There are others who attempt to expedite the steps by providing automated tools for various tasks in the process, but there are certain human endeavors, such as listening to end-users discuss their needs for an information system, that just don't lend themselves to automation. Even in those areas that did lend themselves to automation, such as software generation, most of the products produced in this area used tomorrow's technology to produce yesterday's applications. Quite a number of years ago, at a conference on this very subject, I was told that truly effective computer-aided software engineering tools are the technology of the future and they always will be. For the front-end phases of applications development such as planning, requirements, and design, that is still true today.

So, if you can't eliminate steps and you can't expedite steps, how do you speed up the process on large development projects? The simple answer is this: don't do large development projects! Oh, you can still eat the whole elephant, just stop trying to take it all in one bite, or as your mother used to tell you, "Take small bites and chew!" Just because you have grand plans doesn't mean that you have to lock up a herd of developers for years grinding away at mountains of software until the day comes where you can throw the big switch on the entire pile and see if it works or not. Break it up into pieces, ease one piece at a time into production, and build on your achievements over time. Not only will you have a better product by the time it's completely implemented, you will deliver something much faster, get better feedback, significantly reduce your risks, and reduce the total overall cost in both time and funds compared to producing the same software using the big bang approach.

Start with a vision, not a plan
Successful incremental software development does not require a detailed plan from start to finish. Such a plan, in fact, is contrary to the entire methodology. What it does require, though, is a vision, or a concept, of the ultimate goal state of the application. Everyone involved needs to understand this vision, so that each phased implementation is another deliberate step towards that ultimate goal. Developing software in a rapidly changing, imperfect world means that things will be different over time, so there is no point in laying out specific details for the days and weeks several months out. What is important is that there be a clear and well-understood vision of where things are headed, along with how today's project fits into the overall picture. Twelve two-month projects designed to achieve a clear vision will accomplish far more than any single 24-month project with the same aim. Every incremental deployment during the former approach provides valuable feedback and experience that is simply not available to the latter. It is the vision that guides the process, not a detailed plan.

Make yours a vision of excellence
According to Phillip Crosby, author of a number of books on the subject, "Quality is Free." The basic tenant underlying this "pay me now or pay me later" theory says that you might as well plan to build the best system that you can, because you're ultimately going to face the exact same costs, whether by design or lack thereof. By embracing "best practices" from the onset, you can dictate where and how these expenses will occur as opposed to moving forward with ill-fated strategies that will end up costing you as much or more when you least expect it. In today's environment, the highest quality information systems applications built with state of the practice methods and techniques share the following desirable characteristics: