We have already looked at restaurants one by one. This time we connect the restaurants visited by the same lawmakers. The 11 flagship restaurants in Yeouido were visited by 182 lawmakers, and each core restaurant pair shares 17 to 25 overlapping visitors. What appears is not a collection of separate popular places, but one restaurant zone the same lawmakers keep moving through.
Why relationships, not rankings?
In the previous analysis we lined the restaurants up one by one: the place with the most payments, the most-visited place, the place with the most money. Such ranking treats each restaurant as an independent dot. First and second simply sit above and below one another in a table; nothing tells you whether the same people used both.
Yet the data holds information one level above the dots. If the same lawmaker went to restaurant A and also to restaurant B, an invisible line forms between A and B. Draw all those lines and the ranking becomes a co-visit network. The question is no longer which place is first, but which restaurants attract the same group of lawmakers.
That distinction matters. Anyone can build a ‘Top 10 Most Popular Restaurants’ list from payment totals alone. But seeing which restaurants the same lawmakers visited together requires visit records by point of use. Totals show size; co-visits show overlap. This piece follows that overlap.
Not dots but connections — the co-visit map
Seen from a distance, eleven restaurants sit in a ring, linked by thick lines. Each line is the number of lawmakers who visited both restaurants. Thicker means more overlap; weak links below 17 were excluded from the picture. Even so, the frame is dense with lines. Yeouido’s core restaurants do not behave as separate islands; they are strongly tied to one another.
Read at the macro scale, the map is defined by convergence more than concentration. No single restaurant looms alone while the others scatter around it. Instead, large restaurants share the same visitors and gather into one mass. The strongest link is 25, and even the weakest link left in the graph is 17. The core has very little empty space.
The core figures — 182 visitors, and an unbroken 17–25
Two figures are the firmest this map yields. First, the flagship eleven were visited by 182 lawmakers in all. Second, the main links among those eleven stay in the 17 to 25band without breaking. In a group of around 300, having 182 people inside the same 11 restaurants means lawmakers’ lunch routes fall within a much narrower radius than the raw rankings suggest.
The 17–25 range is not just a number either. In a network, line thickness measures how strongly two nodes are tied. That no core pair drops below 17 means that, whichever two places you pick, at least seventeen lawmakers visited both. This is not one or two people overlapping by chance; one group is rotating across several restaurants.
Gasiri as the hub node
Zoom in one notch, and Gasiri sits at the center. With 76 visitors it is the largest node, and it overlaps by 25 lawmakers each with Namdomaru, Hwadam, and Ido Sikdang. That is why the most numerous and thickest lines gather at Gasiri. A hub is not simply the place with many visitors; it is the place that shares the most visitors with the others.
Behind Gasiri, Namdomaru (60) and Hwadam (60) form the second tier, with Ido Sikdang (50), Hanguk-ui Bapsang (46), Hallyugwan (45), Daebanggol (44), Izumi (44), Sohojeong (41), Unsan (40), and Donghaedo (33) around the outside. What matters is that even these outer nodes are not independent dots. Lawmakers who visited Gasiri often also visited Namdomaru, Ido Sikdang, and Daebanggol. The traffic continues from one restaurant to the next.
So it is more accurate to read this not as ‘several popular restaurants’ but as a single cluster: one spoked mass with Gasiri at the hub and the other ten hung from it by thick lines. Not a list of dots, but one shape with a center.
The picture only this data can draw
Here the methodological distinction becomes clear. A ‘Top 10 Most Popular Restaurants’ needs only payment totals, and anyone can make one. But the relationship between restaurants visited by the same lawmakers can only be drawn when the record shows who visited each point of use. A summary table holding only the visitor count per restaurant can never tell you how many people visited two restaurants. Totals give you dots; relationships give you lines.
That is exactly where political-funds data parts ways with an ordinary food ranking. A standard popularity list takes restaurants as its population; a co-visit network takes lawmakers as its population and links the places they used together. This is not the same restaurants seen through a different lens but a measurement of an entirely different dimension. For that reason, this picture is hard to replicate with a simple restaurant ranking.
What the measurement says — the physical radius of the lawmakers’ world
So saying the radius is narrow is not only a metaphor but a measurement. 182 lawmakers visited the 11 flagship restaurants, and the main links connecting them run unbroken in the range of 17 to 25. That a group of 300 uses the same venues at such a high density shows that the physical radius of the lawmakers’ world is smaller than we assume.
In the earlier ranking analysis we saw that the top restaurants were almost all packed into walkable Yeouido by the Assembly. The co-visit network adds a further layer. The restaurants are not only clustered in one neighborhood; the same peoplefill them in turn. A narrow visitor group overlays a narrow space. Lawmakers’ lunch radius is narrow both by place and by people.
Why does this matter? Political-fund spending is meant to be public, yet disclosed records usually survive only as tables and totals, with the relationship invisible. Totals tell you who spent how much where, but a summary table never tells you which places the same people repeatedly used. That overlap surfaces only when visits by point of use are connected with lines. If the restaurant ranking was a map of movement, the co-visit network is another map: one that shows how much those routes lie on top of one another.
Method & source · The co-visit graph is a separate tally built from the visit records by point of usefor political-funds meals. Nodes (circles) are restaurants, the number on a node is the count of lawmakers who visited it, and edges (lines) are the count of lawmakers who visited both restaurants together, with only 17 or more shown. In keeping with the de-identification principle, lawmakers are tallied only as anonymous ‘counts,’ with no individual identification. Separately, the restaurant data published on this site (food.geojson) holds only the visitor count per restaurant (memberCount) and contains no lawmaker identifiers, so these co-visit edges cannot be recomputed from the public data alone (they rest on a separate tally). It is based on the meal points of use in political-funds expenditure records disclosed by the National Election Commission, and is a snapshot of payment records. Data tally · kookrator.