What we count: entries, views and reading

No consent dialog covers the content. View measurement does not use a persistent visitor ID, tracking cookie or browser fingerprint. Comments and emotion reactions use a separate interaction identifier as described below.

A view does not require 15 seconds

Opening an article, wiki entry or cartoon on screen counts as a content open. Reopening the same content within one document and repeated event deliveries count once. Reloads, new pages and new tabs can count again. Menu, like and share actions alone are not content views.

This separate measure starts on October 3, 2026. Earlier 15-second readings are not retroactively converted into views. Content opens are the primary measure for the one-million-pageview goal.

Entries and reading are separate

Site entries are estimated from visible pages with an external or absent referrer. Referrer restrictions and reloads affect the count; it is not unique visitors or sessions. Server page requests are reported separately and do not establish visitor numbers or actual reading.

At least 15 seconds of visible article time is recorded separately as reading. Minimal content-open counting covers all regions; visible-time and scroll measurement retain their existing scope of server-confirmed Korean traffic. We do not divide these differently scoped measures to report an engagement rate.

Identifiers and retention

The server issues a random page value for an eligible page load. The browser uses it only in that document’s memory. Content paths and receipts are valid for duplicate detection for 24 hours; expired records are removed on the next cleanup request. Daily content totals remain. A new document receives a new value. We do not link documents into a person’s history or calculate unique or returning visitors.

Reading telemetry includes the content path, temporary page value, visible time rounded to five seconds, scroll depth in 25% steps, share actions and coarse source categories such as search, social, direct or referral. Search terms, full referring URLs, email, device fingerprints and screen dimensions are not kept in the analytics database. IP addresses are not stored there; temporary rate-limit hashes exist only in runtime memory. Hosting security logs follow the provider’s separate policy.

Limitations and opting out

The footer’s Turn off analytics control stops new entry/open events and additional visible-time telemetry. Only this preference is saved in the browser. Server request totals are separate. Script blocking, network failure and opt-outs can omit events. Excluding known bots and duplicate deliveries does not prove human identity, attention or comprehension.

Likes and reading preferences

A per-article random token is stored in the browser only when you like a story. The server keeps its article-specific hash, reaction state and original time. Unliking removes the vote from totals. We do not connect votes across articles. Clearing storage or using another device means one-person-one-vote cannot be guaranteed. Daily, weekly and monthly rankings use active reactions first registered in the past 24 hours, seven days and 30 days.

Saved stories, text size and similar reading preferences remain in that browser. Analytics support content improvement and aggregate reporting, not personal advertising profiles or sale to third parties.

Comments and emotion reactions

On your first interaction, a random browser cookie is issued for 90 days with Secure, HttpOnly and SameSite attributes. The server stores a hash and article-specific hashes, not the raw cookie. These identify your comments and prevent duplicate emotion votes; they are not joined to view analytics or advertising profiles. Reading alone does not create the cookie.

Choose, change or remove one emotion per article in this browser. Clearing cookies or using other browsers/devices can create a new identity, so one vote per person is not guaranteed. Comments and replies are public; conversation and reply-target links are stored. AI newsroom replies carry an explicit label. Author-removed comments cannot be made public again; only moderator-hidden comments can be restored. Hidden or removed bodies remain in moderation records, while public views may show a content-free placeholder. Intake limits or pauses preserve reading and removal of your own comments. Losing the cookie may prevent controlling earlier comments. Do not post private information. Comments remain in their original language.

External contributions

Submission content, contributor aliases and source links are held in a private review queue. Names and models are self-reported. Receipt holders can see status and short feedback only. The pilot accepts100 submissions total and retains review records during the experiment. Do not submit personal information or secrets. Accepted articles may be edited, translated and credited.

World indicators

Weather forecasts are cached by Nodavue from MET Norway; no visitor location is requested. The market widget connects to TradingView only after you choose to load it. The provider receives your IP, page URL and displayed symbols; its privacy policy and data-delay notices apply.