Nosok & OneMan
Nosok
I’ve been working on a model to predict the optimal moves in Tetris if you could see the next pieces—think of it as a perfect play algorithm. What would be your strategy for minimizing line clears in a long run?
OneMan
Keep a single column wall on the left edge and never let any piece touch the bottom except that wall. Drop every I piece vertically in the wall so it just adds height. Use J, L, T, S, Z to fill gaps in that wall or leave a hole that can never line up. Keep the right side empty. If a piece would complete a line, rotate it so it stacks on top of the wall instead of across the bottom. In the long run the stack stays low and no lines clear, so your count stays at zero.
Nosok
That sounds like a very efficient way to keep the score flat, but in practice the random piece stream will keep trying to jam you into a corner. The wall strategy works for a few moves, but as soon as an O or a T lands against it you’ll have to break the wall anyway. Also, keeping the right side empty is a good way to lose all the useful Tetris space, so eventually you’re just stacking a wall of I’s. I’d suggest allowing a small buffer zone in the middle to absorb those unavoidable shapes—then you can still keep the stack low without sacrificing the entire playfield.
OneMan
Add a buffer strip two cells wide in the middle. Use the left wall for I’s, right side stay clear, and place O, T, S, Z in the buffer so they don’t touch the walls. Rotate them to fit without completing a line, and when a shape would force a line clear, drop it into the buffer’s lower half. Keep the buffer shallow enough that the stack never rises past the top; when the stream forces a corner you can still flip the buffer into a new orientation and continue. This way you keep the field low and avoid sacrificing the whole playfield.
Nosok
That’s a clever tweak—adding a 2‑cell buffer is like building a moat around your wall. The only thing that worries me is the buffer’s “new orientation” rule; if you keep flipping it you’ll start to create a pattern that the RNG could exploit. Also, you still need to handle the occasional S or Z that will fit exactly into the buffer’s bottom half and accidentally clear a line, even with your rotation rules. In theory, though, the strategy keeps the stack low and you avoid the brutal line clears, so I’ll give you a 3.2 out of 5 for the idea and a 0.8 for the execution risk.
OneMan
Got it, I’ll shrink the buffer to one cell and lock any S or Z that would finish a line before they hit the floor. The only thing that matters is that the stack never rises above the bottom three rows; if a shape threatens that, I’ll force a rotation that keeps it in place. That keeps the RNG from finding a repeatable pattern. Simple, efficient, no teamwork needed.