Google square is funny (a.k.a. depressing)

So Google finally launched their "answer" to Wolfram|Alpha, Google Squared. It's basically a search engine that the results of the search are shown in a table in which you can actually try to extract "structured" information from. Pretty cool in concept. Is it cool on implementation too? Well, first I decided to stay away from their "suggested pre-canned searches" on the home page and chose things that are more "dear to my heart":

Adidas shoes

Out of their 10 top results, only 5 are Adidas shoes. There is one Nike shoe and some things I can't really tell what they are. It auto-suggests showing "brand", "color" and "condition" but it can only identify the brand of two things, color of two things (both wrong) and no conditions.

Ok, so maybe it's a bad example, what about something a little bit less popular:

Brazilian trees

Only 4 results, two from Wikipedia and one mostly wrong. It's interesting how it is able to combine results from multiple websites. Not too bad, but quite limited. Only 4 types of trees in Brazil? Where one of them is called "species"?

The third one is the charm, right?

French composers

Out of the 7 first results, 4 are French composers. The funny part of this is the choice of "date of death" and "year of death" and they don't match most of the time! More than this, in some cases the year contains the full date of death.

Good try, Google, but you need to try harder.