can fix? Is there a flatter seat that scores less now and keeps the board alive?

Read the queue, and spend hold on a real problem

The next pieces are part of the decision, not a decoration beside the board. If a line piece is coming soon, the open column is a plan. If it is not in the queue, filling the well because you are bored creates the problem you will blame on luck later.

Hold, where the game has it, is insurance. Park the piece that truly does not fit. Store an I-piece while you finish the rows beside the well. Delay an S or Z that would make an overhang right now. Do not park a piece only because a prettier one might arrive, and do not swap so often that you stop reading the queue. A held piece you never use is a hole somewhere else on the board.

One Tetris is worth more than four tidy singles

In classic NES scoring at level 0, a single is 40 points, a double 100, a triple 300, and a Tetris 1,200 (Tetris scoring). Four singles, the same four lines, are 160 points. The four-line clear is 1,200. Modern guideline games add back-to-back bonuses, combos, and T-spins, and the tables differ by version, but the tilt remains: a planned four-line clear is paid more than constant small cleanup.

Beginners do not need the T-spin yet. They need to stop turning a half-built Tetris into two singles because the well closed. For one session, tally singles, doubles, triples, and Tetrises. If singles dominate, ask why. Did the well close? Did height feel scary before it was actually dangerous?

Burn lines when the plan has gone stale

There is a height where waiting for the I-piece stops being discipline. If the stack is near the top, the well is blocked, and the queue does not show a quick repair, clear whatever lines you can. A tall clean stack with an open well can still be played. A medium stack with holes, overhangs, and no well is already in trouble. A double that buys air beats a theoretical Tetris that never arrives.

Near the top, read the current piece, the queue, and the landing surface before you press hard. If a placement needs a last-second spin, the stack has stopped offering easy seats. A slower decision can save several future ones.

Name the first hole

When you top out, do not start the next game immediately. Look for the first hole, not the final mess. Maybe you covered the well. Maybe an S or Z tucked under a ledge. Maybe you ignored a piece the queue had already shown you. The last twenty seconds usually explain less than that earlier cell.

Play three games with one added rule: you may not cover the well, and after each loss you name that first hole out loud. Then use a short session instead of grinding on autopilot. Three minutes of flat stacking with no covered holes. Three minutes of keeping the same side well open. Two minutes of pausing before every S and Z. Two minutes of naming the first bad placement from the last loss.

8020 move: For the next ten games, choose one habit only: keep the well open, avoid covered holes, or use hold less. Add a second habit after those ten, not before.

What can wait

A T-spin video is a fine pleasure after you can clear a line on purpose. A new skin does not change a covered well. Restarting the moment the board looks ugly skips the only information the loss was offering. Two misreads are worth retiring: that the randomizer hid the I-piece, and that the piece which finally touched the top was the real mistake.

Reading a whole board, not just the falling piece, is the same pleasure in slower form on chess. A mistimed jump in Mario punishes you the way a covered cell does here: the error is small, and the next few seconds are already worse.

Where this stays honest

• Guideline randomizer: bags of seven, one of each piece, documented at Tetris Wiki. Twelve pieces is the long drought. NES Tetris uses a different randomizer and has no hold.
• NES level-0 points (40, 100, 300, 1,200) are the classic table. Later levels multiply them. Modern games pay differently.
• A flat stack, an open well, and naming the first hole are practice, not a laboratory result.