No Silver Bullet – Essence and Accident in Software Engineering

The Werewolf and the software project

• Both seem straightforward and familiar at first•Then they [can] become a monster.•For software, the monster is missed schedule, blown budgets and flawed products, etc.•We then cry for a silver bullet, an easy fix to the problem. But it is never there.

-First we need to see that problem is not that software progress is too slow, but that hardware progress is too fast.

-Second, we need to understand what rate of progress in software by looking at two different forms of difficulty:1. Essence – the difficulties inherent in the nature of the software2. Accidents – the difficulties that, today, attend its production but are not inherent.

The Properties of Essence

1. Complexity of the software-Leads to problems such as product flaws, cost overruns,

difficulty of enumerating, and unreliability.-complexity of functions makes it difficult to invoke said

functions. -complexity of structure makes it hard to add on new

functions-There also arises difficulty in management; complexity

makes it project hard to find and control loose ends, learning burdens, and so on.

2. Conformity-the difficulty in this comes from conformation of software to other interfaces.-This cannot be simplified out by any redesign alone.

3. Changeability-Change in software constantly occurs to extend its functionality.-The pressure for extended function come from users who like the basic function and invent new uses for it.

4. Invisibility-The reality of software is not inherently embedded in space.-We find that it to constitute not one, but several, general directed graphs, superimposed one upon another.

Breakthroughs Solving Accidental difficulties

1. High – level languages-It frees a program from much of its accidental

complexity.-The most a high-level language can do is to

furnish all the construct the programmer imagines in the abstract program.

2. Time-Sharing- It preserves immediacy, and hence enables us to maintain an overview of complexity.-The principle effect is to shorten system response time.

3. Unified programming environments-They attack the accidental difficulties of using programs together, by providing integrated libraries, unified file formats, and pipes and filters.

Technical Developments – Potential Silver Bullets

1. Ada / Other high languages- Not only reflects evolutionary improvements in

languages concepts, but embodies features to encourage modern design and modularization concepts

However- Biggest payoff came from the first transition, up

from the accidental complexities of the machine into the more abstract statement of step by step solutions

2. Object - oriented programming- Two data types: abstract and hierarchical; aka classes- Each removes one more accidental difficulty from the process.-gives designers freedom to express essence of his design without so much syntactical material- Unfortunately, it cannot remove the complexity of design; that in itself is essential, not accidental.

3. Artificial intelligence-There would seem to be hope in a breakthrough.-But, what AI provides, as of this point is the ability to make the program solve a problem the way humans seem to; how to recognize.-Hard thing about building software is deciding what to say, not sayng it.

4. Expert systems-A program containing a geveralized inference engine and a rule base – takes input data assumptions and explore the logical consequences, yielding conclusions and advice.

-For software building: suggesting interface rules, advising on testing strategies remembering bug-type frequencies, offering hints, etc.

-Difficulties: 1) making easy ways to get from structure specifics to auto generation of diagnostic rules. 2) knowledge acquisition

5. “Automatic” programming-Idea of programs that solves problems based on a statement of problem specifications.

-Nothing to the exact extant has been made at this point.

-But there are similar examples: Systems for integrating differential equations have also permitted direct specification of the problem; assessing parameters, chose from library of functions and generates solution.

6. Graphical programming-Application of computer graphics to software design.-Not a very promising possibility, considering that visualizing some programs is close to impossible.

7. Program verification-Proving the design is correct before effort is put into implementing ad testing.-A major amount of work is required to be able to fully verify a design, and even then, bugs can still potentially occur.

8. Environment & Tools-hierarchal file systems, generalized tools, etc.-Useful in programming, but the return at this point is marginal.

9. Workstations-It’s helpful to have faster running machines.-Don’t, however, expect a magical programming enhancement.

Promising attacks on the Conceptual Essence

1. Buy Versus Build- Any product is cheaper to buy than to build

afresh.- Cost of software has been development cost.- Many users operate own computers without

writing a program, yet can solve new problems with the programs they have.

2. Requirements refinement & rapid prototyping- Hardest part of building software is deciding what to build. -Most important function that software builders do for their clients is get from them, the requirements of the product.-A prototype software system is one that simulates the important interfaces and performs the main functions of the intended systems. -Much of present-day software acquisition procedures rests upon the assumptions that one can specify a satisfactory system.

3. Incremental development – grow not build- The system should first be made to run, even though it does nothing useful except call the proper set of dummy subprograms.-It is then fleshed out, with subprograms in turn being developed into actions.

4. Great designers-Software construction is a creative process.-Best designers produce structures that are faster, smaller, simpler, cleaner, and produced with less effort- Like Mozart and Salieri

Any Questions?