My first boss Larry Ragan often dismissed “new” research that wasted money and time “telling us what we already know.” Ole Larry must have been rolling over (laughing) in his grave last week, reading this:
A New York Times analysis of a month’s worth of rally speeches from each candidate showcases Mr. Trump’s freewheeling oratory and Ms. Harris’s disciplined messaging. Using an artificial intelligence model to assess semantic similarity, The Times found that Ms. Harris’s speeches on average contained remarks that were 84 percent similar, compared with 75 percent of Mr. Trump’s.
Methodology
Data was collected for transcripts from 11 rally speeches from Ms. Harris and seven speeches from Mr. Trump between Aug. 3 and Sept. 6. The Times analyzed remarks from these transcripts using an open source text similarity model that can calculate semantic distance between pairs of text. The overall similarity percentage for each candidate was derived by examining each remark in a candidate’s speech, identifying the highest similarity score between that remark and remarks in the candidate’s other speeches, and then averaging those maximum similarity scores.
Leave a Reply