So last night was the launch soiree for version 2.0 of Grokker, a search interface tool built right in my own backyard of Marin. It’s backed by Paul Hawken, and sports some impressive advisors, investors, and board members, including Paul Saffo and John Seely Brown. Grokker “gives the big picture” on large collections of data – it’s a visualization tool, and from what I saw last night, the new version is quite elegant. I saw a demo of Version 1.0 about a year ago and they’ve really made a lot of progress. I am a huge fan of the word “grok” – from its original Heinlein origins through Wolfe and Wired (where I helped the author of Wired Style write the definition) and the Standard (where we launched a series of newsletters and a magazine called “Grok“). It’s neat to see it making another round through the vernacular.
In any case the party was quite subdued by usual product hype standards, with an odd assortment of well-to-do investor types wandering shoeless through an extraordinary Mill Valley home, and Bonnie Raitt also in attendance (Bonnie Raitt?!), sporting a very down-to-earth and pleased-to-be-here manner. Groxis CEO RJ Pittman gave me a demo. Grokker takes datasets created by a keyword or phrase queries on sites like Amazon or “the Web” (it hits six search engines) and runs them through a second filter which displays results as clusters of nested orbs, each with tags derived on the fly from the data. It’s quite a seductive interface, and I can see it working for any number of search needs. And as a trend, I applaud this kind of development – building new applications based on search as a platform, rather like an OS. (Tim Bray among others have noted in the past how search today seems stuck at the level of the DOS C: prompt. Groxis might be seen as an attempt to go GUI.)
But the most interesting thing about Groxis is how it is approaching versioning – this release comes after more than a year out in the open, soliciting feedback (some of it quite negative) from the search community. With Version 2.0, Groxis plans to again listen and learn from how the product is used, rather than try to force it into a particular bucket of revenue (though they do have enterprise and licensing deals). This is due to the angel investors behind the company, Pittman said, who are not demanding a rigid, pre-determined approach to how the product will make money. That was exactly how Google did it in the beginning (and, it seems, how Friendster is doing it now). The Mac version will be ready to beta in a few weeks. When I get it, I’ll post more.
3 thoughts on “Grokking Groxis’ Grokker”
This sound interesting, but I’ve always had a difficult time with graphical interfaces to data sets. We researched/played around with a number of these types of programs at Wired (Steve Steinberg was always interested in this topic).
Toward ‘mapping the Web’, Xerox built a wheel-spoke interface several years ago, I think Apple and/or Macromedia did a 3D browser also. Early stuff – very creaky.
As with any map, they all introduced significant overhead for the user, the ability to interpret symbols and spacial relationships through a window on a computer screen is not that easy. And many of them inadvertantly would obscure their own text labels, which in my opinion is an automatic failing grade.
I always thought that what made Google so great is that its filtering algorithms obviated the need for “exploring” the data set. But I can certainly see the need to see a little further over your search’s event horizon.
In general, I’m a massive fan of public API’s — google, amazon, UPS, USPS all have created excellent tools that create huge strides in what web programmers can do (easily).
I’d also like to see more work on building Web apps that improved my ability to customize my results, as well as see statistics about those results. But fancy visual interfaces just get in the way, IMHO.
[Groxis: Why not a Web-based app?]
Its a GREAT product
I want to buy Grokker, robot spider or crawler that will
datamine internet WWW sites for all the trade boards such as
An estimated 20,000 such sites exist on the WWW
The spider should also automatically register/enroll/list
a company at each of the trade boards with, name, address,
tel, fax, email, URL, by product category and its products
together with graphic files at each trade board
if you have such software available I also want to
partner in the sale of an estimated 1000 copies
Dr. Peter Palms PhD
Palms & Company, Inc.
Palms Harbour Lights Building, Suite 203
515lake Street South
Kirkland (seattle), Washington State USA 98033
Tel 1 425 828 6774
Fax 1 425 827 5528