>Hi Terry
>
>>Good leadership, a sexy project, incentives and equity might overcome prima donna issues. If there is a pot of gold at the end of the project a team of prima donnas could do it.
>
>You may be right at that.
>
>>The real trick [as always] is in the planning - assessing the market and translating those requirements to a simple, understandable plan that will deliver the project on schedule and on budget!
>
>I agree, that's crucial. However your last phrase is the kicker. "on schedule and on budget"
>
>The most difficult thing in any software project (of significant size) - tracking and assessing 'progress'. To do so you need to have good estimates of the work scope and the time required. As a project manager (I project managed Civil Engineering and other non-software specific projects before becoming a programmer) the one thing that I find that continually frustrates software projects is the Estimation/Progress cycle. It comes down to to two key questions:
>
>[1] How to come up with a reliable estimate of time (and hence costs) for developing a software application
>[2] How to ensure that the actual development is on schedule
>
>The consistent inability in the software industry to consistently and reliably handle these two elements is why the majority (and I have seen estimates as high as 85-90%) of all major software projects are late and over budget.
>
>If someone could come up with a reliable way to objectively assess how much has been done, and how much remains to be done, and to estimate the times required to the levels of accuracy that are achievable in other industries, there is a fortune waiting for them <bg>

Andy;

In my previous life I was an electronics engineer. I find that the term “engineer” associated with software to be very inaccurate and misleading. True engineering has predictable results that can be proven using mathematics and physics.

Personally, I think that software is more of an abstract phenomenon, requiring mathematical, and business skills, as well as creativity, to be successful. There is no true model that one can emulate in software and reproduce a new success with predictable results. There are too many variables to consider as I see it. The more complex the project the less control you have over the end result (delivery time and cost).

I think my problem with software is attempting to use predictability in a discipline where you have no true concept of the total number of variables or which variables to consider. :)

Tom