GoLike All articles
Internet Culture

Please Send Me Your Worst Recommendation: In Defense of Messy, Biased, Gloriously Imperfect Taste

GoLike
Please Send Me Your Worst Recommendation: In Defense of Messy, Biased, Gloriously Imperfect Taste

Photo by Photo by Vitaly Gariev on Unsplash on Unsplash

The other day a friend told me to watch a movie that was, by every objective measure, not good. It was long. The dialogue was weird. The ending made no sense. I loved it.

That's the thing about recommendations from humans. They're not optimized. They come with bias, history, and a little bit of risk. And that's exactly why they're still the best discovery engine we have.

The Algorithm Never Embarrasses Itself

Your feed is designed to be agreeable. It studies what you click, what you linger on, what you save, and then it serves you more of the same. That's useful if you want another show like the one you just finished. It's less useful if you want to be surprised.

Algorithms are safe. They don't have a friend who insists you'd love a band that sounds like a lawnmower. They don't have an uncle who swears by a diner off exit 47. They don't have a coworker who sends you a link to a 40-minute video essay about a topic you've never considered. Those people are not optimized. Thank goodness.

A Bad Recommendation Is a Tiny Act of Trust

When someone recommends something to you, they're not just sharing a link. They're saying, 'I think you might like this, and I'm willing to be wrong in front of you.' That's a small vulnerability. It's also a bid for connection.

A perfect recommendation from an algorithm costs nothing. A weird recommendation from a friend costs a little pride. If you hate it, they have to hear about it. If you love it, you have a new inside joke. Either way, you have a conversation.

The Weird Stuff Is Where the Good Stuff Hides

The best recommendations I've gotten in the last few years were not the obvious ones. They were the ones that made me raise an eyebrow. A specific brand of frozen dumplings. A hiking trail with no view until the last five minutes. A book about a subject I didn't know I cared about. None of those would have landed in my feed, because I hadn't told the algorithm I was interested yet. That's the point.

Humans can see around corners. They know your history, your moods, your weird little preferences. They can connect two things that don't look connected. An algorithm can do that too, but it's doing it with data. Your friend is doing it with memory.

Let Your Friends Be Wrong

The internet has made us a little too good at curating. We untag bad photos. We delete old posts. We keep our feeds tidy. But taste doesn't develop in a tidy place. It develops when you try something that doesn't fit and decide whether you like it anyway.

So let your friends be wrong. Let them send you the song you'll skip. Let them drag you to the restaurant with the questionable carpet. Let them recommend a movie that everyone else forgot. The worst that happens is you have a funny story. The best that happens is you find something you love.

Your Assignment: Send One Weird Recommendation This Week

Here's the challenge. Before the week is over, send one person a recommendation that is deeply specific and slightly risky. Not a popular show everyone already knows about. Not a restaurant with a perfect rating. Something that made you think of them. Something you're not 100 percent sure they'll like.

Then do the harder thing: try one recommendation someone sends you. Not the safest one. The weird one.

We spend a lot of time asking algorithms what's next. Maybe we should ask each other more often. The best stuff is rarely perfectly tailored. It's a little messy, a little biased, and a lot more human. That's the kind of discovery GoLike was built for—and it's the kind we should keep alive.

So go ahead. Send the weird link. Make the questionable recommendation. Be wrong in front of a friend. That's how taste gets better, and how friendship gets weirder in the best way.

All Articles

Related Articles

15 Human-Picked Things to Try Before 2027 (No Algorithm Required)

15 Human-Picked Things to Try Before 2027 (No Algorithm Required)

Saving Is the New Flex: Why the Bookmark Button Became the Internet's Real Taste Test

Saving Is the New Flex: Why the Bookmark Button Became the Internet's Real Taste Test

I Reviewed My Suburb's Grocery Stores Like Theme Parks and One of Them Got a 'FastPass'

I Reviewed My Suburb's Grocery Stores Like Theme Parks and One of Them Got a 'FastPass'