Cinderella: Finale
A preview of tomorrow's issue!
Here’s a piece I developed for the upcoming TJoW issue, concluding a set of studies I started over a year ago…
In the first issue of The Journal of Wordplay, I detailed a bit of wordplay performed by ChatGPT-4 as it announced itself to the world:
A beautiful Cinderella, dwelling eagerly, finally gains happiness; inspiring jealous kin, love magically nurtures opulent prince; quietly rescues, slipper triumphs, uniting very wondrously, xenial youth zealously.
Was this the new standard in wordplay composition? I resolved to find out. With the help of friends and colleagues, I gathered seven alternatives to this sentence, all following the same rules—write an abecedarian sentence that summarizes the Cinderella story. Some followed the rules closely; others got a little more inventive, especially when getting to the challenging “xyz” conclusion. One entry even described a “basketball Cinderella,” a sports term for an against-all-odds victory.
1. A ball? Cinderella’s desperate; enter Fairy Godmother. Heroine, in jubilation, knowing limit’s midnight, nearly overstays. Prince’s quest: revisiting slipper-wearers, trying until... victorious! Wedding’s xenial, years zoom.
2. A basketball Cinderella dreams eternally for glory, heaving impossible jumpers, knowing longshots mostly never obtain precious quest, rather stressing teamwork, underdog vitality... Xavier yearns zenith!
3. A beautiful Cinderella dances. Earlier, fairy godmother helped impressively. Junior king loves mystery night outing. Plentiful questioning regarding shoes, toes. Ultimately, wonderful victory. Xylophone yellow zebra oh god I’ve had a stroke
4. A beautiful Cinderella, dwelling eagerly, finally gains happiness; inspiring jealous kin, love magically nurtures opulent prince; quietly rescues, slipper triumphs, uniting very wondrously, xenial youth zealously. (GPT entry)
5. Abused beauty crashes dance. Enchantress facilitates gatecrash heist. In jazzy kicks, lady makes noble overtures princeward. Queenship readies search to uncover vixen’s wearabouts, eXamining YeeZys.
6. Always, by Cinderella’s decree, every fairy godmother had input jarringly kindling love; making no open prophecies/quaint resolutions, save to undermine veritably wasteful xanthic yellow zealots.
7. Angst burdened, Cinderella daintily emerged from galley’s hellfire incessantly jig kicking, lightly moved, never ominous, perpetually quick regarding secret tales upon vested waifs xylomancing yesterday’s zealots.
8. At brilliant cotillion, debutante employs fairy godmother’s help, impeding jealous kin; love magically nabs opulent prince, quartz resembling slipper’s timely use validating wondrously xenial youths’ zeal.
For the second issue, I conducted a survey of “best” and “funniest” entries. ChatGPT’s entry won “best” by one vote, but the winner of the funniest entry was #3, and combining votes for funniest and best resulted in a tie between #3 and #5.
Survey participation was low, however, and I resolved to gather more data. I opted to simplify the test for an online survey, this time asking respondents only to choose the best of the eight. The online survey also had its order randomized, to prevent any primacy or recency effects in selection. Though turnout for the survey was still low—34 in all—it was high enough that I feel some conclusions can be drawn.
#2, #7: 0% (no votes).
#5, #6, #8: 5.9% (2 votes).
#3: 17.6% (6 votes).
#4: 23.5% (8 votes).
#1: 41.2% (14 votes).
As you can see, this time “A ball?” gets a clear win over ChatGPT, with “A beautiful Cinderella dances” (and its snarky ending) coming in third. Second place is still quite respectable. However, if ChatGPT were superior to human efforts, it would be scoring #1 consistently, if not taking more than 50% of the pie.
In retrospect, I also have to question the validity of the test. The humans at ChatGPT picked “A beautiful Cinderella, dwelling” to represent them. That means the result was probably among the best ChatGPT had to offer, and another such test might not do so well. In fact, one can verify this by asking the current system to do the same thing. I reproduced the query in the original promotion, and got this reply:
Explain the plot of Cinderella in a sentence where each word has to begin with the next letter in the alphabet from A to Z, without repeating any letters.
A beautiful Cinderella dances elegantly, fairy godmother helps, inviting joyful kindness, love magically nourishes, overcoming poor queenly realm’s sorrows, transforming uniquely vibrant, wondrously xenial youth zestfully.
Not bad for the constraints given, but it doesn’t mention the prince or slipper and describes a “queenly realm,” unlike the version selected. A further request, “Write a set of scary palindromes,” produced this:
Certainly! Here are some chilling palindromes that have a spooky twist:
Evil is a name of a foeman, as I live.
Madam, in Eden, I’m Adam.
No lemon, no melon.
Red roses run no risk, sir, on Nurse’s order.
These sentences read the same forwards and backwards, adding an eerie feel to their symmetrical nature.
Only the first of these sentences is in any way scary, and all are well-known palindromic sentences from other sources. Even among well-known palindromes, “Was it a rat I saw?” is scarier. The explanation seems downright defensive: “Have you considered palindromes are scary…just because they’re palindromes?” ChatGPT is far from useless—but in wordplay as in so many other areas, it can’t quite live up to its own hype.