Methodology

No ranking has been published yet. The index opens once 200 visitors have rated 6 items or more. Everything below describes how it will be computed when that happens — there are no sample figures to report yet, and inventing some would defeat the point of this page.

Who answers: Visitors to this site, unpaid and self-selected. This is a convenience sample, not a probability sample: people who look up Korean culture online are not a random draw of everyone who has been to Korea, and the results carry that skew.
Measurement axes: Each item is rated on two independent axes, both on a 1–5 scale: how surprising it was, and how good it felt. Something can be startling without being pleasant, so the two are kept apart and only combined at the end.
Which answers build the ranking: Only sessions that rated 6 items or more (the “core sample”). A ranking only means something if the items were measured with the same ruler — that requires the same person rating several of them. Ratings from visitors who rated fewer are shown on the individual item pages and are labelled there as not counting toward the ranking.
Formula: index = round((surprise_avg / 5) × (goodness_avg / 5) × 100, 1)
Responses we discard: From the core sample we drop sessions that gave the identical score on both axes across every item, and sessions where most submissions arrived in under 3 seconds. Both rules were fixed before any data was collected, and neither has a threshold that could be quietly tuned afterwards to get a nicer result.
Tie handling: Items are ranked strictly (no shared ranks). Items whose index scores are close may be statistically tied within sampling error — check each item's valid_n before treating a small gap in rank as meaningful.

Limits worth knowing before you cite this

The 6-item threshold is a judgement call, not a statistical derivation. Comparing items requires the same person rating several of them, but there is no settled answer for how many are needed to establish a personal baseline. We picked 6 because it is half of twelve, and we fixed the number before collecting anything.
“I have been to Korea” is self-reported and we cannot verify it. Someone who has never been can tick the box. We report the answer rather than pretend to check it.
Duplicate protection is basic on purpose. One rating per item per browser session. A private window, another device, or clearing cookies gets around it. That is a deliberate limit, not an oversight — the alternatives (accounts, CAPTCHAs, device fingerprinting) cost visitors more than the abuse costs us. The real deterrent is that skewing the ranking means rating 6+ items every single time.