Changing a 2D array variable changes the original 2D array variable as well. Why?

Why is matrix change as well? Go variables are passed or copied by value, not reference unless explicitly specified using pointers vs values. Is there something that I don't understand?

import "fmt"

func main() {
	matrix := [][]int{
		{100, 20, 30, 10, 11},
		{15, 100, 16, 4, 2},
		{3, 5, 100, 2, 4},
		{19, 6, 18, 100, 3},
		{16, 4, 7, 16, 100},
	}
	var matrix2 [][]int
	matrix2 = matrix
	matrix2[0][1] = 11111111
	fmt.Println(matrix)
	fmt.Println(matrix2)

	// Output :
	[[100 11111111 30 10 11] [15 100 16 4 2] [3 5 100 2 4] [19 6 18 100 3] [16 4 7 16 100]]
	[[100 11111111 30 10 11] [15 100 16 4 2] [3 5 100 2 4] [19 6 18 100 3] [16 4 7 16 100]]
}

Go slices are references to an underlying array. A slice is like a pointer to a value, not like a direct value. If you use arrays instead of slices, you will see different behavior:

package main
import "fmt"

func main() {
    matrix := [5][5]int{
        {100, 20, 30, 10, 11},
        {15, 100, 16, 4, 2},
        {3, 5, 100, 2, 4},
        {19, 6, 18, 100, 3},
        {16, 4, 7, 16, 100},
    }
    var matrix2 [5][5]int
    matrix2 = matrix
    matrix2[0][1] = 11111111
    fmt.Println(matrix)
    fmt.Println(matrix2)
}

Output:

[[100 20 30 10 11] [15 100 16 4 2] [3 5 100 2 4] [19 6 18 100 3] [16 4 7 16 100]]
[[100 11111111 30 10 11] [15 100 16 4 2] [3 5 100 2 4] [19 6 18 100 3] [16 4 7 16 100]]

You can see that because I used an array [5][5]int of arrays [5]int instead of a slice [][]int of slices []int, the assignment matrix2 = matrix made a copy of the entire matrix.

If you want to use a variable-size matrix, you will have to write your own function to create matrix copies. The above code only works if you know the size of your matrix at compile-time (e.g. 5x5).