You probably can use machine metrics to have a good overview of some properties of the text; one of the more useful machine metrics that comes to mind is vocabulary estimation.
I guess personally I'm a bit put off by the UI. I'm reminded of a scene in the recent movie Her, where the AI offered some suggestions but only after being explicitly asked to do so. In HemingwayApp, the immediate red/yellow highlight will nudge the user into fixing the paragraph, whereas a more subtle colour and/or style might allow him more space to think "the computer is wrong this time, I like it more the way it is".
I guess personally I'm a bit put off by the UI. I'm reminded of a scene in the recent movie Her, where the AI offered some suggestions but only after being explicitly asked to do so. In HemingwayApp, the immediate red/yellow highlight will nudge the user into fixing the paragraph, whereas a more subtle colour and/or style might allow him more space to think "the computer is wrong this time, I like it more the way it is".