We went through every question in the game
Blip Party has sixty games, and most of them run on a list of something: quiz questions, prompts, words to draw, pictures to name. Those lists grew fast, in lots of small batches, and the batches never checked each other. So we sat down and read all of them. This post is what we found and what we changed.
The same prompt, six ways
The biggest problem was repeats. Most Likely To had six versions of "hit snooze". Poll Star asked you to name a musical instrument nine times. Dictate had carrot six times, so if you played a few rounds you drew carrots. And 95 of 238 Punchline prompts started with "The worst…", which is a lot of worst.
We kept the best version of each idea, rewrote the ones with a good idea and a flat line, and wrote new ones where a game was running thin. Then we checked every list against itself as a whole, looking for the same idea in different words, not only the same sentence twice. "Tea or coffee" and "coffee or tea" count as the same question now.
A smaller, better quiz
Quiz Hero went from about 4,600 questions to about 3,600, and about 1,370 of those are new. It got smaller on purpose. Geography had a flood of "what's the capital of…" questions and Cinema kept asking who won Best Picture. Those are cut back so a game has some range. Records that go out of date now say when they were true.
We also fixed questions that were simply wrong. A couple of examples: KSI has left the Sidemen, and Johnny Logan is no longer the only singer to win Eurovision twice. Loreen did it in 2023. If you lost a point to one of those, sorry. You were right.
One more quiz fix, in the daily Quiz Five: the right answer was the first option about 84% of the time. If you'd noticed, well spotted. The options are shuffled now.
Answers that scored wrong
Some fixes change who wins, so they're worth listing on their own.
- Fibber: six of the fake answers turned out to be true, so picking them marked a correct player wrong. Two of the "true" answers were false. Every Fibber fact was checked against a source.
- Rattle: 77 prompts had no answer list, so any real word scored. Every prompt has one now.
- Spectrum: 98 of 133 targets sat at the top of the dial. They're spread across the whole range now.
- Hedge: the right answer sat on the same card too often. It moves around now.
Smarter Blips in the clue games
Blips, our bot players, give clues from a list we write for each word. In Namedrop, 88 words had no clue list, so a Blip clued them with a random word, or in emoji rounds a random emoji. That made them feel broken rather than just a bit dim. Every word has clues now, and we also fixed a few Cabal clues that gave away their own word.
New pictures
The pixel pictures in Rebus, Render and Punchline's photo mode were all redrawn in one style, with proper shading. There are more of them too:
- Rebus: 24 new picture puzzles, like horse + shoe and key + board.
- Render: 26 new things to name before the picture finishes loading.
- Punchline photo mode: 12 new odd photos to caption.
Daily players get a small one as well. We added 76 Ballpark questions, so the daily one now takes about 66 days to come around again instead of about 40.
One spelling
The lists were a mix of US and UK English, sometimes in the same game. Most of our players are in the US, so everything is in US English now. Games where you type an answer still accept the UK spelling, so "colour" and "crisps" still count.
If you find a question that's wrong, a prompt that falls flat or a word that's too hard to draw, tell us with the Help button in any room.
Free, no app and no sign-up to join — players hop in with a 4-letter code on their own phone.