Visual walkthrough — Determinant of 2×2 matrix
We only need one idea from before: a vector is an arrow from the origin to a point. We write the arrow to as the column : the top number is how far right, the bottom number is how far up.
Step 1 — Two arrows make a corner
WHAT: We plant two arrows at the origin, (blue) and (yellow).
WHY: The determinant is going to measure the area they enclose, so first we must actually draw the shape they enclose. Two arrows from a shared corner sketch out a parallelogram — that is the shape.
PICTURE: Look at the blue arrow leaning right, the yellow arrow leaning up-left, and the faint parallelogram they open up between them.

Step 2 — Trap the parallelogram in a box
WHAT: Draw a rectangle whose width reaches the farthest-right point and whose height reaches the highest point. Its width is and its height is , so its area is
WHY: We chose a rectangle because area of a rectangle = width × height, no cleverness required. Everything left over inside the rectangle will be triangles and small rectangles — also easy.
PICTURE: The green dashed box hugs both arrow-tips. The blue parallelogram sits inside it, and there are coloured off-cuts in the corners we will remove next.

Multiply the box area out (this is just expanded — each term is one word):
Step 3 — Name the off-cuts we must throw away
WHAT: Each off-cut piece has an area we can name:
- Triangle under : base , height → area . There are two of these (top and bottom), together .
- Triangle beside : base , height → area . Two of these, together .
- Corner rectangle: width , height → area . There are two of these, together .
WHY: We separate the box into "the parallelogram we want" plus "junk we can compute". If we can total the junk, then
PICTURE: Each off-cut is shaded a different colour and labelled with its own area. Count them: two blue triangles, two yellow triangles, two red corner rectangles.

Step 4 — Subtract the junk and watch it collapse
WHAT: Total junk area = the two blue triangles + two yellow triangles + two red rectangles:
Now subtract junk from the box area:
Cancel term by term — this is the whole magic:
WHY: The and pieces exactly cancel (the box under each arrow was never part of the parallelogram). The pieces don't fully cancel: we counted one inside the box but must remove two red rectangles, leaving . That surviving minus is the whole story.
PICTURE: The animation-style panel shows junk pieces sliding out of the box; what remains is exactly the blue parallelogram, stamped with its area .

Step 5 — Where did the sign go? (the flip case)
WHAT: Take , : (counterclockwise). Now swap them, : . Same parallelogram, area still , but orientation reversed.
WHY: Area itself can never be negative — that is why the parent note writes for actual area. The signed value carries one extra bit: which arrow is "first". This is exactly the orientation flip described in the parent's Example 3.
PICTURE: Two side-by-side parallelograms of equal size. On the left the corner labels run counterclockwise (green, ); on the right, after swapping the arrows, they run clockwise (red, ).

Step 6 — The degenerate case: area zero
WHAT: Let and . Notice . Then
WHY: With no area, the transformation squashes the plane onto a line — information is lost and no inverse exists. This is the parent's singular / linearly dependent case, and Cramer's rule (Cramer's rule) breaks precisely here (you'd divide by ).
PICTURE: Both arrows lie on one line; the "parallelogram" is a flat sliver with area label .

The one-picture summary
Everything above compressed: the green box , the red twist removed twice, the blue kept-area , giving — and the small sign badge showing (counterclockwise) vs (clockwise) vs (collapsed).

Recall Feynman retelling (say it to a friend)
I want the area of the slanted shape made by two arrows. Slanted shapes are annoying, so I trap the whole thing in an upright box — width is total-right, height is total-up, so the box area is . But the box is too big: it includes triangles under each arrow and two little corner rectangles. When I subtract those off-cuts, the pieces belonging under each arrow ( and ) cancel perfectly, and the corner rectangles leave behind a single . So the true area is . If I list my two arrows in the "wrong" rotational order, the same sum comes out negative — that minus sign is just remembering that I flipped the shape over. And if the two arrows sit on one straight line, there is no shape at all, so the answer is exactly . That is the determinant: kept-area minus twist, with a sign for flips and a zero for collapse.
Recall
Where does the come from in the box picture? ::: It is the part of the enclosing box that actually belongs to the parallelogram — the "kept" area. Why does only survive and not ? ::: The box already contained one ; we must remove two corner rectangles, so one is left over as a subtraction. What does a negative determinant mean geometrically? ::: Same area magnitude, but the two arrows are in reversed (clockwise) order — the shape is flipped over. What does look like? ::: The two arrows are collinear (parallel), so the parallelogram is squashed flat to a line with no area.
Connections
- Area of parallelogram — this whole page is its box-minus-junk proof.
- Linear independence — Step 6: zero area ⟺ dependent columns.
- Cramer's rule — fails exactly at the Step 6 degenerate case.
- Inverse of a 2×2 matrix — exists only when Step 6 does not happen.
- Transformations and scaling — the box→parallelogram map is the transformation.
- Cross product in 3D — the 3D big brother of this signed-area idea.
- Determinant of 2×2 matrix — parent topic.