John D. Cook • 9/7/2026

Ngram error rate

This article examines the etymology and usage of the word 'grok', coined by Robert Heinlein in 1961, and contrasts it with Google Ngram Viewer's data showing usage before that date and a recent spike. The author attributes these inaccuracies to OCR errors and incorrect date attribution, noting that the Ngram data is unreliable for tracking word popularity. The post also touches on the word's relevance in tech circles and its resurgence with the Grok AI, concluding that while the effects are large, they are directionally correct after adjusting for error rates.

0 comments

#Data Quality #Text Analysis #Ngram