The death of the food critic: why For-You feeds replaced four-star reviews

The death of the food critic: why For-You feeds replaced four-star reviews

The food critic profession did not die because critics were wrong. It died because, by May 2026, a 23-year-old in Queens with a phone could publish a verdict on a restaurant before the reservation hold expired, and 90 million people could agree or disagree with her by the time the print edition went to bed. Speed beat taste. It was never close.

I have been arguing this with food editors for two years, and the holdouts keep retreating to the same defense: the critic offers context, history, palate, a writer's voice. All true. None of it matters when the reader has already booked the table from a 14-second video at 11:47 p.m. the night before.

Pete Wells stepped down from the New York Times in 2024, and his farewell column read like a man describing the slow collapse of a building he loved. The Boston Globe cut its dedicated restaurant critic. The Chicago Tribune kept the title but stopped funding the travel. Eater's reviewers now openly cite TikTok creators in their copy, which is the equivalent of the cardinal asking the parishioners what scripture means. It is over. The question worth asking is what replaced it, and whether the thing that replaced it is better.

I think it is. Let me show you why.

Distributed palates beat single palates, every time

The classic argument for the food critic was epistemic. A trained reviewer eats five times a week for thirty years. She tastes things you and I cannot taste. Her single anonymous visit, multiplied across decades, produces a verdict closer to truth than any individual diner could approach. That was the bargain. We paid the salary; we got the calibrated palate.

The bargain broke in two stages. First, around 2018, the anonymous visit collapsed — every restaurant kitchen in Manhattan had a phone alert ready when a Times critic was spotted on the floor. Service tightened. Plates got walked. The "anonymous" verdict became a verdict on the restaurant's A-team. Second, around 2022, the multiplication argument got obliterated. One critic eating twice at a restaurant is a sample size of two. A TikTok For-You feed pushes a place that has been visited 400 times in six weeks by 400 different mouths, 38 of whom posted video. That is a sample size of 38 with continuous service variance, kitchen variance, server variance — and you, the reader, watching the variance directly.

I would rather see eight people eat the cacio e pepe than read one person describe it. The eight people cannot agree on a script. The eight people cannot all have the same allergy, the same date energy, the same Tuesday-vs-Saturday window. The aggregation is more honest than the curation ever was.

The food critic's response, when I have pushed this in conversation, is that aggregation produces tyranny of the median: every restaurant becomes the same restaurant because the algorithm rewards the dish that photographs well at 9 p.m. There is something to this. The fluffy Japanese pancake had a moment in 2023 that I do not want to relive. But the critic argument assumes the alternative — the four-star review — is not itself a tyranny. It is. It is the tyranny of one Manhattan-trained palate, usually trained at the same five tasting menus, applied to a city with 8,000 restaurants. The TikTok median is at least a median of 8,000 visits a week. The Times median was a median of one visit a year.

The takeaway: distributed review is noisier, but the noise is the signal. You are watching 38 different humans react to the same kitchen, and the variance tells you whether the place is consistent. A single critic cannot give you that, no matter how good her palate is.

The For-You feed is a better recommendation engine because it is a worse oracle

Here is what print food media misunderstood about TikTok for the entire 2020-2024 period: they thought it was a content platform. It is not. It is a recommendation engine that happens to use video as its payload.

A restaurant review in print answered the question: "Is this place good?" The For-You feed answers a different and more useful question: "Will you like this place?" Those are not the same question, and the second one is the one diners were actually asking the whole time. The critic answered it badly because she had no way to know who you were. The algorithm answers it well because it has watched you scroll for 600 hours.

I am not romanticizing the algorithm. It does ugly things. It privileges restaurants with photogenic interiors over restaurants with great food in fluorescent-lit rooms. It punishes places that cannot or will not allow filming. It feeds the same six dishes back to itself in a citation loop until every Italian-American spot in Bushwick is serving rigatoni vodka in a way that looks identical at 4K. These are real costs.

But notice what the algorithm does not do. It does not refuse to review a place because it is "not on the critic's beat." It does not wait three months between visits. It does not give a single dish a single sentence and call that coverage. It does not require the restaurant to mail a press kit. It does not insist that the chef have trained in a French kitchen to merit attention. By 2024, the algorithm had already done more to discover undercovered cuisines — Sichuan-Hui hybrids in the San Gabriel Valley, the Senegalese stews in Harlem, the Yemeni mandi rooms in Hamtramck — than the entire American food press combined.

"This place is doing 90 covers a night and no critic has stepped through the door in four years."

A creator I follow, @sichuanboy, put that caption on a video of a strip-mall Chongqing restaurant in Rosemead last year. He was right. The Times still has not been to that restaurant. The For-You feed brought it 2.3 million views and a six-month reservation wait. That is what replacement looks like.

The takeaway: the For-You feed is not pretending to be the oracle. It is admitting that taste is local — to your palate, your budget, your neighborhood, your weeknight — and routing around the oracle problem entirely. The critic was trying to be right for everyone. The algorithm is trying to be right for you. Those are not the same job, and the second one is what the diner needed all along.

What survives, and what the next critic looks like

I do not think criticism dies. I think the role dies and the function migrates. The function — telling a reader whether the meal will be worth the money and the night — does not go away. It just gets unbundled.

Three pieces survive, each in a new home.

The first piece is taste history — knowing why the chef at a new Sicilian spot in Greenpoint is doing pasta alla Norma a particular way, what tradition she is echoing or breaking. That migrates to long-form Substack writers, food podcasters, and a small number of academic-adjacent voices who never had a print column anyway. Adam Reiner, Helen Rosner's newer essays, the people doing 4,000-word substacks on a single regional cuisine — they keep the lineage alive. The audience is smaller than the Times's was. The work is better.

The second piece is consistency reporting — whether a place that was great in January is still great in May. That migrates entirely to TikTok and Instagram. A creator who lives in the neighborhood and posts every two weeks is structurally a better consistency reporter than a critic who visits twice. This is the area where the print review was always most exposed; restaurants change in eight months, and a review written in March was already lying by November.

The third piece is trust calibration — the reader knowing whose opinion to weight and why. This is the piece that the For-You feed handles worst, and it is the opening for what comes next. The algorithm will surface a video; it will not tell you whether the creator has a sponsorship deal, whether the restaurant comped the meal, whether the creator has eaten there twice or twenty times. The new critic is the person who builds that trust-graph on top of the algorithm — not replacing the videos, just annotating them.

That third piece is part of what we are building at GeoTok. Our job is not to second-guess the creators; the creators are doing the reviewing. Our job is to organize their verdicts by place, neighborhood, and consistency over time, and to put them on a map so you can see which Brooklyn block has nine creators converging on the same omakase room and which has one promotional post in a sea of silence. The video stays the review. We just make it possible to walk down a street and know which doors have been opened.

A note on the dollar question, since I get it every week: yes, this changes restaurant economics. A four-star Times review in 2008 was worth roughly $1 million in incremental revenue over a year — that figure circulated widely in industry analyst circles and was treated as a back-of-envelope benchmark. By 2024, the equivalent inflection — going from 12 thousand TikTok views a week to two million — was worth more than that, faster, and to a different demographic. The critic's economic power did not vanish; it diffused. A thousand creators each move a slice of it. That is not a tragedy for restaurants. It is a redistribution.

The takeaway: criticism does not die. It splits into history, consistency, and trust — and only the first of those still wants to live in a newspaper.

What this means for how you eat next week

If you are still reading restaurant reviews to decide where to eat tonight, you are doing it wrong, and you have been doing it wrong since roughly 2021. The reviews are a beautifully written record of a meal somebody else had eight months ago. They are not a forecast of the meal you will have tomorrow. Use them for what they are good for — context, history, the writer's voice — and stop letting them tell you where to spend $180 on a Saturday.

For the actual where-to-eat decision in May 2026, the workflow is: open a video feed, watch six creators, find the place that gets shown three times by people whose taste you have already learned to trust. Then check whether the place is still doing the thing the videos are about — last week, not last quarter.

That is the workflow we built GeoTok around. When you see a place in a video and want to know what every other creator has said about it, where it is, and whether it is still operating the same kitchen tonight, the app is the answer. One tap from the video, not three searches and a screenshot.

The professional food critic did beautiful work for ninety years. The people doing her job now do it faster, in more places, in more languages, with more sample size and less pretense. That is the trade we made. I think we got the better end of it. Published May 2026, written from the future the critics did not see coming.