extreme vs rad

if I have an option I use a mixture of procedural and object oriented instead of either.
if I have to work with one I will use the Extreme model, as it allows for better control over code quality.

the rad model is bad code.

the extreme model is better code, but is targeted at custom one off apps. with the ides of adding functionality the customer wanted at a later date.
great business model, get base functionality with good code, add desired features afterwards as they become available. [ screams for a plugin driven app doesn't it? ]

if it's a company owned app for wide spread distribution, my preffered method of defining the objective and taking as long as needed to get desired functionality with good code seems best.
better resulting app with fewer update releases to add desired functions. this allows the app's next release to be worked on with fewer coding breaks to finish functionality that should have been in first release, which keeps the app having better code and fewer bugs.

Never agreed with writing code I know won't work, that's not RAD, it's a stupid waste of time. Scaling is a problem I've seen a few times, but that to me is a design error, not a model one. Every time I've seen it, it should have been no quicker to do it wrong.
Never worked in an extreme environment but from the sound of it I'm probably not far off, except for the refactoring whenever possible bit. I do refactor existing code when it's broken (usually too slow). Refactoring for other reasons, adherrence to new standards, readability, maintainability I will always try to do, but usually in any unused contingency. Working in 'live' environments most of the time, the if it ain't broke don't fix it syndrome is quite well entrenched.
As for better models, that depends on the project, the budget and the environment, not the model. Sometimes I have to use Waterfall simply because the fully documented design is a requirement. I can see it coming back with offshoring as well.
The only time I can see a use for pure RAD is when quick and dirty is a requirement. I've seen this done a few times but as far as I'm concerned it's a short term fix and a long term problem until it's done properly. I get a good deal of work on the latter front.

I'm working in a collection of apps.
and, other than base libs for the language.
I'm not using a single lib where I didn't code every line.

by the time the collection is ready for release, it will take a dvd to hold the apps.

unless it's an open source lib, where the code can be edited to remove functionality not needed, or add functionality that is.

I worked with both models in different companies and XP is the best for most projects. Although each project is different, the XP approach gave me the best results. It is usually a very consistent process that gives you control and flexibility.

One very interesting aspect of XP is that when properly used it could put you on par with CMMI certified companies with the advantage of being able to react quickly.

But there is a breaking point. When the project is big enough, XP and RAD just don't work. At least in my experience.