Strong recommend

Algorithms can help you find things — but serendipity makes them unforgettable

Barcelona (Creative Commons)

Let’s call it a learning experience. On one of my first parent-free trips abroad, I did something that would make present-day Jack simply cease living from cringe: I asked the hotel receptionist for a restaurant recommendation. Look — it was late, I was young and we were staying in a not-so-great part of Barcelona.

The man behind the desk swiftly picked up the phone, made a booking and scribbled down the address on a piece of paper. On finding the restaurant — I don’t even know whether my phone had Google Maps functionality at the time — we could see it was completely empty, save for one rather forlorn-looking couple. This wasn’t the vibe.

We walked on, and soon hit a bar that was heaving with twenty-somethings. We sheepishly opened the door (I was young enough to still be worrying that the first restaurant would report our no-show to the hotel and there would be… consequences?) found the last table and enjoyed what remains the best huevos rotos of my life.

I mention this partly because I am contractually obliged to begin every newsletter with a story about food, a quote from Tony Blair or a reference to the fiscal cost of repeated fuel duty freezes. But also because human recommendations have been flawed long before the rise of algorithms.

I Choo-Choo-Choose You

Algorithms get a bad rap for reasons you are already aware. They optimise for engagement, not truth, they create filter bubbles, amplify extremes, encode bias, blur editorial responsibility, shift power to platforms and encourage passive consumption over active seeking.

At the same time, they’re just a new method of recommendation. One that solves the problem of abundance (without any sort of system, Spotify might just default to suggesting already popular artists), they personalise at scale, enable discovery and reduce friction. These are all really useful qualities! But like people, algorithms are not neutral. It all rather depends on what they’re optimising for.

On YouTube, the goal has largely been to maximise watch time and retention. On Facebook, it’s engagement and advertising and on Amazon it’s conversion and sales. My Catalan receptionist had his: financial — either for him or his mate. To be fair, whether the food would have been good, bad or indifferent, my presence did not exactly scream ‘repeat customer’.

But I was optimising for vibe, intuition and discovery. For that reason, if the receptionist had recommended the huevos rotos place, we still would have eaten well. But the experience would have been quite different. It wouldn’t have been a story, or a life lesson, even a moment of destiny.

You see, I stumbled upon that place, and was proud of the achievement in the way your spouse might be terribly pleased with themself for assembling a piece of IKEA furniture, even if the drawers don’t quite align and the nails inflict tiny scars.