KnowledgeWare

~ the goodbye ~

James Martin founded Database Design Inc in Ann Arbor in 1979 to sell his own methodology, which he called information engineering. Martin was the most commercially successful technology author alive at that point, a man who published books faster than most people could read them and filled hotel ballrooms with data processing managers at seminar prices. Information engineering held that if you modeled the business correctly, the code was a mechanical consequence of the model.

Fran Tarkenton played eighteen seasons at quarterback, retired holding the NFL passing records, and went into business software. In the mid eighties he merged Tarkenton Software into Martin's company, moved the headquarters to Atlanta and kept the KnowledgeWare name. Martin brought the method and the reputation. Tarkenton brought a sales organization and a face that opened doors.

The product was IEW, the Information Engineering Workbench, later succeeded by ADW, the Application Development Workbench. You drew diagrams of your business on a PS/2 and the workbench generated COBOL out the other end. This was CASE, computer aided software engineering, and for roughly five years the enterprise IT industry believed in it with real money. Sales reached $92.3 million in 1990 on $9.8 million of profit, up about 1,700 percent in four years.

In the United States the company was sold partly on who was running it, and that is the part that does not travel. Tarkenton is in the Pro Football Hall of Fame. In Turkey that sentence is a row of English words. American football has no foothold here whatsoever and never had one, no league and nothing on television that anybody watched, so the framing that made this company legible to an American CIO in 1990, the famous athlete who turned out to be a serious operator, would have arrived in Ankara as static.

~ the chronicle ~

~ what it sold ~

CASE promised that building applications could stop being craft work. Draw the entity relationship diagram, specify the process model, press the button, get the program. What it reliably delivered was diagrams. The generated COBOL was verbose and unpleasant to maintain by hand, which meant that once a shop went in it could not easily come back out.

The tools cost a fortune and the training cost more. They were bought for exactly the enormous multi year projects that fail for reasons no diagram has ever addressed, and when those projects failed the tool was standing right there to take the blame.

~ the funny part ~

Tarkenton taught classes at the company's Peachtree Street offices, and the line he is remembered for there is his own verdict on the thing he was selling. A fool with a tool is a faster fool. He was correct, and he said it while running the company whose tool it was.

~ the long tail ~

The third quarter of fiscal 1994 came in at a $25.8 million loss. A quarter of the workforce, 240 people, went out the door. Sterling Software announced a deal in august 1994 valued at around $143 million and closed it on 30 november at roughly $100 million. Full year sales were $132.5 million, higher than 1990, which is the shape of a company that grew into costs it could no longer carry.

Then the accounting caught up. Former shareholders sued in january 1995 alleging that executives had deliberately misstated earnings between november 1993 and august 1994, and Sterling set aside $15 million for the legal bill it had inherited. On 28 september 1999 the SEC filed in Atlanta against seven former KnowledgeWare executives over a scheme that inflated reported earnings for the fiscal year ended 30 june 1994. Tarkenton settled without admitting or denying the charges, paying a $100,000 penalty and giving back $54,187 of incentive compensation plus interest.

Sterling Software was itself swallowed by Computer Associates in 2000, which is where a great deal of this era ends up. Information engineering did not survive as a discipline and the workbenches went with it. The idea underneath keeps returning every decade or so wearing a different name, most recently as a promise that you will describe the software and something else will write it.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »