Editorial Methodology

How we decide what to rank, where the images come from, how we keep voting clean, and what we do when something looks wrong. The short version of how Hot Ranking actually works.

1. How we choose poll topics

Every poll on Hot Ranking is curated by an editor. We do not auto-generate ranking topics from trending searches or scrape from third-party "what to write about" feeds. Topic selection follows three rules:

• The question must be debatable. If the answer is universally agreed (e.g. "Is gravity real"), it's not a poll. We want topics where reasonable people genuinely disagree.
• The option set must be bounded. A poll with 200 candidates is not a poll — it's a list. We aim for 8–20 high-quality options per ranking, ideally covering the full canonical set for the topic (every Taylor Swift album, every Mario game, every World Cup winner).
• Every option must have a clean image. If we can't find a real, rights-clear image for a candidate, we don't include it. The visual is part of the ranking.

We update poll coverage continuously as new content drops — when a major album, film, or game releases, we add it to the relevant existing ranking within a week. We also retire topics that have become irrelevant (no more "Best TV Show of 2024" polls in late 2026).

2. Where the images come from

We use a small set of public APIs as image sources, in this order of preference:

• iTunes Search API — for music (artist photos and album art) and movies (theatrical posters). Highest-quality source for these media types.
• Wikipedia REST API — for biographies, TV shows, video games, foods, landmarks, and most other categorical entries. We pull the article's lead thumbnail.
• Wikimedia Commons — for items where Wikipedia's article has no thumbnail but the underlying image file exists.
• PokeAPI — for Pokémon-specific polls (official sprite art).

Once fetched, every image is verified to be at least 1.5KB, no larger than 6MB, and a real image format (JPEG or PNG, not SVG or HTML error pages). We then hash each image with SHA-256 (first 32KB) and compare against every other item in the same poll — if two items hash to the same value, one of them got the wrong picture, and we re-query with a more specific search term. Polls do not ship until every item has a distinct, working image. This check runs continuously: a nightly job re-scans every poll and re-fetches anything that broke.

3. Attribution and takedowns

We use item images under fair use principles for editorial commentary and ranking. Album art, posters, and character portraits are inherently associated with the work being ranked, and we display them in a way that clearly identifies the underlying media (not as standalone art). We do not edit or alter the source images.

If you own an image and want it removed from the site, send a takedown notice to dmca@hotranking.app with the URL where the image appears. We respond within 48 hours and remove infringing content as soon as practicable.

4. Vote integrity

We use three layers of vote-fraud prevention:

• Browser fingerprint: a hash derived from user agent, screen size, timezone, and a random per-browser identifier stored in localStorage. Used to prevent the same browser from voting twice on the same poll.
• Database-level unique constraint: the poll_id + fingerprint pair is unique at the database row level, so duplicate vote attempts fail at the storage layer, not just the application layer.
• Rate limiting: any single fingerprint is limited to 20 vote submissions per minute. Sustained brigading triggers a temporary block.

We do not require accounts. We do not store personal information about voters — the fingerprint is not reversible to your identity. We also reserve the right to discard votes that appear to be the product of coordinated manipulation (e.g. botnets, sudden 10,000-vote spikes from a single ASN), although in practice this is rare.

5. Results and aggregation

For tier-list polls (S/A/B/C/D/F), we compute the percentage of voters placing each item in each tier, and surface the modal (most common) tier as the "consensus" rank. For ranking polls, we average each item's position across all submissions. For this-or-that and hot-take polls, we show the percentage split with a sample-size footnote.

Results update in real time after each vote is submitted, with a 1–2 second cache to keep the database from being hammered on viral polls.

6. Featured polls and the daily rotation

The "Today's Battle" poll on the homepage rotates every 24 hours using the UTC day-of-year as a deterministic seed across our pool of editor-featured polls. This is automatic — no human picks the daily poll, which keeps the rotation fair and predictable.

The "Viral Right Now" section uses a recency-weighted heat score: vote count multiplied by a recency boost (3× for polls under 3 days old, 2× for under 7 days, 1.5× for under 30, 1× for older). This surfaces polls that are gaining votes now, not just polls that have accumulated votes over time.

7. Editorial corrections

We make mistakes. When we ship a poll with a wrong image, a typo in an option name, or a missing canonical entry, we fix it as soon as we hear about it. If a correction would meaningfully change the results, we note it in the poll's description with a date stamp. We don't delete or silently rewrite past polls.

Report errors to corrections@hotranking.app.

8. Funding and editorial independence

Hot Ranking is supported by display advertising via Google AdSense. Ads appear in fixed slots on the homepage, category pages, and poll pages, but never inside the active voting interface. We do not run pop-ups, autoplay video ads, or paywalls.

Advertisers do not see, influence, or have any input into poll selection, item lineups, or rankings. If a brand wants a sponsored ranking (e.g., "Best [brand category] of 2026"), we publish it as a clearly labeled sponsored poll, and the option set is still curated by us, not by the sponsor. We currently run no sponsored polls.

9. Who runs it

Hot Ranking is operated by a small independent team. We don't publish individual editor bios — the site is collaborative — but every editorial decision goes through a human. There is no AI auto-generation of poll topics, item lineups, or descriptions. AI tools are used in two specific places: bulk-fetching candidate images from public APIs (so we can scale the catalog to hundreds of polls without manual download labor) and drafting first passes of editorial descriptions (always reviewed and edited by a human before publication).

10. Changes to this page

We update this page when our methodology changes meaningfully. The current version is dated April 27, 2026. Past versions of the page are available on request.

Last updated: April 27, 2026 · Questions? hello@hotranking.app