I just spent some time with GoogleBase and was amazed at just how deeply Google has embraced standard information architecture concepts and trends. We have categories, facets, tags. I kid you not. Google of the simple search box with a go button has come a long way indeed.
A few top level categories
12 top-level categories (e.g., course schedules, Events & Activities) show up when you post a piece of content. Interestingly they do not get exposed directly in the search interface, but get used indirectly. Google uses your initial query to place you in a particular category, and show the relevant facets. For example, the facets presented when you search for a vehicle name are different than when you search for a recipe. This is a good approach, but it does mean that Google needs to guess the right category to present the appropriate set of facets. Google Product Manager explains that “Google Base suggests attributes and item types based on popularity, which you can use to define and attach your own labels and attributes to each data item.” So there is some magic going on behind the scene to decide what sets of facets to show.
Apart from the 12 categories, Google also allows you to name your own category. This is smart – its like a constant closed cardsort study with a “Did not fit it” category that allows you to find out the weaknesses of your category scheme.
A rich browse experience: Search and browse is integrated
I like this. I searched for “prius”. It gave me a flat list of results, and options to refine by several facets (or attributes as it refers to them) “price” “condition”, “make”, “vehicles”, “color” etc. You can go back and forth between search and browsing. It also shows query previews (how many results are available for each attribute). In this case, these are not very useful. Since its different views of the same set of results, it shows “114” for all the facets. This defeats the purpose of query previews – whose goal is to help you decide what path to take depending on the number of results. (A better use would be to show previews for choosing paths within the facet).
When I choose one of the facets (I chose “price”), then it gave me some price ranges to refine by. (Note, the interaction for choosing the specific year is not very elegant. I don’t know how to make it better though.)
Like with the categories, Google allows you to name your own facets. I expect that they will use information about facets that are used, and new ones created to constantly refine their facets scheme. Smart indeed.
Refining and Broadening
It gets better (at least more complex). Just for fun, I go back and choose “vehicles”. What happens next I don’t understand at all. Now the interface shows that the facets in use are “vehicle”, “make” and “location”. I only wanted to filter according to vehicles – why did “make” and “location” come into play?
I also notice that it offers me the option to broaden, by clearing one of the choices to broaden the query. That’s a good way to use facets – allow you to filter, sort, refine and broaden in a continuous fashion without having to go to the search box. A well done faceted interface can anticipate what you want and offer you choices in a natural manner. Note how different this is than a paradigm mostly relying on a search box. I for one, had not expected that Google would enable such a “browse” experience.
Hide all your tags, Google is coming
As if facets are not enough, GoogleBase also has tags! And tags are used in a very different way than they are generally used (apart from being called labels as with Gmail). The only reason I realized these are tags is because of the tag like descriptions on the front page, e.g., and in the post interface.
Trust Google to use tags in a completely different manner. When you click on what seems like a tag, then you are essentially doing a search on that keyword. I hated this. Its completely not what I expected, and it left a very bad taste in my mouth. Google does not seem to get that tags are for discovery.
Google ends up defining a lot of standards. And if they use tags for keywords searches, then its likely that a lot of people will come to regard them as such.
I was confused by when tags are clickable or not. Sometimes in item descriptions, I see tag like terms but they are not clickable. Sometimes they are clickable.
I am appalled by this way of using tags. I hope this does not become a trend.
Its interesting how the different companies have embraced tags. The contrast between Google, Yahoo and Amazon’s use of tags goes to the heart of the difference between the companies. Amazon sees it as yet another tool in their arsenal of getting people to say interesting things about their catalog, Yahoo (MyWeb2.0) sees it is a way to connect people, to let people to rely on their friends for searches, and Google, well Google just sees it as more fodder for its search algorithm.
Craigslist & Google
It is also interesting to compare with Craigslist (specifically the Bay Area Craigslist – thats what I use and find compelling). In some ways, the GoogleBase experience is a lot more complex that Craigslist, which has geographical & item categories, and some ways to sort/search. Beyond that you navigate the flat list organized by time of posting. Google could have gone the Craigslist route. Instead it chose to get a lot more structured and fancy with facets and all. (Admittedly, Google has a more challenging design problem – their service is not local, and some items are time bound, but others are not).
Another contrast with Craigslist . Craigslist is a community. GoogleBase comes across like a database. It does not seem to have the soul, the personality that Craigslist SF has. Partly, that’s because Craigslist is local. If I post an ad in GoogleBase, it would feel like it went somewhere into the Google netherworld, findable by Google only. In Craigslist, I feel as if I know where it went and who would be looking at it.
Its in keeping with the Google way to dream up a generic global classified (& other content) service. And also to think that it does not need a local / social component.
I am finding it difficult to sum up my thoughts about GoogleBase. Its very ambitious, and in spite of my criticisms above, overall I find the GoogleBase vision compelling. People want to publish content without having a full fledged site. Some of the content is short-term, some is more long-term. Its a generic need. GoogleBase caters to it. We in Bay Area are spoilt by CraigsList. What about places Craigslist does not serve?
I am fascinated by Google’s embrace of this much complexity in the interface. Clearly, they realize the search, and go paradigm is not always enough.
I like that they are using facets – it seems very appropriate for the type of data they are supporting. Having worked with facets, I know how complex faceted interfaces can be, and I quite like their solutions. This being Google, I suspect they usability tested it to death, and we can assume that users do get it.
I *hate* their use of tags. I hope their interpretation does not catch on. I think it does a disservice to users, in asking people to input tags, but not giving them the most compelling usage of tags – links to other individuals who thought similarly.
Finally, what’s most interesting about GoogleBase is that it corroborated what I have believed all along – tags and search don’t replace facets, categories and other information architecture concepts. I can imagine a tag-based, del.icio.us like GoogleBase – input your information and find others’ information serendipitously. But Google wants people be able to find informaiton in a predictable, intuitive manner and for that it has to rely on a combination of categories, facets and tags. It did not reinvent Yahoo’s directoy – instead its an innovative way of using all three. Tags are a brilliant innovation and can do many better things better than other methods. But the rumours of the death of classification, facets etc., have been greatly exaggerated by experts such as Clay Shirky.