100th Blog Post: What I know about Software Development and Crisis Management

I started writing this blog post about this…

Opinions on policy and politics aside, this article on the struggles of healthcare.gov tells a classic tale of large software development project failures, and how not to react when trying to solve the issues.

But as I continued to write my thoughts, this became more about my views on software development and crisis management. So, enjoy (or ignore, or comment).

Hopefully, it is worthy of my 100th blog post (which this is).

On software development

  • Software development, no matter how much research is done on development methodology effectiveness, is as much art as it is science. Treat the artists like factory workers, and you get what you deserve.
  • Not all software developers are great at developing all applications in all languages, using all tools. Web development, especially when connecting to legacy systems, is a specific discipline. The best Web developers don’t work at large contract system development houses, in my experience. That’s usually because there are more interesting projects in the world for them to work on.
  • In software, one of the most common reasons for failure is complexity. For large projects like the one cited in the article, you have to work very hard to come up with an architecture that makes concepts and aspects of the application simple to extend and use. This is not cheap—it takes time and significant energy from artists to achieve. But, when it is achieved, it is a beautiful thing. You owe it to yourself to experience this at least once in your career.
  • Never ignore or otherwise skimp on quality User Experience Design (UXD). Never.
  • Designs change. New needs and unidentified technical risks are discovered. Requirements evolve. But if you want to get something done on time, quit changing the (in scope) requirements. If a good design forces you to re-write significant parts of your requirements, that is a good sign that you didn’t write effective requirements. In fact, they may not be true requirements at all. And, BTW, a list of features from a competitor’s brochure are not requirements. If you don’t how to write effective, (mostly) design-neutral requirements, learn. Start here. Then read this.
  • Well-written use cases—that everyone on the team has read and understands—solve a lot of problems. So does identifying the operating environment and performance goals sooner rather than later. Tip: Print this stuff out and hang on the walls were the developers and testers sit. Or (if you want to save some trees) make a JPEG of it and ask everyone to set it as their desktop wallpaper for the duration of the project.
  • Make developers create unit and integration tests for their code. Have them do code reviews with peers. Make them use code comments and consistently format their code. Make them fix their own bugs. Hold them accountable to the quality of software build they provide to the test team. Make them perform design checkpoints with their peers (or even customers) for any significant component. Anything less and they are not a serious professional developer.
  • Developing for security and performance are specific skill sets, but these are most often “want to” issues. As in, if a talented developer truly wants their product to be secure and fast, they will figure out how to make it happen.
  • Do daily stand ups. Make the product manager show up.
  • A good tester is as valuable as a good software developer. So is a good technical documentation person. They are artists too.
  • Give them all good, fast computers with at least two monitors and reliable Internet access. You want something done faster, remove inexpensive barriers.
  • Don’t make them pay for coffee. And don’t buy the cheap stuff.

On crisis management (in IT)…

  • The comment in the article about the “war room” is spot-on. I call it “the body cannon”. When **it hits the fan, there is usually some executive wanting to pull anyone and everyone off their current work and throw them at the problem, with no real plan. They are simply hoping that through volume, that the problem will be solved faster, when in fact it usually has the opposite affect. The better approach: select a small team of talented, trusted artists (that know the code!) and simply ask them: “What do you need?”, and then get it for them ASAP. Then, be prepared to pay the price of pulling these people off their current project (usually in the form of a schedule slip), once the crisis is over. These things are always about choices, not decisions.
  • Yelling at software developers doesn’t work. People that choose a career creating art by typing all day with headphones on are not the type that react well to yelling. If you don’t know how to convey a sense of urgency without yelling, the problem is you. If you don’t understand the problem or software really well, get out of the way.
  • If you have hired talented people, they can become heroes. Let them.

And, finally, if you are making an application for use in healthcare, take it seriously. Lives are at stake. It can still be fun and rewarding, but the problems within healthcare are large and demand our best efforts all the time. Now, go be great.

3 thoughts on “100th Blog Post: What I know about Software Development and Crisis Management

    • Hi Peter, Thanks for the comment. I did not list the items in order of importance–I just wrote them as I thought of them. I often cite the importance of UXD, but recall that not all applications have much of a human-computer interface. Some applications perform a huge number of complex transactions (where scalability, security and performance are key) with no human interacting with them for weeks or months.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s