How ‘hyperselection’ is making modern life efficient, pleasant – and boring

It’s difficult to imagine a song like Queen’s 1975 Bohemian Rhapsody topping the charts today. Almost six minutes long, it moves through several radically different musical sections and famously ignores the conventional structure of a pop song. It seems almost designed to break the rules of what a hit single should be. Fifty years later, popular music seems to have moved in the opposite direction. Once artists and producers figure out what makes a commercial hit – from its length and structure to its rhythm, melody and how quickly it reaches a catchy hook – songs can be optimised to reproduce them. This tendency is especially pronounced today because streaming platforms allow millions of songs to compete for our attention. This creates intense pressure to stick to formulas that are known to work rather than risk experimenting with something different. As a result, the musical landscape becomes filled with songs that sound alike. This is an example of what my colleagues and I call hyperselection in a new paper published in the journal Frontiers in Psychology. This idea refers to what happens when competition becomes so intense, and we become so adept at identifying and optimising what succeeds, that we repeatedly select for the same characteristics while alternatives are eliminated. When we become too good at figuring out what works Beyond music, hyperselection is making many parts of our world more efficient and optimised while simultaneously making them more predictable, mediocre and sterile. In professional football, teams constantly search for tactics that give them an advantage. Once a strategy proves successful, other teams follow suit. The game becomes increasingly competitive and tactically sophisticated, but teams also become all the more similar in how they play while leaving less room for individual flair. The same happens in academia. Scientists are rewarded for publishing papers, receiving citations and attracting research funding. These are reasonable indicators of scientific performance. But intense competition can encourage researchers to pursue studies that are easier to conduct and more likely to be published, rather than risky ideas that might fail but could also produce important breakthroughs. Urban settings provide another example. Shops, restaurants and developments that are commercially successful are copied and reproduced, while those that cannot compete disappear. Cities thousands of kilometres apart can therefore become filled with the same Starbucks, Apple stores and luxury malls. Variety is valuable One consequence of hyperselection is the loss of adaptive variance. This refers to the variability that comes from allowing different products, behaviours and strategies to coexist, including those that may not be the most efficient. This matters for more than whether our music, cities or football matches become boring. Most of human evolutionary history unfolded in wild and messy natural settings, and we evolved to respond to such conditions. Modern societies, however, have become increasingly effective at creating cleaner and safer environments. Sanitation, antibiotics and other advances have greatly reduced infectious disease. Yet exposure to a diverse range of microorganisms also plays a role in the development of our microbiome and immune systems. Excessively sanitised environments may therefore remove some of the microbial diversity that our bodies need. Something similar may happen with psychological wellbeing. Modern environments are designed to minimise risks and maximise comfort and convenience. At the same time, experiences of difficulty, uncertainty and failure provide opportunities to develop coping skills and resilience. By removing the chance to encounter such experiences, particularly during the formative years of childhood, people may become less capable of dealing with them when they inevitably occur. AI could accelerate hyperselection Artificial intelligence (AI) could intensify hyperselection because optimisation no longer needs to proceed at the speeds afforded by humans or previous technologies. Algorithms analyse the behaviour of millions of people and continuously learn which songs, videos, products or other content attract the most attention. Generative AI can then produce enormous quantities of new content based on patterns that have already proven successful. This creates a feedback loop. AI identifies what works, produces or recommends more of it, and uses the resulting information to optimise its outputs even further. Importantly, much of this process can occur without humans having to identify the successful formula themselves. As AI becomes integrated into more areas of life, hyperselection could therefore occur faster and across a much wider range of domains. Despite having access to an unprecedented volume of information and content, people may find that everything from entertainment and news to advice and creative work begins to look, sound and feel increasingly alike. What can we do about it? The lesson is not that optimisation is bad. Making things safer, cleaner and more efficient has produced enormous improvements in human life. Rather, we need to recognise that there can also be costs to optimising everything. One way to address this is to treat the preservation of adaptive variance as a design principle. For example, universities could make room for risky research rather than rewarding only predictable scientific output. Cities could preserve unique local businesses and spaces even when they are less commercially efficient. More generally, we can create more room for exploration and mastery, rather than treating every failure as something to be avoided. If we become too good at eliminating everything that does not work, we may also lose the very thing that makes life interesting, enriches our experiences, and provides the conditions for growth and flourishing.