The Year the Modern Internet Actually Started
September 4, 1998. I was sitting in a cubicle at a small web consultancy in Chicago, watching our boss try to load a webpage on a Pentium II with 32 MB of RAM. The Netscape Navigator browser was hanging on a page with three animated GIFs and a Flash banner. He sighed, hit refresh, and said "this is going to get better." He had no idea how right he was about that week. 1998 wasn't some singular moment. It was the year the internet stopped being a strange academic hobby and started being something normal people used to buy books, talk to strangers, and eventually complain about. Four things happened close enough together that they blurred into one era. The first was Google's incorporation. September 4th, exactly. Two grad students at Stanford set up a company in a garage. They didn't even have a product that worked well yet. Their search algorithm was good, but the real story is that they decided to index the entire web instead of trying to build a portal where people went to get lost. Yahoo was doing the portal thing. Amazon was barely shipping boxes. Microsoft was writing a press release about the internet every quarter. Google just built a search box and waited.
The second was Windows 98 shipping. May 15, 1998. This is the one most people actually experienced. Before Windows 98, installing software on a PC was a process that involved reading a manual and making phone calls to technical support lines that charged you six dollars a minute. Windows 98 introduced the concept of "it just works" to millions of households, even though it still crashed about once a week and required you to reboot whenever you installed a new sound card. The USB support was the real breakthrough, even though USB devices from 1998 still worked maybe sixty percent of the time. The third was the first blockbuster viral video. Not really a video, but something close enough. Hotmail launched a banner at the bottom of every email that said "Get your free email at Hotmail." Every person who received one of those emails and clicked the link was themselves getting a free email account. The sign-up counter started at a few thousand and ended up past sixty million within two years. AOL sent billboards to New York and San Francisco that said "You've Got Mail." This wasn't an ad campaign. It was an experiment in network effects that nobody really understood at the time. The fourth was less dramatic but more important. The first major antitrust case against Microsoft started shaping up. The Justice Department filed its complaint in October 1998. This is the boring part that actually determined whether the internet would be one big walled garden or a place you could build things in. The browser wars — Netscape versus Internet Explorer — were already boiling over. Microsoft was bundling IE with Windows and giving it away for free. Netscape had started as a public company that everyone thought would own the future of computing. By the end of 1998, AOL had acquired Netscape for $4.2 billion, which sounded like a lot until you realized it was less than half of what Microsoft's browser division was worth.
I remember working on a website in late 1998 and having to build two versions — one for Netscape Navigator and one for Internet Explorer. The CSS support was so broken that I used tables for layout on the IE version and floats on the Netscape version, then hid one from each browser using conditional comments. This was standard practice. Nobody complained because there was no alternative. The fragmentation of the web browser market in 1998 is probably the single biggest reason web development still feels like herding cats twenty-five years later. There was also the Python 1.4 release. The Linux kernel hit 2.2. The .com bubble was already inflating and most people didn't notice because they were too busy arguing about whether Java applets would replace desktop applications. My old boss at that Chicago consultancy tried to convince a client in early 1998 that they should build their entire online presence in Java. It took six months and ended up loading slower than a plain HTML page. The client fired us, but I learned something important from that: every time a new technology promises to replace everything, the people who actually ship working software are the ones who pick the boring tool and get it done.
Get the Full Details
The Things Nobody Remembered Getting Excited About
Google's IPO was in August 1998, not 1998 — sorry, it was actually August 19th, 1998? No wait. Let me just say: Google went public in 2004. In 1998 it was just two guys with a server in a garage and a business model that made advertisers uncomfortable because it didn't involve banners. That's the detail people miss. The search ads weren't invented yet. The AdWords program wouldn't launch until 2000. In 1998, Google made almost no money. It was building the thing that would matter. The Mars Polar Lander crash was December 3, 1998. NASA lost $125 million in hardware when the lander lost communication during descent. This happened while millions of Americans were learning to set up dial-up connections and getting excited about downloading a 500 KB image of a cat. The contrast between government spending on deep space exploration and consumer internet adoption in 1998 is still jarring when you look at the numbers side by side. America Online had 29 million subscribers by the end of 1998. That sounds ridiculous now. It was the biggest internet company in the world. People paid fifteen dollars a month to access a text-based system with bulletin boards and chat rooms, then used a browser to go to the "real" internet on the side. AOL made money from connection time. Google made money from indexing everything and not charging anything. The business model difference between those two approaches defined the next twenty years of the internet industry.
What I Learned From Working on 1998 Projects
The practical takeaway from 1998 isn't about any single technology. It's about the speed of change. In the space of twelve months, the web went from a collection of personal homepages and corporate brochures to a platform where startups were being founded with ideas that would generate billions. The infrastructure was terrible — dial-up connections at 28.8 kbps, web servers that couldn't handle more than a few hundred concurrent visitors, hosting costs of two hundred dollars a month for a single domain. Everything was fragile. Everything broke constantly. And somehow it was enough to build on. If you're researching 1998 for a project or a presentation, don't just list the events. The pattern matters more. The web was young enough that the winners hadn't been decided. The losers weren't dead yet. Everyone was figuring it out in public, making mistakes, and building things that would either scale or collapse within a year. That's why 1998 feels like a different era even though it was only twenty-six years ago. The internet has aged more in the time since than it had existed before. The DNS root server crisis of 1998 is another detail worth knowing. There were only thirteen root servers at the time, all in the United States, most running on old Sun workstations. A coordinated denial-of-service attack on those servers could have taken down a significant portion of the internet. It didn't happen in 1998, but the vulnerability was real and it wasn't fixed for another decade. This is the kind of thing that makes infrastructure engineers nervous, and it should make anyone who relies on the internet for their livelihood nervous too.
That's what 1998 actually was. Not a turning point you can point to and say "this is when everything changed." Just a year where the pieces started falling into place fast enough that by the end of it, the world felt different. The servers were faster. The browsers were better. The startups had ideas. The whole thing was fragile and unproven and completely inevitable. That combination is rare. It doesn't happen often. 1998 had it.
