"Most economical," maybe, but that doesn't make it good.
Frequently it does. The application that works is more "good" than the application that doesn't work. The application that exists genrerally works, while the application that doesn't exist does not. Thus the application that has been written is more "good" than the one that hasn't been written.
If I can cut the development time in half by using bloated code and thus be able to write two applications instead of one, then the second one that being economical made possible is pretty much always going to be better than it would have been if we had not gone for economy.
Much of my code at work is bloated. My team puts very little effort into optimizing on first run of applications. Why? Because the applications requested have 20% chance of never actually being used. Sad, but true. Many of the applications are requested so that the person requesting them can look like they are doing work. We still have to write them, but they won't get used.
The other factor is that my and my teams time is worth more than the cost of buying faster computers. Add to that that most speed slowdowns are on the user side, not the computer side. Things like waiting for input. Finally, the rate that new applications that get heavily used end up with so many change requests in the first year that trying to heavily optimize the code would mean that by the time the optimization has been done, it is no longer needed.
That doesn't mean that we never optimize. We put our optimization resource to the places that will get the most bang for the buck. At optimization time, we look first at applications that cannot scale well to meet demand. Then we look at applications that are heavily used. Then we look at applications that are just slow.
An application that takes the unreasonable time of 60 seconds to save a document is not going to be high on our list if it is only used twice a year for a semi-annual recording. You can say that the application isn't "good", but it is WAY more "Good" than if it didn't exist at all, which would be more likely if heavy optimization were the requirement.
