I hope we're not going to argue about whether Kenpom is right or wrong on this site. This is how arguing about whether climate change exists came about. These are numbers from a computer that uses a formula that is adjusted every year based on 16 years of data from all D-I teams during that time. There is no subjective opinion involved. It's just the baseline result derived from every game played in D-I college basketball since 2001-2002. We use Kenpom to inform our opinions, not to set them exactly for us.
Whatever opinions someone has about team A or team B may end up being more accurate than the Kenpom numbers over the entire season. The good thing about sports is that players often overperform or underperform when making a single play, or in a single game, or even over an entire season, because they are not computers. This can move everything about that player and his team from the baseline to something much better or much worse. Duke fans are hoping for a lot of overperformance (e.g., the entire Duke team vs 91 UNLV, Laettner vs 92 Kentucky, Battier/JWill in 01, Zoubek in 10, Allen vs 15 Wisconsin II), because the objective expectation for 2018 is to be a top 10 team that isn't a dominant #1.
My personal opinion is that Kentucky should be higher than Duke because, based on the past few seasons, Kentucky should be expected to play defense at a significantly higher level than Duke, while Kentucky's offense should also be expected to be excellent. The mainstream college basketball media are people who aren't good enough to cover the NBA, so it's not surprising that they would ignore defense and rank Duke #1.