Should We Re-Develop Our Entire Application?

Apple'€™s Christmas ads to "Give up on Vista"€ once again got me thinking about whether a company with a mature commercial product has a greater chance of success by:

• Re-developing its entire product

        or

• Re-developing components of the product in a piecemeal fashion

The opportunity to re-build an application under new, modern, technology conditions is seductive.  It'€™s the chance to build the best application everyone can imagine.  The technical staff will make better architectural decisions, eliminating those too-expensive-to-fix bugs and making future enhancements easier.  The UI will be modern and useful.  The code will be clean and supportable.  The product will be better product able to reach further in the market because it can integrate and interact with other programs and technologies built on the newer platforms.  Marketing will love the new product. They can put "€œNew & Improved"€ on all the packaging; they can show off the new interface in pictures.  The entire company will be proud to have a product far better than the competitors'€™ have. 

Meanwhile, modular re-development is limiting.  Developers can'€™t necessarily choose the latest and greatest technologies.  They have to work within the existing architectures and existing limitations of the product.  Most of the current bugs will remain bugs.  Users will be pleased with the improvements but still be frustrated with the overall product.  Marketing will want to know why more couldn'€™t be accomplished.  Everyone in the company will want the technical staff to "€œdo more".

In my experience, wholesale re-development is often a dream that turns into a nightmare as the re-development gets bogged down under the sheer weight of (and the mention inherent in) implementing all of the infrastructure, features, and enhancements of the mature product.  The project gets off schedule and eventually the company is forced to release something that isn'€™t nearly as good as the previous product. 

The reality of modular re-development is accepting incremental updates over time, eventually replacing the entire product so that little of it is brand new and little of it is incredibly old.  (Similar to the maintenance of an aircraft.)  Developers may feel frustrated and disappointed at the start of the project (rather than mid-way through). But, the company's risk is significantly limited.  With a piecemeal approach to re-writes there are fewer dependencies on unknown, un-built parts of the application and fewer moving parts.  The new code will be written.  It will be fine-tuned.  It will be integrated or sometimes grafted into places. (The Bionic Man vs. Frankenstein.)  Even if the worst-case happens and the module gets off schedule and the company forces the release, at least only a part of the application is imperfect rather than the entire product.

Neither of these options seems too encouraging. To be fair, technical development is always a challenge.  That'€™s why we make the big bucks. Good development is about making smart, sometimes daring and sometimes unpopular decisions. Don't let a a re-development effort provide your competitor with the perfect opportunity to steal your market. 

I suspect Apple is feeling pretty good about itself right now.  Certainly all its shareholders are.