The theory is that it would be possible to estimate the amount of work needed to get a Windows app to work on Wine by calculating the number of API's that is not working or only half working and estimating the time it would take to correct these API's.
This would maybe work for all 'old-style' Win32 APIs... But as more and more new APIs are exported via COM, it's starting to get impossible to get 'static' counting of what is used.
Lionel