In February 2005 three friends—Chad Hurley, Steve Chen, and Jawed Karim—created a video-sharing platform. Karim uploaded the inaugural video on the platform titled Me at the zoo on April 2005. In a little less than two years Google acquired the platform for $1.65 billion. YouTube now operates as one of Google’s subsidiaries.

Fascinating story indeed, but here’s the interesting bit. Had the founders tried to execute the exact same idea ten years earlier it wouldn’t have been so successful. In fact, most likely it would have been a spectacular failure.

For starters, the vast majority of web users were on painfully slow dial-up connections in 1995. Sometimes it would take minutes to download a small image. The average two-minute-long YouTube clip would have taken as much as an hour to download on then-standard 14.4 bps modems. Nobody has that kinda patience.

Another key factor to YouTube’s success was that its video service was based on Adobe’s Flash platform which had already become the standard at that time. This meant that the founders could focus on the user experience of sharing and discussing clips, and not spend time and effort developing a whole new video standard from scratch. Interestingly, Flash itself wasn’t released until late 1996, and didn’t support video until 2002, thereby making the 2000s the ideal time for a video-sharing platform such as YouTube to come into existence. In the words of Isaac Newton, YouTube had the shoulders of giants to piggyback.

Let’s take another example. Charles Babbage, the father of modern computing created the blueprint of the Analytical Engine—the world’s first programmable computer—in the late 19th century. He toiled on it for the last 30 years of his life, but to no avail. Babbage had most of the system sketched out in his lifetime, but the first true computer to use this programmable architecture didn’t appear for more than a hundred years.

On the other hand, Babbage’s Difference Engine—a fantastically complex fifteen-ton contraption, with over 25,000 mechanical parts, designed specifically to calculate polynomial functions—debuted in 1855, and went into mass production within two decades.

Why did the Analytical Engine prove to be such a short-term dead end, given the brilliance of Babbage’s ideas, whereas his Difference Engine didn’t? Well, Babbage simply didn’t have the spare parts of such a machine back then. He was effectively sketching out a machine for the electronic age during the middle of the steam-powered mechanical revolution. In other words, Babbage didn’t have any giants’ shoulders to stand upon.

In Where Good Ideas Come From, Steven Johnson explains the idea of the adjacent possible pioneered by scientist Stuart Kauffman to explain the set of first-order combination of spare parts that lead to innovation.

Johnson suggests that ideas are constrained by the skills and technologies that surround them. We have a natural tendency to romanticise breakthrough innovations, but ideas are structures that can be created only with the available parts. Real world breakthrough happens only when we take what is available with us and put them together to create a something new.

Babbage’s design for the Difference Engine was a work of genius, no doubt, but it did not transcend the adjacent possible of its day, whereas the Analytical Engine did. It was far ahead of its time—in the remote possible.

Similarly, YouTube would have been inconceivable in the 1980s since there was no home internet. It was impossible in the 1990s due to slow browsing speeds. It wouldn’t most likely have been possible in the 2000s had there not been video support on Flash. Video-sharing on the internet was an idea that was just waiting to happen as soon as the spare parts became available.

“The adjacent possible is a kind of shadow future, hovering on the edges of the present state of things, a map of all the ways in which the present can reinvent itself,” Johnson writes.

By this logic, we can think of innovation as opening a door. We begin in a room with four doors, each leading to a new room that we haven’t visited yet. Those four rooms are the adjacent possible. But once we open one of those doors and stroll into that room, three new doors appear, each leading to a brand-new room that we couldn’t have reached from our original starting point. If we keep opening new doors, we will eventually have built a palace.

Framing problems as the adjacent possible is liberating. The founder of Shopify Tobi Lütke once said: “Predicting the future is easy, but timing it is hard.” Sure we know that autonomous vehicles are coming, but we have no idea exactly when they’ll become ubiquitous. Therefore, whenever you find yourself in a spot with your eyes on something that’s a few doors away, it’s time to reconsider your approach. However, not all doors are equal, so try to find the doors that would lead to the biggest possible expansion of the adjacent possible—be it with an invention or a startup.