Niklaus Wirth: I tell you a story. I had an excellent PhD student [..] So I recommended him to join Microsoft. [..] He had subsequently climbed up the ladder to Bill Gates and presented his proposal to him [to start the Word system new from scratch]. He was so convincing that Gates said „Yes let’s do that. You start with your group.“ And they worked busily on this. Their software was supposedly very good, and then Microsoft sent out sales people into the field to ask the customers what they would think of it. Everybody said „Oh thank God, yes, we are so tired of this unreliable Word system. We will be very happy to adopt it under one condition: it has to be able to process all our texts and it has to be upward compatible“, essentially saying “ We have invested so much in learning to use this system that we don’t want to loose this investment“.
Edgar D. Daylight: The programming language papers presented at top conferences are very formal using, for instance formalized type systems. But, the techniques have to work on C or on C++. It is a bit ironic because as a researcher you thus have to work a posteriori: start with the bad languages and then try to make good.
Niklaus Wirth: And preserve upward compatibility, which means that the type system may detect errors or it may not.
Edgar D. Daylight: It’s discouraging to get into software.