Summarize this content to 2000 words in 6 paragraphs in Arabic Strange, isn’t it, the things that haunt us? Not the large incidents or the endings, which weigh on us without mystery. What I mean are the loose ends, the partial histories that linger in our minds because they lack resolution or explanation.In my case, on the evening of December 15 2023, a taxi appeared in front of me on Wythe Avenue in Brooklyn at the exact moment I exited an apartment building completely blitzed on champagne and holiday cheer. I got in, put in my earbuds and hit shuffle on Spotify. “Maps”, the indie-rock love ballad by the Yeah Yeah Yeahs, came on. I hadn’t heard it in maybe a decade and, just as I marvelled at how shuffle had managed to unearth a deep cut, I saw my ex’s building, dark except for a band of second-storey windows shining thin light into the street. He was home, probably with someone else.My mind raced. What had made the taxi arrive when it did? What had compelled the driver to hang this particular left, which wasn’t conducive to the destination? What was going on with Spotify? Instinctively, I tried to piece meaning into these elements, despite understanding, very clearly, that their constellation was accidental and, in fact, random. But actually, what is randomness?The core principle of shuffling is, of course, randomness. Specifically, randomness in a statistical sense, which is defined as “involving equal chances for each of the actual or hypothetical members of a population”. Is there anything more absurd than googling “is randomness real?” Unfortunately there is, and it involves asking ChatGPT if “true randomness exists in the universe”. I’m in so deep that this seemed like the quick and dirty way to establish an overview of the topic. From what I can tell, it is unclear whether randomness is only a concept or an undeniable component of the physical world.According to writer and technologist James Bridle, “True randomness is a slippery thing. It is a property not of things in themselves, like individual numbers, but of their relationship to one another. One number is not random; it only becomes random in relation to a sequence of other numbers.” Personally, the difficulty of all this comes down to the fact that it is impossible to determine if an event is truly random or the product of a larger mechanism that, at present, eludes human comprehension.Spotify’s first iteration of its shuffle feature was dictated by a decades-old algorithm that generated unbiased randomness from a finite sequence of elements. Breathtakingly efficient, the Fisher-Yates shuffle, developed by Sir Ronald Aylmer Fisher and Frank Yates in 1938, was employed by Spotify to dismantle user playlists and reassemble them into new, unpredictable orders. From the developers’ perspective, the task of creating this feature was masterfully accomplished with just a few lines of code. From early users’ perspective, shuffle was a travesty.This discrepancy was bewildering for both parties, but mainly for the developers, who had objectively delivered a mathematically perfect version of randomness. Perfection turned out to be the problem. The algorithm captured a Platonic ideal of randomness instead of one compatible with the human mind. As a species, our intuitions about randomness are tragically incorrect. Evolution has sculpted our brains into pattern-seeking organs and, because of this, we presume that randomness must always present as chaotic. However, as randomness is unpredictable, it can (and will) at times give the impression of order.In the case of the Fisher-Yates shuffle, users expected the randomness to behave in a certain way that excluded any sign of organisation. For instance, clusters of songs from the same album or by the same artist were interpreted as evidence of shuffle’s dysfunction. But a shuffled playlist without clusters is statistically impossible. In a playlist of 100 songs shuffled by this algorithm, each has an equal chance of occurring in the first track position and so on through the list, ultimately meaning it is reasonable (and random) for several Taylor Swift songs to play back to back. When users complained about clustering, or shuffle not being random, what they were complaining about was reality.In a 2013 lecture titled “The Search for Randomness”, mathematician Persi Diaconis explained that coin tosses are a problem of physics and not one of probability. Essentially, a coin toss can be reduced to its revolutions, velocity or movements in physical space, which are objectively quantifiable. Once values are inserted into these variables, “Newton tells us the rest,” Diaconis said, and the mystery of heads or tails is no longer 50/50. To prove this, he built a cointossing machine that could perfectly replicate the conditions of a toss 100 per cent of the time, leading to fully predictable outcomes. Humans, of course, do not have the same level of dexterity as machines, though some exhibit above-average muscle co-ordination.Now, picture the Monte Carlo Casino, August 18 1913. Underneath a decadent golden chandelier, a frenzied mob of gamblers encircles a roulette table, desperate to place bets on red. The ball has landed on black an unprecedented number of times, so it seems a red streak will erupt at any second. Except, of course, it doesn’t. This is a strictly human projection. A roulette wheel, being inanimate, has no memory or desire to change its behaviour based on previous events. Later this error in judgment will be known as the Monte Carlo fallacy, or the gambler’s fallacy. According to the OED, it is the inaccurate belief “that the probability of any particular outcome to one of a series of repeated but independent events of chance is inversely dependent upon the previous outcomes”When Spotify users interpret the appearance of clusters of songs to mean that shuffle isn’t random, they’re suffering from the same ailment as the gamblers in the casino. On that night, the ball fell on black 26 times consecutively, an event that has the probability of one in 66.6 million.It occurs to me that my moment in the taxi may not have been random or an issue of probability. It was a problem, rather, of physics. If a woman travelling north at 30mph taps on her phone to initiate a cascade of electronic operations that result in the playing of a love song that happens to begin at the exact coordinates where she lost her heart, will a rip be created in space and time?To quell the backlash against the Fisher-Yates shuffle, Spotify re-engineered its algorithm to better harmonise with the human perception of randomness. In other words, the new algorithm was fashioned to stimulate the Monte Carlo fallacy. It purposefully eliminated clusters and distributed artists across a shuffle in a proportional manner to create a smoothly randomised arrangement of music. Further details were uncharacteristically and publicly shared on the platform’s blog in 2014. Unsurprisingly, this version of shuffle was also maligned, not only on account of its inauthentic nature, but also because many users felt it still failed at its job. But that was more than 10 years ago.Spotify’s algorithm captured a Platonic ideal of randomness instead of one compatible with the human mindToday, what Spotify shuffle is and how it works is anyone’s guess. The company is notoriously tight-lipped (it did not respond to a request for comment as to how its shuffle algorithm operates). As a result, theories abound across the internet. Videos on YouTube offer homegrown experiments that attempt to reverse-engineer the feature and, while none of this is conclusive, the suspicion is that shuffle malfunctions because of “payola”, the system by which record labels pay to have songs promoted on influential playlists. I mean, it’s 2025. We know how the world works. Alternatively, shuffle might be under the influence of matrix algorithms weighted towards frequency. The technology is eating itself.The only thing unequivocally random about shuffle is where you happen to be in your life — geographically, emotionally, psychologically — when you hear a song.Computers are incapable of producing true randomness. As the website random.org explains, “Surprising as it may seem, it is difficult to get a computer to do something by chance. A computer follows its instructions blindly and is therefore completely predictable. (A computer that doesn’t follow its instructions in this matter is broken.)” In other words, computers are designed precisely to eliminate randomness, and yet randomness is required for many applications, including the encryption and decryption of data. The workaround has been the advent of pseudorandom number generators, algorithms that conjure randomness from a “seed” or initial value, which is then manipulated by a deterministic process. However, if the seed or process is known, the “randomness” can be reverse-engineered, aka hacked.True random number generators are almost impervious to this, but they require a somewhat mind-bending aspect: an instrumental hook-up to the environment. This could be a microphone on a roof that picks up atmospheric noise or a Geiger counter monitoring radioactive decay. In at least two instances, the undulations of lava lamps have been recorded by cameras. The generators extract certain data points from these signals, which are then transmuted through computation into seeds. Selected from highly volatile phenomena, the information is virtually unhackable. When shuffle does what it is supposed to, it shields us from the seething mess of the real world. Our catalogue of Liked Songs is, in many ways, the story of our lives. Certain tracks not only reflect different personal and geopolitical contexts, they also become formalised into our own mythology, expressing ideas and visceral emotions we long to articulate but cannot. When we hear them as we go about our day, they press us forward into action, responsibility and reverie, providing reliable support in a way unmatched by most things in reality. There’s a solidness to this and a power. The present moment is being soundtracked to what we already know. Outcomes become more predictable, coherence is injected into the unknown.“Maps” is a song of only a few words, yet they manage to convey the unspoken seed of every relationship: “Wait, they don’t love you like I love you.” This phrase throbs, not so much as a chorus but a transmission. Hearing it repeat as I saw my old life slip away may or may not have been random. But I knew, at least, that it was true.Heather McCalden is an artist and writer. Her memoir “The Observable Universe” is published by Fitzcarraldo EditionsFind out about our latest stories first — follow FT Weekend Magazine on X and FT Weekend on Instagram
