The determinant of a matrix is the signed volume of the box its columns span — and it is an invariant in two deep senses. It is unchanged by a shear (adding a multiple of one row to another leaves it fixed: a slanted box has the same volume as an upright one of equal base and height), and it is unchanged by a change of basis (similar matrices P⁻¹AP have equal determinants). That second invariance is why the determinant is a property of the linear map, not of the particular coordinates you wrote it in. It measures how a transformation scales volume — and the transformations that don't scale volume are exactly the ones it cannot see.
A 2×2 determinant, then the same matrix after a shear (row₁ += 2·row₂). The demo re-measures — the signed volume is unchanged: live demo
“Shearing a shape changes its size.” — a shear slides the top of the box sideways without changing its base or height, so the volume — the determinant — is exactly preserved. The determinant is built to be blind to shears. cited
Adding one row to another slides the box but keeps its volume; changing coordinates renames the map but keeps its determinant. Two invariances, one number — a property of the map, not the grid. Cauchy 1812
On the canonical compiler, det[[2,1],[1,3]] = 5; after the shear row₁ += 2·row₂ the matrix is [[4,7],[1,3]] and the determinant is 5 again — exactly: