We had Internet access since 1993. Our office was a small rented room in the former Soviet Academy of Sciences Institute of Cybernetics. One day a friendly guy from the neighboring office walked in and asked: “Do you want Internet to connect with the world?” I immediately said yes. I still remembered how important our telecommunication lines had been during the independence movement — connecting to the world mattered.
Because we were running a Novell server, I had to switch our office network protocols to TCP/IP and connect a cable from our server’s serial port, through a mysterious little box, into their office. Somehow Novell even managed to do IP routing, so I received a block of dedicated IP addresses for my local network.
I had no idea what exactly I had just bought, but I started spending my nights in the office discovering the new possibilities. It was funny that I could suddenly connect to AOL, and various BBS systems running in some American guy’s bedroom — without paying the huge international phone bills that had been normal before. So I spent my nights talking there, and slept on the office floor in my sleeping bag, just like in the old days when I programmed Soviet clones of the IBM/360.
Our Novell server also ran Gupta SQL Server, which we used to test systems for the emerging banks and telecoms of newly independent Estonia. So when someone offered me a chance to run their WWW server module inside my Novell infrastructure, I took it. Novell had always been a closed, corporate environment, and anything new there was a challenge.
That’s how I got involved with the World Wide Web. After hearing about a US company called Netscape, I flew there and made a deal to represent them in Estonia. I had done something similar earlier — flying to San Francisco to knock on Steve Jobs’s door and ask for a NeXT computer to introduce in Estonia. He was excited to give me one.
The guy running early Yahoo from his bedroom was happy to list my corporate web pages in his catalog. We were the 12th company in the “computer software” category.
But as the web exploded, Yahoo’s manual catalog couldn’t keep up. So we discussed automating the process. The same guy who wrote the Novell WWW module said he could write another module that scanned new DNS records and checked whether port 80 was active.
I asked my “Internet provider” — a small office with a Sun server — whether I could access their DNS zone file. He said yes.
So we had a module that scanned the expanding Internet and stored every new website into my Gupta SQL database running on Novell.
We stored every HTML tag as a separate SQL record. If a page had <b> and <i> tags inside the <body>, those were separate rows. It was a very early form of structured indexing.
Then I hit a problem. My D: drive — normally used for tiny customer test databases — kept filling up. After running the module overnight, the disk was always full. Luckily, a guy in the same building ran a computer‑components shop. As hard‑drive prices dropped, I frequently visited him in the morning to buy a bigger disk. Seeing my sleepless face, he always joked: “Again you’ve written too much code.”
With a growing database of newly discovered web pages, we discussed what to do with it. Yahoo couldn’t keep up anymore. So we wrote another module that generated SQL queries and exported the results into a file accessible via HTTP. Coming from HTML, we standardized the file into a tag‑based syntax — something like early XML or RSS.
I contributed the SQL logic: complex formulas, tag weights, keyword‑occurrence calculations, ORDER BY and GROUP BY tricks. He wrote the Novell module. Together, the system could answer a query. Several people ran similar modules on their own infrastructure.
We never built a UI — it wasn’t important. It was a proof of concept. We were all early‑web enthusiasts, using AOL handles instead of real names. I was always “Veiko,” and Americans always struggled to spell it correctly. That was normal in 1995.
Because I was still in Estonia in 1995/1996 and couldn’t participate in Silicon Valley meetings, someone from our enthusiast group eventually took the concept and developed it further at Inktomi. Our system had already been running — distributed crawling, tag‑based indexing, SQL‑based ranking, and a proto‑XML export format — long before search engines became an industry.
And after Marc Andreessen got his “golden parachute,” leaving many Netscape developers empty‑handed, I had enough of the corporate world and moved into alternative music instead.
To my surprise, Google later took the same general concept — distributed crawling, structured indexing, relevance ranking — and made it successful.
