What's the meaning of "permuting feature values"?

Here’s the link to how permutation is done on feature values.

Intent is, should a feature be unimportant, changing its values within its domain shouldn’t have considerable impact on the target score.

Say you have a employee_id column and in your dataset and the rows are in ascending order of employee ids. Now, should you swap values of employee_id column alone (without moving rows around), you’ll arrive at a new bunch of rows where the employee_id column is permuted.

Here’s an example

Assume that the sample data below is the original data:

employee_id,age,weight,plays_sports
1,30,150,1
2,40,160,0
3,50,170,1
4,60,120,0

One of the permutations of theemployee_id column would result in

employee_id,age,weight,plays_sports
4,30,150,1
1,40,160,0
3,50,170,1
2,60,120,0