Why this game exists
Modern online news has a serious vagueness problem. Headlines are engineered to extract a click without giving anything away — “You won’t believe what happened next,” “The internet is freaking out about this,” “Doctors hate this one trick.” They’re effective because they work on a very specific quirk of human attention: we’ll click almost anything to resolve an open loop in our heads.
We thought it would be more interesting (and a little vindicating) to invert that loop. Instead of the headline winning the second you click, you’re shown the bait and asked to guess what the mundane truth is before you give up any of your points. If you can read a clickbait headline and reliably figure out what the article actually says, you’re already harder to manipulate online.
What “winning” looks like
We don’t score you on speed alone. We score you on two separate skills:
- Pattern recognition — can you guess what the article is really about from a deliberately vague title?
- Calibration — can you tell when a headline is mildly misleading vs. egregiously misleading? That’s the spectrum bonus round at the end of every game.
Players who score consistently high tend to share the same habit: they don’t take the headline at face value, they make a quick mental list of what the article could plausibly be about, and they look for the boring option. Clickbait answers are almost always boring.
Where the headlines come from
The headlines you play with are real. We pull them from the RSS feeds of clickbait-prone publications — ScreenRant, BuzzFeed, The Verge, IFL Science, BBC News, Bored Panda, and a few dozen others across eight categories: TV & Film, Celebrity, Tech, Sport, Science & Health, Business, World News, and Food & Lifestyle.
Each article is processed by an LLM (Anthropic Claude) which reads the full article body and produces a one-sentence summary of what the piece actually says, plus three plausible-but-wrong distractor answers. A human reviews every headline before it goes live. We reject roughly half of what comes through, usually because the headline is too obvious, the article is too thin, or the distractors aren’t convincing enough.
A small fraction of headlines are AI-generated when a category is running low. These are tagged internally so we can keep the live pool weighted heavily toward real publications. The goal is for the game to remain a tour of the real internet, not a generated one.
We send traffic back to the original article
After every round, you’ll see a link to the original article. This serves three purposes:
- If a headline genuinely catches your interest, you should be able to go read the full piece.
- We’re using publishers’ headlines in a game. The least we can do is drive readers back to them, which is the opposite of how most aggregators behave.
- Over time, the click-through data per publication gives us a credible pitch for branded headline packs — if the game sends a publisher meaningful traffic, that’s a real partnership conversation.
Who we are
For Clicks Sake is built by a tiny independent team in the UK. It runs on Next.js on Vercel, with a Supabase backend and a separate Python pipeline that handles scraping, AI processing and editorial curation. We don’t take VC money. We don’t sell your data. The plan to keep the lights on is a combination of unobtrusive ads, an optional ad-free Pro tier later on, and eventually a handful of sponsored headline packs from publishers who like the idea.
If you want to get in touch about a bug, a partnership, or just to tell us a headline drove you to genuine fury, the address is hello@forclickssake.app.
What’s next
We’re still adding to the game in public. Near-term: a badges system for spectrum accuracy and category mastery, category-filtered leaderboards (“Best Tech Clickbait Reader”), and a daily-challenge archive so you can replay missed days. A bit further out: a “Bury the Lede” mode where you guess how deep into the article the actual point is buried — that data is being collected from day one, so when the mode ships there will already be a real corpus behind it.
The current build is best understood as the foundation. The fun part — turning every clickbait pattern into its own mini-game — is just getting started.
Ready to play? Try a Quick Play game, read the full rules, or check the FAQ.