A Series of Miscalculations

"Arguing in My Spare Time" No. 2.18

by Arnold Kling

May not be redistributed commercially without the author's permission.
Index

One of my beliefs about competition is that the business world is very forgiving of mistakes. This is based on my own experience.

In late April of 1994, I quit my job with a large company in order to start a business on the Web, called The Homebuyer’s Fair. On the advice of Rob Main of Electric Press, who developed the first version of the site, we simplified the URL to homefair.com.

I had no business plan. Even if I had known what an "elevator speech" was, I could not have given one. I simply had a concept of providing articles about buying a home and engaging in some form of disintermediation in the real estate and mortgage business.

If I was vague about my entrepreneurial strategy, at least I was clear about my fallback position. I did not burn my bridges with my former employer. Moreover, I thought that even should my business fail, I would enhance my human capital by learning about the Internet. My willingness to launch The Homebuyer’s Fair was based more on my confidence in these fallback positions than on my belief in the venture itself.

One of the first things I needed to do was obtain a connection to the Internet that would enable me to use a graphical browser called Mosaic. This in turn meant that I needed to install a software "socket" for my Windows computer. Configuration of this "Trumpet Winsock" taxed my technical acumen beyond the limit. Eventually, Dave Stoddard, the President and founder of us.net, drove to my office and installed the software.

At this point, I should have realized that I was out on a limb. If someone with my education and experience in computer programming could not readily gain access to the World Wide Web, how many homebuyers or mortgage borrowers were likely to come to my site? I remember feeling very deflated early that summer, when Jeffrey Dearth of the Electronic Newsstand told me that although there might be 20 million Internet users, he thought that fewer than 500,000 had used Mosaic. He was not even concerned about putting his site onto the Web, because there were so many more users of Gopher, a hierarchical, text-based protocol that was popular at the time.

Also that summer, I hooked up with Abbot Chambers and Dale Dougherty of "Global Network Navigator," a premature portal. I wrote a couple of articles for Abbot’s "Personal Finance Center." In one article, I predicted that a solution soon would be found for the problem of unsolicited email. In the other article, I predicted that micropayments (payments under one dollar, or even less than a penny) would become widespread, and that this would lead to the rapid growth of alternative banking institutions on the Internet. This morning I received half a dozen spam emails. I have never made a micropayment.

By January of 1995, I was having mixed feelings. On the one hand, the Internet was generating a lot of "buzz," and traffic on the site was growing. On the other hand, none of the three major online services—Prodigy, CompuServe, and American Online—offered Web access. I was not sure that AOL really would provide access to the Web. My opinion was that if they did, the result would be the end of their ability to earn revenue from businesses and ultimately the death of their service. In August, they did provide Web access, and this was one of the best things that ever happened to them.

I asked Eddie Snyder, the head of the firm that did my accounting, for advice. He told me that he thought my company was worth $20,000, and that if I worked at it, in five years it might be worth $1 million. That did not sound like a comfortable position from which to plan the first of three bat mitzvahs, so I began to think in terms of going back to my old job.

By this time, banner ads had been introduced to the Web. Here is my initial reaction to them, which I posted to a discussion list. "I don't think [banner ads are] a solution, because [they are] so stupid. Does anybody bother to click on the sponsor's ads? Most people surf with some sense of purpose; when they wander off in some direction, it is because they are following a thread, not because they saw an ad for something that has nothing to do with what they are surfing for."

As a result of this thinking, I made no effort to obtain banner ads for The Homebuyer’s Fair. Not that I had any other reliable source of revenue. My sponsors consisted of a rag-tag collection of a few mortgage lenders, some apartment relocation services, New Homes Guides for two cities, and one home builder.

Just as other entrepreneurs were starting to see the potential of the Web, I executed my fallback strategy and went back to my old job. I was ready to give up the web business, but my wife, Jackie, felt that since revenue from sponsors was covering costs I should stick with it for a while. So I would spend an hour in the morning, an hour at night, and much of my weekend time maintaining The Homebuyer’s Fair.

In August of 1995, I had another lunch with Jeff Dearth, whose Electronic Newsstand had indeed moved to the Web as Enews.com. I wrote an article for their site about Netscape’s initial public offering, in which I predicted that the romance between Wall Street and the Internet would not last. Today, one would say that this romance had not even begun.

Also in 1995, I was asked by Cliff Kurtzman of Tenagra.com to participate in choosing award-winners for Internet commerce. In the end, I dissented from the choices, because they included Yahoo. I believed that the Yahoo directory model was flawed, and that it would not scale as the Web grew. I was concerned that within a year Yahoo would have fallen out of favor, and I did not want to be embarrassed by having participated in giving an award to a defunct company. These fears of Yahoo’s death were just a bit exaggerated. Eventually, Yahoo somehow overcame the sting of my refusal to make its selection unanimous, and it now has a market value of close to $50 billion.

Meanwhile, I had hooked up with a small relocation company in Scottsdale, Arizona. They had good industry connections, and we were able to obtain real estate company sponsors for a free cost-of-living comparison tool that we developed for The Homebuyer’s Fair. In 1996, I traded my company for a one-fourth stake in a new joint business with the relocation company.

We needed to develop new applications for our site, and I had a number of good ideas. However, I did not know Perl, the standard language for dynamic Web programming. I attended a seminar at which Frank Hecker, a salesman for Netscape, casually mentioned that Netscape’s next web server supported server-side Javascript, which appeared to be an easier solution. It seemed to me that with this easier scripting language, Netscape would soon dominate the web server market.

We purchased the Netscape server, and I went about development enthusiastically. Also, I started a user’s group in the Washington, DC area for the server. At the user’s group, I met Dirk Reinshagen, who immediately struck me as highly knowledgeable. We eventually hired Dirk as a consultant.

The Netscape server turned out to be a disaster. The server-side Javascript applications that Dirk and I developed resulted in stress to the server that was way beyond what it was built to handle. Our server was crashing every few minutes, and taking a minute or so to restart. Apart from our site, no other company, including Netscape itself, attempted to make extensive use of server-side Javascript.

Late in 1997, at Dirk’s suggestion, we tried the JavaWebServer from Sun. Although this was not recommended for production web sites, we already were in desperate shape, so we had little to lose. It turned out that the JWS, which was not advertised as being able to operate in a high-stress environment, was able to do so. Conversion from Netscape to the JWS took a year.

By this time, I had quit my day job again. The Scottsdale relocation company had hooked up with the largest newspaper publisher in Arizona, which invested in our business. I was now full time with homefair.com (like many Internet companies, we changed our name to match our URL).

Also in 1997, we noticed that one of our web partners, Mapquest.com, was undergoing some upheaval. We considered offering to buy the company, but when we learned that they had accumulated a few million dollars in debt, we passed. They have since gone public, and as of this writing Mapquest has a market capitalization of close to half a billion dollars. Although we did not buy the company, we did have fruitful discussions with some of their former technical staff, who set up a consulting firm, part of whose mission is to support homefair.com.

In 1998, we merged with theschoolreport.com. Because their business model was so similar to ours, I assumed that it would be easy to merge our systems. The cost of merging our two systems turned out to be higher than the cost of building the systems in the first place.

On October 11, 1999, our business, along with the Scottsdale relocation business, was sold to homestore.com for $85 million. My share was quite dilute by this point (prior to this sale, the newspaper company owned 80 percent, and the rest was divided among numerous partners in Scottsdale and with theschoolreport.com). But a couple of percent of $85 million is still real money, particularly considering the sequence of mistakes, miscalculations, misjudgments, and erroneous forecasts that led to it.