Feedback is our weekly column of bizarre stories, implausible advertising claims, confusing instructions and more
Letters of note
Pleased by the, errrm, feedback about our coverage of research on the comparative mortality rates of chess pieces (17 August), we turn to what over-engineered computer modelling can tell us about Scrabble, the game that has been both delighting players and igniting lexicographical arguments since 1938.
Feedback can鈥檛 be alone in finding that one of the joys of the game lies in making up words and swearing blind they are in the dictionary, just not that dictionary. Also contentious, however, is whether the values given to individual Scrabble letters are fair.
Advertisement
In the English version of the game, these values were set by Scrabble鈥檚 creator, Alfred Butts, based on the average frequency with which letters appeared on the front page of The New York Times. But a found that certain tiles were overpowered, unfairly boosting the chances of players who have them in their selection. Z is easier to play than Q, for example, but both score 10 points. Lewis鈥檚 calculations showed a Z should only be worth 6 points.
However, software engineer Kevin McElwee that Lewis鈥檚 proposed values actually make the game slightly less fair. In his AI program, which oversaw hundreds of matches between two equally skilled computer players, the new values increased the overall spread in final scores, suggesting a greater element of luck. Bizarrely, though, McElwee found that if Scrabble鈥檚 tile values were assigned to letters randomly, the spread of final scores in the game remained unchanged 鈥� possibly because people strive to play high-value letters, counteracting the 鈥渄ifficulty鈥� of playing them.
Where the luck of the draw really does seem to count is in whether someone manages to play all seven of their tiles at once, earning an extra 50 points. Remove the ability to score these 鈥渂ingos鈥� and the average difference in final score between two equally skilled players was significantly reduced. So there you have it: stop debating whether Hoover is a real word and hold out for the tile you need to score big with overhot. It鈥檚 in the dictionary.
Rook-y error (again)
Speaking of chess, Frank Warnock wades into the almost metaphysical debate raised by our throwaway assertion thatnon-drawn chess games end with a king鈥檚 demise. 鈥淭he word 鈥榗heckmate鈥� came to English through Old French and Arabic, from the original Persian phrase 拧膩h m膩t, 鈥榯he king is dead鈥�,鈥� he writes. To which we may only add, long live the king.
Rave to the grave
Rest in peace? Think again. Forensic scientists at the Australian Facility for Taphonomic Experimental Research 鈥� now there鈥檚 a job to die for 鈥� have documented the surprisingly active afterlives of human corpses. by Alyson Wilson and her colleagues, who used time-lapse cameras to film the slow dance of the dead over the course of 17 months. 鈥淲hat we found was that the arms were significantly moving, so that arms that started off down beside the body ended up out to the side of the body,鈥� Wilson told reporters.
The findings will be used to inform police investigations, which had previously assumed that a body lay in the same position once a person died. The rest of us can live happily in the knowledge that once we die, we might finally catch up on our exercise.
Drafting error
Adding to our list of extraordinarily expensive mistakes, we note with our all-seeing eye the archived example that appears on page 27 of this issue. Meanwhile, Peter Jacobsen draws our attention to the Diablo Canyon nuclear power plant in California. Shortly after construction began in 1968, geologists discovered a fault line running under the sea nearby.
To insure against the increased earthquake risk, a design for additional supports to the cooling systems was drawn up and built accordingly. Unfortunately, somebody misplaced the note instructing engineers to flip the transparent blueprint over when using it to build supports for the second reactor, a mirror image of the first. Consequently, the earthquake proofing was built in the wrong place. A well-known online reference tool notes that despite the design errors, the plant was approved to open.
鈥淚 suspect this one typo won鈥檛 be the worst,鈥� says Peter, 鈥渂ut it鈥檚 legendary among engineers.鈥� We鈥檙e sure everything is going to be just fine.
A case of booze
Feedback is pleased to note that hangovers are a medical illness. That, at least, is the ruling of a court in Frankfurt, and we are counting on Teutonic exactitude in reaching it. The decision came after a company was accused of making unverified health claims about its powdered supplements and liquid shots, which were sold as hangover cures.
Under German law, food and drinks can鈥檛 be marketed as preventing or treating illnesses. This includes the after-effects of alcohol, the court declared, noting that the condition even has a medical term: veisalgia.
Good news for the thousands of tender heads currently attending Munich鈥檚 beer-fuelled Oktoberfest. File under useful phrases: Herr Doktor, ich brauche einen Krankenschein.
Terms and conditions at
Got a story for Feedback?
Send it to New 杏吧原创, 25 Bedford Street, London WC2E 9ES or you can email us at feedback@newscientist.com
You can send stories to Feedback by email at feedback@newscientist.com. Please include your home address. This week鈥檚 and past Feedbacks can be seen on our website.