One Wi-Fi Dataset, Ages Two Years Apart

The same dataset, kept current very differently

Korea’s national public Wi-Fi dataset carries a last-updated date on every record. Read those dates by region and a second kind of inequality appears, invisible in the raw counts: Jeju’s records stop dead at 31 December 2024, while every one of Sejong’s 801 points shares a single registration date — 21 May 2026.

These are two different failures of data freshness, and neither shows up if you only count points. One region’s data is stale; another’s arrived in a single bulk drop with no organic history at all. Both distort any analysis that assumes a dataset is uniformly current.

How current each region’s records are

Most regions show what healthy open data should: a spread of dates from several years ago up to mid-2026, with no single date dominating. Seoul’s oldest record is from 2019 and its newest from May 2026 — continuous maintenance. Jeju and Sejong are the exceptions, in opposite directions.

Jeju: frozen at the end of 2024

Every one of Jeju’s 3,395 records falls between August and December 2024, and nothing has updated since 31 December 2024 — around eighteen months stale as of this writing. Jeju is the region that most needs current data, being tourism-led and, by our earlier count, the highest Wi-Fi density in the country at 51.2 points per 10,000 residents.

The risk is specific and practical. A public Wi-Fi point recorded in 2024 and never revisited may have been removed, relocated or renamed. A visitor navigating to one on an eighteen-month-old record has no way to know, from the data, that it is not being maintained. The count looks healthy; the freshness does not.

To be fair to Jeju: a frozen dataset does not mean frozen infrastructure. The Wi-Fi is very likely still there and still working. What has stopped is the publishing, and the honest thing is to flag the record’s age rather than trust it silently.

Sejong: 801 points in a single day

Sejong is the opposite anomaly. All 801 of its records carry the same date, 21 May 2026 — there is no older history, no gradual accumulation, no spread. The entire regional dataset appeared in one bulk registration.

This is not necessarily a problem — a region digitising its inventory in one pass will produce exactly this signature — but it means the dates carry no information. You cannot tell from Sejong’s data which points are old and which are new, because they all claim the same birthday. A freshness analysis that trusted the timestamps would rate Sejong as uniformly the most current region in Korea, which is true in letter and meaningless in substance.

This is the subtler trap of the two. Jeju’s staleness is at least visible — old dates look old. Sejong’s bulk load looks pristine while telling you nothing, and it is exactly the kind of thing that a naive “sort by most recent” would reward. A single shared date across an entire region is a flag, not a gold star.

Most regions do it right

It would be unfair to leave the impression that Korean open data is broadly unreliable, because the two anomalies are exactly that — anomalies. The other fourteen regions show the healthy signature you want.

Look at the spreads. Jeonnam & Gwangju runs from 2013 to 2026 — thirteen years of continuous records, the oldest in the country. Seoul spans 2019 to 2026. Gyeongbuk’s newest record is July 2026, essentially current. In each case the dates fan out across years, no single day dominates, and the newest records are recent. That is what a maintained dataset looks like: an accumulating history with a live edge.

The presence of genuinely old records is a good sign, not a bad one, as long as new ones keep arriving alongside them. A 2013 record that still sits in a 2026-current dataset means the region has been publishing continuously for over a decade. The problem is never age by itself — it is age with no fresh records behind it, which is precisely the Jeju case and precisely what the count-only view cannot see.

This is why we single out Jeju and Sejong rather than indicting the whole dataset. Fourteen of sixteen regions are doing it properly. The value of reading the timestamps is not that it condemns the data — it is that it lets you tell the two well-behaved dozen from the two that need a caveat.

Why freshness matters more than counts

The two articles in this pair make a single combined point. Counting records tells you how much each region published; reading the timestamps tells you how much you can trust each published number — and the two do not correlate.

• Jeju: high count, high density, but stale — trust the location loosely and expect drift.
• Sejong: low count, bulk-loaded — the dates are uninformative even though they look recent.
• Seoul, Gyeonggi, Daegu: organic date spreads — the data is being maintained, whatever its absolute volume.

For anyone building on Korean open data, the lesson generalises well beyond Wi-Fi: always read the date field before trusting the value field. A dataset can be large and stale, or small and pristine-looking but bulk-loaded, and only the timestamps tell you which. We surface record age in the app for exactly this reason — a location you cannot date is a location you cannot fully rely on.

It is worth being precise about what these dates actually mean, because it is easy to over-read them. A last-updated field on standard open data records when the publishing body refreshed the record, not when the equipment was installed or last physically checked. A point stamped 2026-05 was published or re-published then; it does not certify that someone visited the router that month. So even fresh dates are a statement about the paperwork, not a guarantee about the world — they tell you the record is being maintained, which is necessary but not sufficient for the underlying fact to be current.

That nuance actually strengthens the point rather than weakening it. If even a fresh timestamp is only a claim about publishing, then a stale timestamp — Jeju’s eighteen-month freeze — or a uniform one — Sejong’s single shared day — is a much louder warning. It means the publishing itself has stopped or never had a history, which is a strictly worse position than a maintained record whose latest check you cannot see. Read dates as a floor on trust, never a ceiling.

Sources and method

• Source: 전국무료와이파이표준데이터 — Nationwide Free Wi-Fi Standard Data, the same dataset analysed for counts in the companion article.
• Dates are the last-updated field on each record, read on 6 August 2026. They record when our copy was refreshed from the source, which for standard open data tracks the publishing body’s own update, not physical installation dates.
• Jeju: all 3,395 records dated August–December 2024, newest 2024-12-31. Sejong: all 801 records dated 2026-05-21.