Convergent Thinking

The phrase means blank slate, but the object was stone.

Some problems get easier as you go, and others can’t start until they’re already solved.

What separates training methods that scale from those that don’t?

Round-to-nearest makes the same rounding error every time.

People optimize for local objectives and lose sight of global ones.

“Attention normalizes the wrong norm” showed that attention output variance collapses as context grows.

The hardest box to escape is the one you cannot see.

We treat attention as a solved primitive.

When people say “X is the new oil,” they mean the bottleneck.