An Open Web

I saw my first web browser in 1993 when a friend called me down to his dorm room to show me a web page that the Vatican museum had put up. I’d been on the Internet for about a year using text-based tools like Telnet, FTP, Gopher, and Usenet, but the world of hypertext was an absolute eye-opener. I remember my first reaction to seeing the World Wide Web being, “Well, Gopher is done for”. It felt like a new world opening up.

I built my first web page about six months later, hosted on my 486 running Linux and connected to the Cornell University residential Ethernet. Campus Ethernet in the dorm buildings was still fairly experimental at the time, they just handed out a publicly routed IPv4 address and let you have at it. I wanted to explore and push boundaries, so one of the first things I built was a very basic CGI search engine for a corpus of Old English text that I’d gotten on a floppy disk from the Linguistics department, my first experience with making something actually useful for the World Wide Web¹. Not too long after that, I built a web site for a new student club I’d started, the Cornell University Classic Computer Club (a later snapshot of which, miraculously, still exists on the Internet Archive).

I think about these things today because all of that was made possible due to open standards at every layer, from TCP/IP to HTTP. Those same open standards made the explosive growth of the Web from 1995 onward possible, and it certainly was explosive.

We can build websites in 2023 too, of course, but the landscape feels like it’s changing. Naturally most people don’t build websites any more. Instead, they use the services of big commercial companies like Twitter or Facebook or Bluesky to publish content. In turn, those companies make use of the same open standards that built the web to make walled gardens. As of this writing, a lot of content on Twitter and Facebook and Bluesky is simply invisible if you don’t have an account there and you aren’t logged in. It feels very wrong.

Last week, Google’s “Web Integrity API” proposal made the rounds, and it’s already been built into Chrome without much discussion. It would make it possible for web servers to decline to serve clients they don’t trust. There are other ways to do this, of course — in some sense, user agent sniffing has already allowed a much weaker version of this for decades. But the Web Integrity API goes further, and reaction has been extremely negative. I’m not the only one who cares about the web being open.

I think the next few years are going to be crucial in trying to preserve an open web. I don’t know where we’re going from here, but I think projects like ActivityPub give me a little bit of hope. It’s still possible to build and use an open web, we just have to care enough to actually do it.

The thing I keep coming back to is that the general public rarely cares about these things, and they’re very happy to trade openness for convenience. That’s nothing new. It just reinforces in my mind the need to advocate on their behalf, to keep striving to do better, and to keep building open tools and platforms. If you’re in a position to advocate against walled gardens, please do. If you’re in a position to make content open, please do. Consider building a website, consider writing a blog, consider joining an ActivityPub service like Mastodon or Pixelfed. The only way we’ll keep things open is to actually use open standards.