One of my favourite stories about innovation starts with something surprisingly mundane: finding a parking spot.
Everyone knows that it is easier to find a parking space in an empty parking garage than in a full one. As occupancy approaches 100%, the search becomes disproportionately harder. The same idea has long existed in computer science. When a database becomes crowded, finding an empty space for new data also becomes increasingly difficult.
That made sense. At least, it made sense for about 40 years.
Then came Andrew Krapivin. In 2021, as a 21-year-old student, he looked at a problem that many people assumed was already understood. Little did he know that it couldn't be done significantly faster. So, he did it anyway. Not by a little, but by a lot: 100,000 times faster and more.
Krapivin’s story highlights an often-overlooked danger of patent literature. Patents are frequently described as a valuable source of technical knowledge. And they are. They contain a vast amount of information, creative ideas, engineering insights, and signals about where industries may be heading.
But patents also contain statements that may not be true. At the time of filing, applicants may not yet know whether their ideas truly work, or they labor under a misconception.
This creates a subtle risk. Patent data can teach us what someone believed, proposed, or wanted to be true. It does not always teach us what is true.
To make things worse, these false ideas work as an attractor for new ideas. Once you have seen a solution, it is hard to think out of the box.
Sometimes innovation starts when someone does not know what is supposed to be impossible. What you can't learn from patent data? Starting with a blank slate.
< Back to all blog posts