George Engelbeck

User Interface Researcher/Designer

usAvenue — 1995

Background

usAvenue was a high-end, home shopping Interactive TV (ITV) channel that combined a TV-quality lifestyle entertainment and shopping.

usAvenue - ITV layout

Problem

  • Create a compelling home shopping infrastructure that would attract merchants and shoppers,
  • Support the dynamic switching and updating of video and product content, and
  • Make it easy to view and easy to make impulse buys.

Constraints

  • Be able to work on any ITV system,
  • People sit about 10 feet away from their televisions, and
  • The set top boxes mostly didn't support interactivity well.

Approach

  • Create a design language for the common denominator remote.
  • Iterative prototyping and testing with potential users.

Solution

  • A layout that allowed merchants to put their own look on their store while still looking like usAvenue,
  • A lifestyle programming layer that users could shop from, and
  • A 0-9 menu system with really large text.

Results

  • usAvenue was targeted to launch on U S West's TV trial in Omaha. It was supposed to be rolled out to 10K homes by the end of 1995. usAvenue was canceled in the summer of 1995 along with the rest of U S WEST's ITV efforts. Simply, there weren't enough early subscribers to justify the investment.

Theo - An expert system to predict solar flares — 1985

Background

Theo was an expert system that made solar flare forecasts from white light data. Theo was created in collaboration with Pat McIntosh a solar flare forecasting expert at the National Oceanographic and Atmospheric Associations' Space Environment Lab (NOAA/SEL).

Theo C-class decision tree

Problem

  • Create an expert system that could be used by a novice forecaster to successfully forecast solar flares,
  • Show that expert systems technology could be applied to solar forecasting, and
  • Provide a training tool for novice forecasters by allowing them to experience an expert's reasoning process.

Constraints

  • Use a known expert systems development environment - OPS83, and
  • Complete the project in three months.

Approach

  • Interview the expert using shared decision trees to facilitate the knowledge acquisition process,
  • Iteratively test the decision trees by walking through past forecasts, and
  • Testing the the system by comparing novice forecasters with our expert and the forecast center.

Results

  • The system with novice forecasters performed as well as our expert and better than the forecast center.
  • The system was extended by NOAA/SEL and was in use for over 20 years.