Some initial remarks after playing around with this for a bit: Currently, this seems to only include the etymology for English words. Proper nouns are not included, except those also used as common nouns, like Newton. Nor are most compound words made up of simpler words or affixes such as mainland, snowman, simply, or even disuse. This is understandable because covering all these cases will dramatically expand the number of words that have to be dealt with, and this is probably the limitation of the original source used by Google for the etymology information, which was probably an etymological dictionary with the usual space constraints. But freed from such constraints, it should be feasible in the future to add in the etymological information for such compounds.

What I like is the tree structure of the presentation. This covers not only compound words but weirder cases of combination of disparate etymological sources—check out discombobulate or typhoon. But the tree structure is misleading in cases like apple, where Dutch appel and German Apfel are indicated as if they were parent forms instead of sister forms to the English apple. The branching should occur between Germanic and the daughter languages to make the relationship clearer. Also, note that this doesn't show the Germanic form for apple—it looks like it doesn't do hypothetical forms, restricting itself to attested forms.

A welcome feature would be being able to click through to different words at each stage of the etymology, to be able to trace through to even earlier stages or to see what other words are descended from the same roots.