At the Embedded Systems Conference in Chicago, I took a short class in "Top Gun" management practices by Numetrics, a method in squeezing the best performance out of a team of engineers.
I was impressed with the statistics they gave. They said that 60% of projects worked on, in a traditional sense, slipped the planned deadlines by at least one quarter. But, the so called Top Gun method reduced that problem by 50%. So, now only 30% of projects slip, I suppose.
Here are the 8 points of this method:
1. Compute project complexity statistically.
2. Estimate resources and schedule based on models.
3. Rigorously perform "what if" scenarios for schedule/resource optimization.
4. Benchmark project execution assumptions.
5. Set the most aggressive, yet achievable, targets.
6. Quantitatively asses schedule/resource impact of each feature request.
7. Diagnose root causes of poor performance.
8. Foresee resource shortfalls across the pipeline.
I see that this can be broken down to two steps. First, requirement analysis planning. A typical Mil Spec design practice. And second, remove issues that bring about underperforming engineers.
Planning is the most important step. A few times in the past I was part of a software design project, some where I was in control, and others I was not. In the ones I was not in control, it ended up being a design on the fly. Code was generated from just sitting down and writing it out. Convoluted code was always made, and problems followed. The schedule was slipped. Where I was in control, I spent more time working on the requirements and flow charts than actually coding. The code was just generated based on my diagrams. And what happened? I made the deadline. But all along the way, management was angry with the time I spent on planning. There was a meeting once where I was told not to plan anymore. I didn't listen, and I delivered. The aftermath of that ordeal is a story for another time. But what you can gather from this is that the planning stage has been, for the most part, so overlooked that groups like Numetrics had to come in and make companies plan.
What I like about the Top Gun method is number 5, set the most aggressive, yet achievable, targets. Though many projects, and even in school, the challenge of reaching certain milestones definitely gives a boost to morale. As we all know facing failure and frustration in projects is all too common, and often leads to a give up attitude. That is probably the number one cause of project slippage.
In the multitude of studies that Numetrics has performed, it appears that the human factor was completely removed. People's performance has been broken down into just numbers and graphs. In fact, a whole section of the class was about how to look at charts to figure out how to manage the teams progress. Although I feel that the planning in this Top Gun method is essential, I do not agree with tracking engineers via charts. The speaker grimly stated, "after review the graph and performance of the project, you can identify performance issues and make adjustments to the team easily." I think we all know what this means.
As welcomed as procedural project planning may be, let's keep the dystopian, Ninteen Eighty-four, future to a minimum. Ok?
Cabe