The paper in question Better article The whole thing is bullshit anyway. Stinks to high heaven of marketing. - Bell tested 16 arrangements but published results for 5 - Bell never tested their keypad against the dominant keypad (standard on adding machines for 70 years at that point) - Bell tested 15 employees at a time, not blindly, and the study was never arranged dispassionately - Group 1 is "adding machine with 70 years of experience" against "random bullshit marketing pulled out of their ass" - Group 2 is "adding machine the Brits use to get around patents" against "random bullshit marketing pulled out of their ass" - Group 6 is "we won't get sued by NCR" against "the Brits will sue us" against "our design isn't going to be substantially worse than arranging the numbers the way our customers actually expect them based on 70 years of training" - If you look at the actual data, the design they went with was actually beat out by "our design isn't substantially worse than customer expectation" and "we'll be sued by the Brits." They don't PUBLISH what came in first. - GROUP 1A is the ten-key arrangement used by adding machines for 70 years which never appears again.