Disposable information systems: the future of software maintenance?

This paper is a summary of Voas s keynote address Are COTS Products and Component Packaging Killing Software Malleability at the International Conference on Software Maintenance ICSM in Bethesda MD on November The State of the Industry The news coming out of the Information Technology IT community today is mixed The good news stems from predictions of continued industry growth IT revenues for were B and are predicted to be B in B in B in B in and B in This represents a near doubling of revenues in a year period Also corporate IT spending is up In IT budgets were of gross corporate revenue in that gure is expected to be A serious worker shortage in the IT industry is the bad news According to ITAA s IT Workforce Convocation in January there are un lled IT jobs Software Productivity Research estimates that the US shortage in software engineering personnel is between and They further go on to state that if you add in Year needs that number may be as high as If you work in the IT industry it may not be clear why this is problematic After all it does suggest high employabiliy without regard for quali cations While that is ideal from a worker s perspective the downside for the industry is that numerous new projects are never even started It is not clear however whether there is a real or arti cial personnel shortage An arti cial personnel shortage occurs when for example people are on strike or people are wasting time while on the job A real personnel shortage occurs when there is truly not enough manpower to complete waiting tasks Capers Jones has analyzed wasted personnel time in the IT industry His analysis looked at how workers actually use their average working days this excludes weekends vacations holidays sick days training department meetings i e time on activities not directly tied to working on software and found that testing and defect repairs accounted for days Y K and related repairs accounted for days and time on projects that will ultimately be cancelled accounted for days This means then that days or of the total worker time is applied to productive software activities Note that if we could convert the days spent on soon to be cancelled projects the worker shortage could be greatly reduced or even eliminated I would argue that degree of time loss due to ine ciency is creating to some degree an arti cial personnel shortage Further indications that our profession is in turmoil can be seen when we look at the quality of our products Musa et al exposed that defect densities have remained fairly constant during the past years faults per KSLOC There are two interpretations that can made from this nding We can argue that this is good news since systems are larger and more complex Or we can argue a negative case since all of our advanced software engineering theory and tools have failed to decrease defect densities Since the faults per KSLOC gure was based on a xed size unit of code as opposed to a per system basis I would argue for the more negative position What we can summarize from the previous statistics are that Approximately of our time is spent on non productive activities Roughly two thirds of the software workforce is engaged in rework and repairs of the labor force is working on software systems that will not be deployed There possibly would be no labor shortage if software quality could be brought under control and And we can speculate that no other major profession wastes so much e ort From these statistics the project cancellation gures are most troubling Capers Jones reports that of all systems of size around function point are never completed And the numbers only get worse as the number of function points in a system increases For example Capers reports that of all systems of size close to function point are cancelled The Chaos study by the Standish Group found similar statistics They reported that of projects whose costs were in excess of M failed and for projects between M and M the project failure rate was While much of the aforementioned industry data seems pessimistic it is not meant in that way I m simply restating industry metrics Further we did not build skyscrapers after years it took thousands of years of engineering knowledge to reliably produce such physical systems Given the poor quality of software what is surprising is the rapid move toward acquiring Commercial O The Shelf COTS software from third party vendors In of a typical corporation s IT portfolio was COTS software In that gure is expected to jump to In this gure is expected to be or greater Do we honestly be lieve that software developed elsewhere is better than if we developed it ourselves Whatever happened to the adage if you want something done right you have to do it yourself