As some of you may remember, I’m excited about Google Squared, a project I see as a great first step toward exploratory search at a web scale. Yes, I know that Duck Duck Go, Kosmix and others are already taking on this challenge, but it makes a difference to see Google throw its weight behind such an ungoogley initiative. Plus Google Squared is ambitious, to say the least–the input is free-form text and the output is highly structured.
Since I’ve beaten up Wolfram Alpha for is overreliance on NLP, I can’t give Google a free pass. It would be nice to be able to give Google Square more structured guidance (yes, I’m still an HCIR fanatic). But Google Squared seems to achieve far more robust query interpretation than Wolfram Alpha’s–perhaps because supporting exploratory search is less brittle than question answering.
The quality of the tables that Google Square produces as results is still spotty, but it is a major improvement from the initial release. To those who wrote off Google Squared in June, I suggest you take a second look.