Skip to main content

Longevity of software

Reading Windows 95 — How Does it Look Today? made me think about how few internet services have survived over the past 25 years.

When the author tested the browser built into Windows 95, the default URL – – failed to resolve. This is a good example of how URLs regularly change. At some point in the last 0-25 years, that URL was broken, probably earlier than later. Broken references are a problem in academia even with documents that have DOIs which are expected to remain available forever. They don’t.

Major websites like Google and Medium also didn’t work. We have become used to building highly complex websites with all sorts of shiny new technologies. HTML and CSS and JS and React and icon fonts and WebAssembly…layers upon layers upon layers. This has allowed the development of sophisticated applications that can work amazingly well in the browser, such as Google Sheets or Figma, but do we need that advanced level of technology for everything?

What technology that we rely on today will simply cease to work in a few years? This made me think about what services and platforms we use today that we consider crucial to our everyday life. Facebook Messenger, Skype, WhatsApp, WeChat, Twitter, Reddit, Signal, Slack. In 25 years I expect few of these will exist. This is not an argument for ignoring progress, but more an observation about how rapidly technology becomes obsolete.

This is why open protocols are important. Email is a good example of a technology that was around 25 years ago, and will likely still be around in 25 years time. Why? It’s an open protocol. Anyone can build an email client, run an email service, manage their own email servers. Clients have come and gone. GMail is all mighty today, but new, innovative services are showing up.

This is because the core protocols still work fine. The Windows 95 post explains how the author was able to connect to the internet (using TCP/IP), and ping still worked. I bet a Windows 95 email client would work too.

The underlying technology might be moving rapidly, but what about the services built on top of it? Services shipped with Windows 95 no longer work – ICQ, AOL Messenger, MSN. All shut down. But how likely is that today? As the internet scales and more and more people get locked into proprietary systems, perhaps it is becoming harder and harder for new systems to displace the old:

Leaving Facebook in the 21st century is like my grandmother leaving the USSR in the 40s: you can go, but your friends and loved ones are all held hostage befhind Zuckerberg’s Iron Curtain, so leaving Facebook means leaving your communities, your relationships. That’s not as hard as kicking opioids, but it’s not easy either. And your presence on Facebook is the reason someone else can’t go.

IP, Cory Doctorow

It’s still possible for proprietary systems to be overturned – a sudden shift, like with the WhatsApp Privacy Policy changes can help – but the power of proprietary platforms operating at scale makes today quite different from 25 years ago.

We don’t need more platforms, we need better protocols (and products built on top of them). This is where government should focus their efforts – not on insignificant fines, not on trying to break up companies that would likely have no effect. Government regulation should stay away from the details of how tech companies operate and act at a higher level by forcing an open approach to data, protocols and networks. Interoperability is the key that helps level the playing field.