— A Philosophical Reflection on the Dujjonku Craze and the YouTube Algorithm
One day, my feed changed.
I opened Instagram, and there it was: a round brown cookie, sliced in half, its cross-section on full display. I swiped through YouTube Shorts, and someone was biting into one with an expression of pure rapture. On TikTok, a line of people waited outside a shop in the cold. I scrolled past it the first time. The second time, I paused. By the third encounter, I was already typing “dujjonku” into the search bar.
Dubai Chewy Cookie — dujjonku for short. This dessert does not exist in Dubai. It was born in Gimpo, a city in Gyeonggi Province, South Korea. A confection of kataifi noodles and pistachio spread wrapped in marshmallow, it first appeared in April 2025 but averaged only about 60 online mentions per month through June. Then K-pop idols posted photos of it. Celebrity chef Ahn Sung-jae featured it on his YouTube channel. Within a single month, online mentions surged 23-fold. Delivery app searches spiked 1,500 times over. The fintech platform Toss built a real-time inventory map so users could track which cafes still had stock.
And within two weeks, the frenzy began to fade.
The Invisible Hand of the Algorithm
It would be easy to dismiss this as just another dessert trend. But a more fundamental question deserves attention: Did we eat dujjonku because we wanted it, or because we were engineered to want it?
YouTube’s recommendation system does not simply display content. It constructs a personalized sequence based on each viewer’s behavioral data. If I paused on a food video for more than three seconds, the algorithm surfaces more of the same. When one person pauses, the content reaches ten. When ten respond, it reaches a hundred. In this process, the quality or truthfulness of the content is secondary. What matters is dwell time — how long a pair of eyes stays fixed on the screen.
Short-form content is structurally optimized for this mechanism. The first three seconds must capture attention. Between fifteen and sixty seconds, an emotional response must be triggered. The moment a dujjonku is sliced open, the gooey stretch of marshmallow, the crunch of kataifi, the expression on the eater’s face — none of these transmit flavor. They transmit visual stimulus. What we consumed was not a cookie. It was an image.
Whose Desire Is It?
Jean Baudrillard argued that in modern consumer society, people do not consume products — they consume the signs that products represent. The university student who waited two hours at a department store: was she truly after the texture of kataifi, or was she after the social currency of an Instagram story captioned “finally got one”?
The question is uncomfortable, but it matters, because the structure of the dujjonku phenomenon reveals a complete marketing cycle. Short, visually stimulating videos spread through algorithmic amplification on social media. Celebrities validate the product, generating curiosity and aspiration. Portals and delivery apps surface related keywords. News outlets turn search volume and sell-out reports into headlines. Convenience stores and franchise chains release imitation products. And at the center of it all, FOMO — the Fear of Missing Out — completes the purchase decision.
How much room does individual free choice actually have within this cycle? Borrowing from Rene Girard’s theory of mimetic desire: we do not desire objects for what they are. We desire what others desire. The idol ate it, so I want to eat it. My friend posted about it, so I need to post about it too. The algorithm accelerates this mimetic impulse to a speed that previous generations of media could never achieve.
Accelerating Trends, Shrinking Experience
There is a striking dataset worth examining. The half-life of Korean dessert trends — the time it takes for public interest to drop by half after peaking — has been collapsing at a dramatic rate. In 2020, the croffle took 163 days to decline. In 2023, tanghulu took 54 days. In 2024, Dubai chocolate took 13 days. In late 2025, dujjonku took 17 days. In five years, the lifespan of a food trend shrank to roughly one-tenth of what it was.
These numbers do not merely reflect fickleness. They reveal a structural limitation of algorithm-driven consumption. When the primary purpose of eating something is not to savor the experience but to produce content — to film, post, and certify participation — the product’s utility expires the moment the upload is complete. Experience is not accumulated. It is exhausted.
The philosopher Byung-Chul Han describes this condition in terms of a “burnout society”: a state of perpetual stimulus-seeking in which deep engagement with anything becomes impossible. The two-hour line for dujjonku looks like passion, but it may in fact be a response to an algorithmically manufactured sense of scarcity. The act of waiting in line becomes content. That content enters the algorithm. The algorithm generates scarcity in someone else’s mind. Desire reproduces desire in a self-perpetuating loop.
The Shadow: Someone Pays the Price
Outside this loop, there is a material reality. Kataifi, the core ingredient, is entirely imported, and the craze drove marshmallow prices up more than fivefold. Small cafe owners rushed to add dujjonku to their menus to ride the wave, but ingredient costs pushed the cost ratio past 30 percent, leaving almost no margin. When the trend faded in two weeks, what remained were unsold inventories and discount listings on secondhand trading apps.
Small business owners now face a persistent pressure: instead of deepening their own craft, they must chase whatever the algorithm elevates next. The logic that governs survival is no longer “make something good” but “make whatever is trending right now.” This is not merely a problem for individual entrepreneurs. It is a structural distortion that algorithm-driven consumption imposes on the real economy.
So What Should We Be Asking?
Whether dujjonku tastes good or bad is beside the point. The real question is whether the decision to eat it was genuinely yours.
An algorithm is not a neutral tool. It allocates attention in specific ways. It amplifies specific kinds of desire. It produces specific patterns of consumption. In the 0.3 seconds it takes you to pause your scroll, thousands of data points are already at work determining what you will see next. Within this architecture, what does free will actually mean?
I am not arguing that we should reject algorithms. That would be impractical, and algorithms do provide genuine convenience. But a minimum level of awareness is necessary — the effort to distinguish whether the video that dominated your feed reflected your taste, or whether it was actively constructing your taste.
Croffles, tanghulu, Dubai chocolate, dujjonku. The names change, but the structure repeats. No one knows what the next trend will be. But when it appears on your feed, consider asking yourself one question:
Did this desire truly begin with me?
