5. The ‘Temporal Design’ Principle: (and design half-life)

The currently best design may become invalid, at any time, due to changes in requirements, or by the addition of new designs to the total set of designs for a system.

All designs are under threat of extinction at all times.

Implications:

Final decisions on candidate designs should be delayed until the advantage from freezing the design, clearly outweighs the disadvantage of having to change the design.

• Any requirement change, even an apparently small change in the target or constraint levels of the requirement, or the delivery timing, can be sufficient reason to change candidate design decisions
   

• All candidate design ideas may need fresh review, for updated information about costs, timings, impacts, experiences, before irreversible commitment to them in the next implementation stages: otherwise they may have become invalid
   

• There is probably a finite half-life for a candidate design idea, depending on its rate of change environment culture, and for any idea that would not have been our first choice 5 years ago, the half-life is probably something like 1-2 years or less. The half-life of a design idea is when it is 50% less likely to be the best idea then when it originally was the best choice.
   

6. The ‘Multiple Value’ Principle

The value a design contributes is a function of multiple quantified performance dimensions. Any attempt to evaluate a design on a single dimension is doomed to risk or cause failure.

 Implications:

• You have to be able to quantify all critical performance dimensions, in order to understand the true value and impacts of a single design idea
   

• Failure to correctly estimate the impact on any single critical performance attribute of a design, may cause the entire evaluation of that design idea to be invalid or less optimal than assumed
   

• Any attempt to evaluate a design on the basis of only one or two performance dimensions, is almost certainly missing useful information, for understanding design consequences and risks
   

• Since it is unlikely that we have access to reliable information about design impacts in many dimensions, we are initially forced to take some risks when making any design choices; but we would be wise to avoid 100% commitment to a design until we can get some feedback on the other dimensions, once it is evolutionarily integrated into our system and field-trialed to some degree
   

7. The ‘Design Costs’ Principle:

A design must also be judged on multiple cost dimensions, including both short term and lifetime costs. Failure to consider all critical cost dimensions dooms your system to risk of failure.

Implications:

• A ‘perfect’ design for performance target levels can fail completely to acceptable because of one or more unacceptable or bad cost levels
   

• A reasonable engineering evaluation of any design will consider not only the performance levels it contributes to, but also the multiple costs it will incur during its lifetime in relation to specified cost constraints
   

• A design, once considered acceptable or ‘best’ may lose such status because of a change in some available resources during its lifetime. If, for example you cannot afford or recruit costly or scarce trained human maintenance staff, you may be forced to switch to another design in mid-life – or the entire system may abruptly have to be replaced, if it is too costly to switch to an alternative design
   

• Designers should design with the concept in mind that the system should be reasonably robust with respect to negative shifts in human, time and money resources during system lifetime – even if this means replacement of some initial design components
   

• The last-made design decisions must be made with respect to actual remaining resources for both implementation and operation, at the time their commitment is finally made. The designer cannot assume that all design is first in the queue, nor that resources are infinite

more...

© Tom@Gilb.com 2005