Or they decided it was an opportunity to test potential improvements. Testing them on a new line where fixing any "improvements" that don't pan out will be perceived as an improvement might well be better than testing them on an old line where any regressions will be immediately perceived as such...