AIM INTERVIEWS

Who are you?

I'm Mark McCahill, a software developer that got into computers and programming when I was in Jr. High School and the first timesharing terminals arrived at my school. I've been fascinated by information technology (and technology in general) for as long as I can remember. For the last 19 years I've worked as a systems architect at Duke University. Before that I worked for the University of Minnesota Computer Center after I graduated from the U of M with a Chemistry degree.

Why did you originally create Gopher protocol in 1991? Was there any inspiration?

Initially, my team developed the Gopher protocol as a campus information system, with the idea that an information system that was distributed across independently managed servers would scale better than the alternative that had been proposed: a centralized system run on the campus mainframe. By treating each Gopher server as a first-class node in a distributed information system, we also saw that each Gopher server could be curated independently and optimized to suit different audiences, while also maintaining access to information published elsewhere.

Another group at the university wanted to centralize control of the information organization/architecture and we saw that as a non-starter. My team had been developing networked applications for microcomputers and Unix workstations, so creating a system that ran on the platforms we preferred was a natural choice. The campus mainframe was an excellent choice for scientific computation/number crunching but a poor option for publishing and organizing documents - and publishing/organizing documents is the core role of a campus information system. We also wanted to avoid what we saw as unnecessary complexity in the designed-by-committee centralized approach. The committee that designed the mainframe-centric approach had little experience with distributed systems, and so made some poor choices. My team had insights that committee lacked (like the importance of full-text search in document collections), so we ignored their design and built what we knew was a better system.

Inspirations: We had been using and developing apps with Apple's HyperCard, so links between pages of information in a collection was not a new idea for us. We had built email clients/servers for the POP protocol, so we were comfortable with writing applications running over TCP/IP. I had seen a demonstration of an Apple tool called "Gold Card" that used a full-text search engine to help support techs locate pertinent documents. We had a couple NeXT computers that came with a built-in full-text search engine to search documents stored on the NeXT machine. All of this convinced me that full-text search would win vs. tagging documents with keywords. We took these ideas and combined them into something that we wanted to use, and designed a protocol that was easy to implement cross platform.

In 1993, Gopher faced competition from the newly launched WWW. What do you think was unique about the WWW that made it ultimately successful? Is there anything you think could have been done differently with Gopher at the time?

Combining text and graphics in a document is well suited for advertising, so this formed a platform for those interested in commercializing the previously non-commercial research-oriented Internet. Combining documents and interactive forms, WWW set the stage for the Internet to become an online commerce platform. We could do these things in Gopher, but because most people started a Gopher session at a Gopher menu, Gopher was seen as lacking these features.

I tried to convince Adobe that if they would add URL support (links) to PDF documents they could be a better WWW than WWW because PDF renders document layouts much more consistently than HTML markup does - and with their tools, good looking PDF was much easier to author than HTML. If I had a larger development team we could have forced the issue, but alas - not enough resources were available, and Adobe was not interested in listening to me.

You also created one of the first graphical email clients POPMail in 1989 and were involved with the standardization of the URL format. You're also often credited with coining the term "surfing the net". If you had unlimited resources (time, money, etc.), how would you spend your time?

50% developing and using open source information systems - anything around knowledge graphs. 25% listening, writing, and playing music. 25% training for marathons and sailing.

Are there any people or projects you particularly admire today?

Alan Kay and his team at Xerox Parc have always been at the top of my list. The Data Intelligence Lab at the University of Hong Kong - we are using LightRAG for several systems here at Duke.

What's a message you have for the world?

Keep your designs simple and clear. Stay organized and don't make a mess. As the chef Anthony Bourdain would say, "Work clean".