Whether scaling is useful for a regression model?

Linear Regression and scaling of data

I find contradiction in two answers to question-30302 regarding effect of scaling.The scaling essentially should help us generate better measures? How can we justify its use ? One of the answers seems to utilize wrong/incorrect method of the scaling the data? Another answer does not disclose the scaling approach.