Eight queens on a chessboard, 1848
In 1848 a German chess magazine, the Schachzeitung der Berliner Schachgesellschaft, printed a problem from a chess composer named Max Bezzel. He signed it only as “Schachfreund”, a chess friend. The task was to place eight queens on a chessboard so that no queen could capture another. It is now called the eight queens problem.
In 1850 Franz Nauck published ninety-two solutions. That is the full list, although Nauck didn’t prove that nothing was missing. The mathematician Carl Friedrich Gauss was working on the problem at the same time and wrote about it in letters to the astronomer H. C. Schumacher, a friend. Gauss had found seventy-two solutions when he learned of Nauck’s ninety-two.
Later writers often gave Gauss the credit for posing the problem or for solving it. He did neither. A 1977 paper by P. Campbell in the journal Historia Mathematica traces how that mistake was copied from book to book. By 1869 the question had been widened to any number of queens on a board of matching size.
What the old problem shares with the new puzzle
Eight queens that can’t capture one another must stand one to a row and one to a column. Firefly Fields asks for the same thing on the same 8×8 board. That is the family resemblance, and it is why one-star puzzles are so often drawn with queens or crowns.
Two things are different. A chess queen attacks along a whole diagonal, while a firefly only rules out the neighbors it touches. So every answer to the eight queens problem keeps its pieces apart in the Firefly Fields sense, but most Firefly Fields answers would fail as queen placements. The second difference matters more: the colored meadows. The eight queens problem has ninety-two answers and no clues. A region puzzle uses its regions as the clues, and they are meant to leave only one answer standing.
Nothing we found shows that the 2003 puzzle was drawn from the chess problem. The link is one of shared ideas. It is not a documented line of descent.
Hans Eendebak and the 2003 championship
The 12th World Puzzle Championship was held from October 14 to 19, 2003, at Papendal, a sports center near Arnhem in the Netherlands. The Dutch puzzle maker Hans Eendebak wrote a new puzzle for it and called it Star Battle. The World Puzzle Federation’s newsletter from that autumn names him as captain of the Dutch team and as an editor of the Dutch puzzle magazine Breinbrekers.
The rules have hardly changed since. Each row, each column and each outlined region holds a set number of stars, and no two stars touch, even at a corner. A Star Battle was among the puzzles in that year’s finals, which Ulrich Voigt of Germany won.
The puzzle belongs to a family that solvers call object placement. Grandmaster Puzzles, an American puzzle publisher, names a puzzle by Tim Peeters called Cattle as its likely forerunner. Cattle had the regions and the no-touching rule, but gave its counts as numbers outside the grid. We found no account from Eendebak himself, so “likely” is as far as the evidence goes.
A championship regular
The World Puzzle Championship has been held since 1992, when the first one took place in New York, and it runs under the World Puzzle Federation. Puzzles written for it are solved by national teams from many countries, and the good ones travel home with them. Star Battle was one of the good ones.
Twenty years on, the 30th championship in Toronto in 2023 gave Star Battle a round of its own. The instruction booklet listed it among a handful of classic types and described it as a “Dutch invention by Hans Eendebak”. The same round carried variants with missing borders and with three stars to a line.
Hand-made Star Battles also became a staple for puzzle publishers. Grandmaster Puzzles, run by the champion solver Thomas Snyder, has published them for years and credits the type to Eendebak and the 2003 championship on its rules page.
Two Not Touch in the newspaper, 2020
On April 20, 2020, Star Battle reached a much larger audience. The New York Times began printing it in its expanded puzzle section under the name Two Not Touch, with an easy puzzle and a medium one every day from Monday to Saturday.
The puzzles came from Jim Bumgardner, who runs the puzzle site Krazydad and had been publishing the same type there as Star Battle for several years. The newspaper’s version asks for two stars in every row, column and region. The new name states the rule in three words.
Krazydad offers the puzzle in three strengths: one star on an 8×8 grid, two stars on a 10×10 grid and three stars on a 14×14 grid. Bumgardner calls the 8×8 the easiest and the place to start. His tutorials also gave solvers some shared names for techniques, including the four-squares rule that LarkSpark’s tips call the two-by-two block.
One star, colored regions and queens
The one-star puzzle is the gentlest form of Star Battle. With a single piece in each row, column and region there is less to count, and a small grid fits a phone screen. Shading each region in its own color, in place of heavy outlines, makes the regions easy to read at a glance.
At the start of May 2024 the professional network LinkedIn added three daily games to its site. One of them, known as Queens, asks you to place one queen in each row, column and colored region with no two queens touching. Thomas Snyder wrote that he helped provide puzzles for its launch and described it as a very accessible, one-star version of Star Battle. He later joined LinkedIn as its principal puzzlemaster.
The chess piece brought the story back to where it began. Krazydad now lists “Queens” beside Two Not Touch as another name for Star Battle. It is the same puzzle under each name, and the 1848 problem is the reason the queen fits.
Hard questions for computers
Placing eight queens is easy for a computer, and it has long been a standard exercise for people learning to program. A nearby question is much harder. In 2017 Ian Gent, Christopher Jefferson and Peter Nightingale showed in the Journal of Artificial Intelligence Research that completing a board on which some queens are already placed is NP-complete. In plain terms, no known method stays fast as the board grows.
Many pencil puzzles have been shown to be hard in the same sense. We looked for a published proof about Star Battle itself and didn’t find one we could read and cite, so this chapter makes no claim about it. None of this touches a daily 8×8 grid, which is small enough to solve by elimination alone.
How LarkSpark plays it
Firefly Fields is a one-star Star Battle on an 8×8 board, with a new puzzle every day. The stars are fireflies and the regions are eight colored meadows. You place one firefly in every row, every column and every meadow, and no two may touch, even diagonally. The full rules take a minute to read.
A click places a firefly, a second click turns it into a × and a third clears the cell. A right-click marks × directly. The crosses do most of the work, as they do on paper. The puzzle is timed, and it finishes when every row, column and meadow has its firefly and none are touching.
The name and the theme are LarkSpark’s own. The puzzle type belongs to Hans Eendebak and to the solvers and setters who have carried it since 2003. For the methods they worked out, see the solving tips.
Sources
- History of the n-queens problem (Bowtell and Keevash, 2021) · arXiv
- Star Battle rules and info · Grandmaster Puzzles
- Newsletter 8, October 2003 · World Puzzle Federation
- The 2003 championship in Arnhem · US Puzzle Championship
- World Puzzle Championship · World Puzzle Federation
- Championship years and host cities · US Puzzle Championship
- Instruction booklet, 30th World Puzzle Championship · WSPC 2023
- Star Battle by Thomas Snyder · Grandmaster Puzzles
- Two Not Touch puzzles in The New York Times · Krazydad blog
- Star Battle puzzles and sizes · Krazydad
- Advanced strategies for Two Not Touch · Krazydad
- LinkedIn adds games · Fortune
- LinkedIn adds three games · Fox 6 Milwaukee
- Star Battle, also known as Queens, by Thomas Snyder · Grandmaster Puzzles
- LinkedIn’s principal puzzlemaster · Entrepreneur
- Complexity of n-Queens Completion (Gent, Jefferson and Nightingale, 2017) · Journal of Artificial Intelligence Research