Thursday, January 3, 2013

Classic Mistakes in Software Development and Maintenance

…the only difference between experienced and inexperienced developers is that the experienced ones realize when they’re making mistakes.
Jeff Atwood, Escaping from Gilligan’s Island

An important part of risk management, and responsible management at all, is making sure that you aren't doing anything obviously stupid. Steve McConnell’s list of Classic Mistakesis a place to start: a list of common basic mistakes in developing software and in managing development work, mistakes that are made so often, by so many people, that we all need to be aware of them.

McConnell originally created this list in 1996 for his book Rapid Development (still one of the best books on managing software development). The original list of 36 mistakes was updated in 2008 to a total of 42 common mistakes based on a survey of more than 500 developers and managers. The mistakes that have the highest impact, the mistakes that will most likely led to failure, are:

  1. Unrealistic expectations
  2. Weak personnel
  3. Overly optimistic schedules
  4. Wishful thinking
  5. Shortchanged QA
  6. Inadequate design
  7. Lack of project sponsorship
  8. Confusing estimates with targets
  9. Excessive multi-tasking
  10. Lack of user involvement

Most of the mistakes listed have not changed since 1996 (and were probably well known long before that). Either they’re fundamental, or as an industry we just aren't learning, or we don’t care. Or we can't find the time or secure a mandate to do things right, because of the relentless focus on short-term results:

Stakeholders won’t naturally take a long-term view: they tend to minimize the often extreme down-the-road headaches that result from the cutting of corners necessitated by the rush, rush, rush mentality. They’ll drive the car without ever changing the oil.
Peter Kretzman, Software development’s classic mistakes and the role of the CTO/CIO

The second most severe mistake that a development organization can make is to staff the team with weak personnel: hiring fast or cheap rather than holding out for people who have more experience and better skills, but who cost more. Although the impact of making this mistake is usually severe, it happens in only around half of projects – most companies aren't stupid enough to staff a development team with weak developers, at least not a big, high-profile project.

Classic Mistakes in Software Maintenance

But a lot of companies staff maintenance teams this way, with noobs and maybe a couple of burned out old-timers who are putting in their time and willing to deal with the demands of maintenance until they retire.

You get stuck in maintenance only if you are not good enough to work on new projects. After spending millions of dollars and many developer-years of effort on creating an application, the project is entrusted to the care of the lowest of the low. Crazy!
Pete McBreen, Software Craftsmanship

Capers Jones (Geriatric Issues of Ageing Software 2007, Estimating Software Costs 2008) has found that staffing a maintenance team with inexperienced people destroys productivity and is one of the worst practices that any organization can follow:

Worst Practice Effect on Productivity
Not identifying and cleaning up error-prone code – the 20% of code that contains 80% of bugs -50%
Code with embedded data and hard-coded variables – which contributes to “mass update” problems when this data changes -45%
Staffing maintenance teams with inexperienced people -40%
High complexity code that is hard to understand and change (often the same code that is error-prone) -30%
Lack of good tools for source code navigation and test coverage -28%
Inefficient or nonexistent change control methods -27%

Many of these mistakes are due to not recognizing and not dealing with basic code quality and technical debt issues, figuring out what code is causing you the most trouble and cleaning it up.

The rest are basic, obvious management issues. Keep the people who built the system and who understand it and know how and why it works working on it as long as you can. Make it worth their while to stay, give them meaningful things to work on, and make sure that they have good tools to work with. Find ways for them to work together efficiently, with each other, with other developers, with operations and with the customer.

These simple, stupid mistakes add up over time to huge costs, when you consider that maintenance makes up between 40% and 80% of total software costs. Like the classic mistakes in new development, mistakes in maintenance are obvious and fixable. We know we shouldn't do these things, we know what’s going to happen, and yet we keep doing them, over and again, and we're surprised when things fail. Why?


Anonymous said...

Hi Jim
Very nice post addressing a topic that is always hot in SW. I really liked the worst practice list and the percentage affection. Four out of six factors are related to the quality of source code, so I'd group them in a more generic worst practice... something like ... Not caring from day #1 of your project about / and continuously improve source code quality :)

windows 7 key said...

Awesome, my deal friend. I am just searching for some information on this and it occurs to me that I have just find your blog and read this article. Ever I want to write this kind of article but it takes time and mood to do, so many thanks to you for that you have wrote what I am thinking in my mind. Happy everyday bro.

Stuart said...

It is true that software development firms often make such classic mistakes, time and again. It takes proper planning, estimation of duration and adequate budget allocation to deliver a successful project.

Anonymous said...

Hi Jim, I would also add the fact that even if everyone is aware of the classic mistakes, not everyone has enough experience and knowledge to realize that the project is doing one of the mistakes.
CTOs, COOs or product owners can think that what they ask is achievable and not wishful thinking even if the developers tell them the contrary. Their reaction would be in that case that the developers saying that are not good enough, and that the hiring was wrong.

Site Meter