Findings Concerning Autocorrect Betrayal
The message was reread before sending. The message was sent. The message contains a word that was not intended and was introduced by autocorrect.
FORENSIC FINDINGS — Autocorrect Betrayal
SUBJECT: Undetected autocorrect substitution in a sent message. TIMELINE: Occurred within a twelve-second window.
The Message
The message was composed as a casual good luck message. The intended text included the word "luck." The word "luck" shares a starting letter with a different word that autocorrect considers a higher-probability completion in some contexts.
The substitution occurred during composition, at the moment of typing. The autocorrect confidence was high enough that the correction was applied immediately without a correction indicator.
The Review
The message was reviewed before sending. The review took approximately three to four seconds. The review covered: the overall message, the tone, the greeting, and the sign-off.
The autocorrect substitution was present during the review and was not detected. This is consistent with the known limitation of reviewing one's own text: the reader often reads what was intended rather than what is present.
The Detection
The substitution was detected twelve seconds after sending, during a casual second read. The second read produced the error that the first read missed.
At 3:16:12pm, the substitution was visible. The receiver had already received the message.
The Response
A follow-up message was sent: "meant luck not duck, autocorrect." The receiver replied with a laughing reaction.
The clarification was received well. The substitution became a small moment rather than a problem. The message's intent was understood.
The Finding
The autocorrect operates during the composition window. The review window follows. Both windows are operated by the same person, who has the same limitations. The review is not fully reliable for autocorrect detection.
This is known. It will happen again.