Search in 2007 – Could it work?
The scope of building something as complex as a search engine, however you define it, is vast and requires a lot of thought. I had two nephews, Mandrake and Nigel (not their real names) who had some real talent in programming and hardware respectively, who I tried to entice into joining me on this project. One had a degree in linguistics, which would play a key role in natural language processing of queries. The other what quite an expert in hardware and network architecture.
We spent weeks going over the various challenges in something of this scope. We convinced ourselves that the regular Internet search engines had the means to create authoritative listings, but they had no real desire to do so, because it was not essential to their business model.
So what would our business model be? Creating authoritative listings would be a herculean task, but if it were broken up in manageable pieces, it could be done over time inexpensively and fairly rapidly (I later proved this to be true). If that task alone were manageable, we had a good shot at making a sensible business model.
Assume that one city with 12,000 businesses could be cataloged. The cataloging means that you have collected the business name, days and hours of operation, address, exact GPS location, contact information and an accurate manifest of their products and services. If you could maintain that catalog frequently enough, it would become authoritative. You could then re-market that data or provide ancillary services based on it. That catalog alone would be worth a considerable amount to interested parties on an ongoing subscription basis.
Consider also that you are collecting accurate GPS data, not just interpolated street address approximations. All GPS devices use line segments, or vectors, to represent streets and roadways. Based on data derived from the US Census Bureau, streets have a beginning and ending street number, odd ascending addresses on the right side (from lower addresses towards higher ones), and even numbered addresses on the left side. No one has collected the locations of businesses or other structures accurately, not even Google street maps, so the GPS devices tend to route you to places that don’t exist, or can be off by almost a mile in some cases.
If you could accurately place a driver at a location without the interpolation errors, what would that be worth? It would save thousands of hours of backtracking, wasted fuel and meaningless frustration. In other words, it would be quite valuable!