Multimedia Object Oriented Specification Engine

This would consist of a library of real-world classes, to be used to specify the user's requirements. A report for example has a typical structure (header, detail, body) and behaviour (print, etc.) Such a standard class could then be sub-classed as appropriate. In this example, it would be linked to a data source, also derived from the library. Icons would show the user the behaviour and properties of these objects, and how they inter-relate. GUI mock-ups would show how information is presented.

I believe that this approach would be much more meaningful to end-users, whilst at the same time offering the possibility of automatic integrity checking. This would highlight any contradictions or dangling items, resulting in (one hopes) a complete and accurate specification. Far better than a lot of words, which nobody will read or fully understand.

A bridge between the informal world of users, and the formal needs of software engineering.

The following has little to do with your (interesting!) idea, but something you might consider. Why icons are evil:

Icons don't normally contribute information. They are merely motifs that symbolize information learnt by the user throughout the life cycle of the application. That is, it is really difficult to convey the concept "Format disk" as an icon (and if anyone has seen the standard Windows icon, you'd agree). However, once the user has read the tool tip, the little pictograph suddenly represents "Format disk" to the user from then on. However, since this requires the user to learn a new language ala hieroglyphics, why not just use the words "Format disk" which has immediate and permanent meaning to the user? How many people spend two minutes waving the cursor over the MS Word toolbars trying to find a function?

The two minutes are well spent, for the next time the symbol and a click are all that is necessary, no need to type some command words, and if the function is used many times, it will not be long when the two minutes can be considered a small price to pay. Such is the road which leads to later productivity.

By the way, isn't VisualBasic half-way where you want to go? And it's simpler than object-oriented languages for end users. Polymorphism and inheritance are non-trivial concepts. Just consider the canonical CircleAndEllipseProblem which many "object-oriented" developers don't get yet. -- SunirShah

It has been stated a picture is worth a thousand words. While Icons are not pictures, they can be effective, symbolic tokens for objects and methods which may be essential in accomplishing the purposes for which one is programming.