The ChatGPT of 1993

The first graphical web browser inspired me to start a business on the Web. People thought I was crazy then, too.

It’s March of 1993. I’m on a Freddie Mac team visiting General Electric’s research lab in snow-saturated upstate New York to exchange information about progress in automated mortgage underwriting. At some point, the higher executives want to powwow by themselves, so a GE software engineer takes me by the elbow and guides me to the basement.

“Let me show you something,” he says. What he shows me, over a high-speed Internet connection, is Mosaic, the first graphical Web browser, released to the public about six weeks earlier by Marc Andreessen and his cronies at the University of Illinois National Center for Supercomputing Applications. I am awed by the demo.

A year later, I can still remember Mosaic, but I don’t have a way to get it. Access to the online world is via dialup. The major providers, such as America Online, do not offer Web access. The only way I can reach the Web is through a dialup connection provided by the local public library, and the only Web browser they make available is a text-only browser, called Lynx. After all, Tim Berners-Lee, who invented the Web protocol, saw it as a medium for academics to share research papers, not as a tool for sending graphics or the user interaction that the NCSA techies enabled via fill-out forms and what was called CGI-bin.

In the Spring of 1994, I am mistreated by colleagues at Freddie Mac. Humiliated, I quit. Just short of my fortieth birthday, despite having a three daughters aged four to ten and a non-working spouse, I leave a well-paying job to start my own business. On the Web.

Note that at this point, the dialup services still have not provided access to a graphical Web browser. I start what I call The Homebuyer’s Fair on sheer blind faith that I have correctly anticipated the future.

Not everyone agrees with me, of course. My former Freddie Mac colleagues are quite concerned, doubting my sanity. Microsoft’s Bill Gates has yet to see the Internet as deserving serious attention. In the summer of 1994, I meet with a sharp businessman who has a successful Internet business, and he advises me that the Web is not ready for prime time, and may never be. Internet connections to households are very slow, and as a result graphics take way too long to download. He runs his business on the text-based Gopher protocol, and he advises me to do the same. I stubbornly refuse.

For more than a year, it seems that I am wrong and the Web skeptics are right. But then in 1995, a dam suddenly breaks. Netscape, the company formed by James Clark(1) along with Marc Andreessen and the NCSA kids, has burst onto the scene with its souped-up version of Mosaic, and its spectacular IPO sets off a financial frenzy surrounding the Web. America Online, fearing mass defections by users anxious to access the Web, finally offers access in August. Microsoft’s long-delayed Windows 95, the first Windows version that contains the TCP-IP protocol necessary to enable PC users to get on the Internet, is released that same month.

The story of my business ends happily. I wrote about it twenty years ago in Under the Radar.

My point here is that ChatGPT today feels to me like Mosaic did in 1993. Skeptics see the problems, the faults, the shortcomings. These issues are valid, just as the issues that were holding back the Web were valid until late in 1995, if not later (the rollout of broadband was gradual).

So what? So many possible uses for ChatGPT can be imagined. Some of them will be realized. Others that are not even imagined today will arise in the near future. The problems will be overcome. Today’s flaws will soon be forgotten. Probably deeper social problems will emerge. For now, don’t lock yourself in to being a nay-sayer. I have the same confidence in ChatGPT today as I did in Mosaic in 1993.

(1) Michael Lewis provided an entertaining profile of Clark and Netscape in The New New Thing, first published in 1999.