From my experience (and I teach both), Tableau's graphics look more professional (for reasons I have not been able to figure out), but Tableau is vastly less effective for manipulating and analyzing data. Tableau has few modeling capabilities, so for things like predictive modeling there really is no comparison. But even routine data cleaning is much clunkier in Tableau than in JMP. Tableau is not nearly as intuitive to use. It's only advantage is the quality of the graphical displays. Graph Builder is vastly superior for exploring data. Its functionality surpasses Tableau in terms of how easily the data can be modified and the views be changed. Most of those manipulations could be accomplished in Tableau, but with much more work - you need to know what you want to display in order to get Tableau to dispaly it, whereas JMP invites you to explore the data without knowing in advance what displays you want.
Another difference appears to be the philosophy behind the products. Tableau wants to present itself as a product that does much of the work for you - hence, there are myriad things like suggested graphs for different types of data (which are disturbingly hard to overcome). JMP expects the user to be a capable analyst - and makes things easier to do without telling you what you should be doing. I would frame this as JMP expects the user to be smart, while Tableau appears to cater to users who don't really know what they are doing.
Those are my views - others may feel differently.
