When software ecosystems die
How much can you rely on the frameworks, tools and libraries you build your software on? And what can you do to reduce the inherent risk of depending on someone else's software?
Years ago I watched a whole software ecosystem die.
Not the slow decline of a programming language that is losing its users, or a no longer maintained library that has a newer, incompatible replacement. This was perma-death: game over, no resurrection, no second chances.
Here's what happened, and what you can learn from it.
The story of mTropolis
Back in the 1990s the Next Big Thing was multimedia, and in particular multimedia CD-ROMs. The market leader was Macromedia Director, a rather problematic tool.
Macromedia Director started out as an animation tool, using a sequence of frames as its organizing metaphor, which meant using it for hypermedia involved a rather bizarre idiom. Your starting screen would be frame 1 on the timeline, with a redirect to itself on exit, an infinite busy loop. Remember this started as animation tool, so the default was to continue on to later frames automatically.
When you clicked on a button that took you to a new screen it worked by moving you to another frame, let's say frame 100. Frame 100 would have a "go to frame 100" on exit to made sure you didn't continue on to frame 101, and then 102, etc.
Then in 1995 mTropolis showed up, a newer, better competitor to Director. It was considered by many to be the superior alternative, even in its very first release. It had a much more suitable conceptual model, features that were good enough to be copied by Director, and a loyal fan base.
In 1997 mTropolis was bought by Quark, maker of the the QuarkXPress desktop publishing software. A year later in 1998 Quark decided to end development of mTropolis.
mTropolis' users were very upset, of course, so they tried to buy the rights off of Quark and continue development on their own.
The purchase failed. mTropolis died.
Market leader or scrappy competitor?
The story of mTropolis had a strong impression on me as a young programmer: I worked with Director, so I was not affected, but the developers who used mTropolis were dead in the water. All the code they'd built was useless as soon as a new OS release broke mTropolis in even the smallest of ways.
This isn't a unique story, either: spend some time reading Our Incredible Journey. Startups come and go, and software ecosystems die with them.
Professor Beekums has an excellent post about switching costs in software development. He argues that given the choice between equivalent market leader and smaller competitor you should choose the latter, so you don't suffer from monopoly pricing.
But what do you do when they're not equivalent, or it's hard to switch? You still need to pick. I would argue that if they're not equivalent, the market leader is much safer. Macromedia was eventually bought by Adobe, and so Director is now Adobe Director. Director was the market leader in 1998, and it's still being developed and still available for purchase, almost 20 years later.
mTropolis may have been better, but mTropolis wasn't the market leader. And mTropolis is dead, and has been for a very long time.
Making the choice
So which do you go for, when you have the choice?
If you're dealing with open source software, much of the problem goes away. Even if the company sponsoring the software shuts down, access to the source code gives you a way to switch off the software gradually.
With Software-as-a-Service you're back in the realm of choosing between monopoly pricing and chance of software disappearing. And at least with mTropolis the developers still could use their licensed copies; an online SaaS can shut down at any time.
Personally I'd err on the side of choosing the market leader, but it's hard to give a general answer. Just remember: the proprietary software you rely on today might be gone tomorrow. Be prepared.