Sportswriters love giving the MVP to the best player on the best team, but in hindsight it often looks dubious. From 1988-1990, Michael Jordan probably had the three single greatest seasons in history and won 1 MVP because Magic inexplicably won twice. Derrick Rose won for leading the bulls to 50 wins, Steve Nash won 2 MVPs for leading a high powered suns offense, and Kobe Bryant won in 2008 for leading the Lakers to 57 wins. In hindsight, all of those MVPs are laughable: Jordan got screwed twice, Lebron deserved it over Rose, KG or Shaq deserved Nash's first MVP, Dirk or Lebron deserved Nash's second, and CP3 deserved Kobe's MVP.

That said, this year is obviously very different. By basically every metric they are neck and neck (win shares, value added, PER, OBPM, etc). I'd personally favor Lebron because he played far more minutes, had better counting stats, some of Harden's metrics are inflated (such as RPM and win shares) by playing on the best team with cp3, and I believe the Rockets would have won 50 games if Harden missed the whole year, while the Cavs probably win under 30 and are a lottery team without Lebron.

But it's basically a philosophical debate: do you penalize Harden for playing less minutes when his team blew opponents out and they locked up the 1 seed weeks in advance, or do you penalize Lebron for playing with jabronis?

We can generalize the scenario as follows. Suppose through the first 60 games, player A adds 13 wins worth of value and player B adds 12 wins worth of value over replacement. Player A's team is in first place in the conference and he plays far less minutes over the final 22 games, while player B is in a tight playoff race and plays heavy minutes and misses none of the final 22 games. At the end of the season, player A added 16 wins worth of value over replacement level players while player B added 18 wins worth of value over replacement level players. Who do you choose as MVP? It may end up defining legacies, so saying it doesn't matter isn't fair to the players.

Strong and valid cases can be made for either player, although Harden is probably ~95% to win because of what sportswriters look for. This year it happens to be fine, but in general I think best player on best team is a bad guideline for who should win.