What's Wrong With Today's UIs and What We Can Do to Help

(Last modified Saturday, 3/25/95 @ 1:50PM PST)

Introducation

User Interfaces have come a long way since the days of punch cards. But there is still a long way to go. Today's User Interfaces do present people with more computing power than ever before, but bla bla bla....

What's Wrong With Today's UIs?

General

Structure generally seperates programmers--those who create application, from users--who use the applications. This philosophy is bad because using a computer is really programming it--telling it what to do.

The user perspective

Having to deal with an application layer does nothing to benefit the user's productivity or happiness with the system.

Today's user interfaces are inconsistant. A command run in one application may have no relation to the same command in a different app. Even the way to run the same command in two different apps can vary widely (ie X windows).

Users are dependent on developers to provide decent interfaces for their applications. There is nothing they can do to make an ugly program look better.

The whole concept of applications and linear flow of control, on which many OSs and applications are based, goes against how the human brain works. These concepts force a good user interface to use abstraction, to hide the "real" system from the users. If the underlying OS had been designed well from the start, you could present them with what was really there.

The programmer perspective

Current UIs are a pain to program for. Not only must the devloper concentrate on program logic, but he/she must also provide painstaking details to the interface. In a windowing system, for example, the programming cannot simply ask for the user to select one item from a list, but must instead display a window, draw each button seperately, and track their individual pressing.

Current UI APIs are monolithic, not broken down in smaller, more useful and replacable components.

Many OS/UI vendors, such as Apple Computer, provide rigerous standards for an application's look and feel to conform to, but provide little or no software assistance to developers for implimenting such standards.

Due to the non-OO nature of most UIs, a helpful (UI) subroutene developed in one application may require changes in the logic of another program, should this other program's developer wish to incorporate it.

How Tunes will be different

Concentrate on how users work with the system, not how the system can work with the user ("Ask not what you can do for your operating system, but what your operating system can do for you.")

Using = programming, just in a real-time sort of why. Users aren't stupid; they just don't have time to muck around with all the silly details of programming. For them, we should have easy-to-use scripting language, for intelligently automating stupid tasks.

Every component in the system--seen from both the user and programmer mindsets--should be replacable. This, done through object-orientation, will make replacing individual, outdated pieces of the OS as painless as possible.

To make different copies of Tunes compatible with one another, there will exist a set of standards that all Tunes systems should support in some way or another. Such standards should allow for IO of a number of different data types.

These standards should concentrate on specifying what should be done, and not how to do it. Instead of specifying the interface with which a user can input a certain data type, a program would ask the interface for a certain type, and it would input it in a user-selectable way.

The APIs will be modular, so that you can access and interface with only the parts that you need for a certain job.