No Fear

Computerworld - Last week, at a conference at IBM’s Almaden Research Center in Silicon Valley, the topic was complexity. The outlook for software development: grim. The message: Even as IT environments keep getting more complex, we keep repeating the same old mistakes, such as buffer overflows, and we’re loath to use modern development tools that would automate those errors out of existence. “The problem,” said software risk guru Peter Neumann, “is that we keep going through the same problems over and over and over again.”

He’s right, of course. What’s worse, we all know he’s right — and we know we’re not inclined to do much about it.

And we all know why, don’t we? We’re afraid.

We don’t use advanced technology to improve software development because we’re afraid of programmers. Mainly, we’re afraid that programmers simply won’t use the hot new technology and that the money we spend on it will be wasted.

That’s a perfectly logical fear. It’s based on experience. We’ve seen the same story played out for decades — with CASE, with methodologies, with code generators, with reuse repositories. Either programmers don’t use the tools at all, or they use them half-heartedly, or they find reasons to subvert them, or programmer productivity drops through the floor. Programmers just won’t accept the new tools.

That’s logical, too. Programmers know from experience that new development fads come along about as often as streetcars, and they’re all based on somebody’s theory of how best to build software. But programmers won’t waste their time trying a new approach unless they really believe it’s better than what they’re already using. And they won’t believe it’s better until they’ve given it an honest, committed try. But they’re not going to waste their time doing that unless they already believe. ...

That’s why fancy new development tools tend to become shelfware and software development quickly returns to the tried and true: pushing one line of code next to another.

So, what to do? Are we forever doomed to watch helpless as greater business demands and ever-changing software requirements suck us down into a swamp of complexity?

Nope. We can stop trying to shove new development technology down the throats of programmers and start using it ourselves.

Look, there are some great tools out there calling themselves “application portfolio management.” That sounds like something you’d use a spreadsheet for, but what these tools really do is automatically comb through code, generating metrics, spotting inconsistencies, hunting down problems and generally analyzing everything in your collection of software.

Automated code inspection isn’t new. But these tools are designed to let managers roll up the metrics into a big picture or drill down to see which projects are working and which have trouble brewing — and how they’ll interact with other parts of the software portfolio.