Reversed Text & Word Flip

Reversing by code unit breaks anything outside the Basic Multilingual Plane. Emoji, mathematical alphanumerics and many scripts are stored as surrogate pairs - two code units representing one character - and flipping those pairs produces invalid sequences that render as replacement characters.

Iterating by code point fixes surrogate pairs but not everything. Combining marks attach to the character before them, so reversing moves an accent onto the wrong letter. Emoji built from zero-width-joiner sequences come apart entirely into their component parts.

Fully correct reversal requires splitting on grapheme clusters - what a reader would call a character - which needs Unicode segmentation data rather than a language primitive. For plain Latin text the simpler approach is fine, and for emoji-heavy text it is visibly not.

Reversing characters produces text nobody can read without a tool, which makes it a mild obfuscation - fine for hiding a punchline, useless as security since anyone can reverse it back.

Reversing word order keeps every word readable while scrambling the sentence, which is a different effect entirely. It is the more useful of the two for puzzles, because a reader can make progress on it.

Puzzle servers often use reversed text as a simple cipher layer. Game masters create riddles where reversing a phrase reveals the next clue. In casual chat it is just a playful way to reply without being too predictable.

Character reversal flips every single glyph, so hello becomes olleh. Word reversal keeps each word intact but swaps their order, so hello world becomes world hello. Pick the one that matches your puzzle or joke.