Technology is continually changing, sometimes evolving, sometimes just mutating with no apparent advantage until it finally dies off, leaving it's happless followers scrambling to find a way forward.
In the animal world this would be the equivalent of the Passenger Pigeon. I know you may not have heard the tragic tale of the Passenger Pigeon so read this carefully, pay attention to the numbers here. The Passenger Pigeons were so numerous at their peak that they would turn day into night as they shut out the sun when they passed overhead in enormous flocks. Estimates put the peak population in the US at as many as five billion passenger pigeons. The enormous flocks, during migration, could span a mile (1.6 km) wide and 300 miles (500 km) long, taking several days to pass and probably containing two billion birds.
The species went from being one of the most abundant birds in the world to extinction during the 19th century when gluttonous, would be hunters, took their sport by shooting aimlessly into the clouds of pigeons with their shot guns, leaving the dropped birds where they fell. The Passenger Pigeon being a gregarious animal could not sustain it's population once it fell below a critical threshold. It needed lots of company to be healthy and breed, a serious evolutionary flaw.
The last Passenger Pigeon, named Martha, died alone at the Cincinnati Zoo at about 1:00 pm on September 1, 1914. How is it possible that something so prevalent, so ubiquitous that it's flocks blacked out the sun and turned day into night, could disappear from the face of the earth? The take away of this story is simple, ubiquity is not a guarantee of longevity, and there are plenty examples of ubiquitous technologies and platforms that have disappeared.
The killing off of a technology is much less tragic, although to those affected, those that have "invested years and tons of manpower developing an application" in say Visual Basic, it can be very tragic but there is always a way to dig yourself out, as long as you have the investment capital as a shovel.
So a clear success factor for any CTO is being able to tell which changes are dead enders and which are ones to follow. Looking back over recent history it's clear there are no fool proof guidelines for weeding out the dead enders from the evolutionary changes. But the example of Visual Basic or VB6 as it is known provides a good example of a technology that became very popular but in the end took on the characteristics of a dead ender.
This from the Microsoft site: "The Mainstream phase will be in effect for six years after the product's general availability date. Visual Basic 6.0 was generally available in January 1999. Mainstream support will end March 31, 2005". If you are not aware of the fire storm created by this edict see http://classicvb.org/.