Somewhere, in one of the Usability books I read ( probably Donald Norman ), there was a suggestion that the normal format for entering credit cards was wrong. The usual format is 16 characters in a long string. However the way that we read then and they are printed on the card is 4 groups of 4. The argument is - and I completely agree with it - that entering the card in 4 lots of 4 would be much easier, both to enter and to verify for the user.
This only needs a little bit of coding to include the dashes in the entered data, and ignore them in the validation. Or to have 4 boxes with focus going to the next one. It is not a complex coding issue. So why does no-one seem to do it? Why is the standard box format so prevelant still?
This comes out of having just tried to use my credit card, and make a mistake, and then had to search through the whole lot to identify the mistakes - tedious, and something that would be much easier if it was properly divided.
Edit: To add some points from comments. I think the proper implementation would need to be worked out - 4 boxes of autoinserted dashes may not work, or may need to be tweaked. So it is not saying xxx is the right approach, but that the current approach does have its own set of problems.