[This is the second post in a series looking back on 1994, the year in which the web transitioned from a platform for scientists to a commercial interaction medium. At the end of my first post, set in January 1994, I revealed that I didn’t know what a web server was.]
How could I, a product manager at a company that sold high-performance UNIX workstations and servers, not know what a web server was? The simple truth is that in January 1994, I had never seen nor heard of the web. Hardly anyone had.
Through a bit of online archeology, I now know that the total number of web servers in existence at the time was less than 8001 – and they were almost all of the “.edu” flavor2, hosted at places like CERN and SLAC. And although there’s no count of how many people were on the web at the time, my best estimate is 3.7 million3.
The web of January 1994 is largely gone, and cannot be re-constructed. But, believe it or not, in that month some folks at Georgia Tech’s Graphics, Visualization, and Usability Center did what must be the very first web-based survey attempting to characterize the users of the “world wide web”. James Pitkow and Margaret Recker were trying to learn things like: which browsers were people using, how frequently were they surfing the web, and some basic facts about who those early adopters were (and to see whether, as they believed, the web would be a better platform for surveys than email had been).
And even more surprising than knowing such a survey was done so early, is discovering, as I did a few days ago, that the original survey and its results are still online!
So, let’s use those survey results to travel back in time to a very different web than the one we experience today. First up: gender. Apparently, the web of January 1994 was, to put it mildly, a bit of a boys club. Males accounted for a whopping 95% of respondents. I assume this says far less about the web than it does about the professions that were among its earliest users. (Physics, I’m looking at you!)
And these folks diverged from the mainstream in another significant way. They were not surfing the early web on commodity hardware from the old “Wintel” duopoly. No, 92% of them were on UNIX workstations (and most likely enjoying always-on broadband connections via Ethernet, versus slow, intermittent connectivity via dial-up).
Remember the browser war, when Microsoft and Netscape fought each other, tooth and nail? Well, that was still a ways off, as Netscape did not yet exist, and Microsoft had no plans for making a web browser. Nonetheless, the browser question offered no fewer than five choices, listed alphabetically (Cello, Lynx, Mosaic, Other, and Samba). As it turns out, that was three choices too many. Mosaic, developed by Marc Andreessen and Eric Bina at NCSA and released less than 12 months earlier, had completely taken over, accounting for 97% of respondents! (The remaining 3% were using Lynx, a text-based browser developed by Lou Montulli, Michael Grobe, and Charles Rezac at University of Kansas).
It’s also interesting what the survey reveals about the utility of the early web. With fewer than 800 total servers on the web, it’s easy to imagine that usage would be fairly infrequent. Quite the contrary; 20% of respondents used their browser more than nine times a day! Another 18% accessed the web 5 to 8 times a day. And another 42% reported one to four times a day. Together, that’s 80% of early users finding the web so essential that they used it every single day.
You can see all of the graphs here, and read the full paper here. Who knows just how representative this data is of the whole of the web at the time? But as far as I can tell, it is the only such dataset of its kind from that time period, so let’s be thankful that it exists, is online, and can be read by modern browsers.
To be continued…
1The number 800 comes from a model I built, starting with data from M.I.T.’s Matthew Gray, which counts 623 servers in December 1993 rising to 2,738 by June 1994.
2Also from Matthew Gray
3The 3.7 million estimate I backed into, based on data for 1995 through 1999 from Internet World Stats, with year-over-year growth of approximately 110%