I just read this story about "MASIS", an ever growing software project in our National Defence department? There are so many red flags with this that any competent I.T. manager would have killed it years ago. Yet, as it's within the realm of government, where there's a seemingly endless supply of [other people's] money, it goes on and on. The easiest way to understand what's wrong with this project is to first look back at the Gun Registry fiasco. The best article I've ever read about why the costs got so out of control can be found here. What many Canadians don't seem to realize with the the Gun Registry is that most of the hundreds of millions of dollars got spent on consulting hours to address the endless changes that government officials wanted in the software. As the founder of two software companies, one focused heavily on consulting, I am very familiar with the entire software development process. When the blueprint (specifications) of a software project is mapped out, there are a few golden rules:
- Try to acquire the resources to get it done in no more than 6 months and certainly never more than a year.
- Other than minor, MINOR changes, do not alter the blueprint whatsoever. In other words, stick to the plan!
Keeping these rules in mind, let's take a look at the MASIS project:
- It was started in 1997 as a $147 million undertaking.
- What started as a very focused effort to handle a single equipment category has mushroomed into a whole lot of other things.
- In 2003, Defence department officials said it would be done by 2006 at a cost of $325 million - 121% over budget.
- Now the software is projected not to be fully implemented until 2011 and officials have no idea what the total cost will be.
Am I the only one dumbfounded by these revelations? If you don't fully understand how outrageous they are, let me share with you some observations based on my years of experience as both a software developer and I.T. executive:
- The project was started in 1997. Eight years ago in software development terms is like the 1950's to the car industry. This project is like starting with the technology to build a 1958 Edsel and then midstream deciding that you really want a 2005 Ferrari and somehow changing around the original design to make it so.
- Trying to dramatically alter a software project along the way is like starting to build a house and then deciding you want to instead build a skyscraper, but not actual first demolishing the unfinished house and having to work around the limitations of its initial structure.
- After not too long, you end up with what's known in the industry as "band-aid code": So many "fixes" have been put into the software that it is never stable, has peculiar inconsistencies that are never fixed, and is generally avoided by all users unless they have to use it.
- The very fact that 2011 is being picked as a completion date is an absolute guarantee that:
- It'll never get finished.
- The price tag will probably double once again.
- Will never do what it's supposed to do.
What's most irritating about the very existence of this software project is that several off-the-shelf pieces already exist that probably do as much, if not more, than MASIS will ever do. While they're not cheap, they will certainly cost less than what has already been sunk into this project and THEY'LL WORK!
I wish that all federal political leaders would agree to pass a law that would forbid any more custom built software projects. Surely we've wasted enough money already?!? Or is it built into a politician's DNA to repeat the same mistakes over & over & over & . . . ?