Groxis and Clusty Matt 09 May 2005

19 comments Latest by GOESE

Google we’ve got a problem. Hardly anyone ever goes beyond the first page of results, even if there are a ton more relevant links. Now a couple of players are stepping up to take a crack at a solution. displays a Yahoo search as a series of categories set in a circular map (via a Java plug-in). It’s neat but seems like the type of thing you’d only use once in a blue moon (who wants to deal with all that hubbub for every search?). may just be onto something though. The site displays a list of folders with relevant categories to the left of the meta-search results (you can cluster by topics, sources, or URLs). It’s a nice way to provide Google-style results while also offering an alternative to filter results down to relevant categories.

Both sites present interesting new search paradigms but will anyone really care enough for them to make a difference?

19 comments so far (Jump to latest)

Dan Boland 09 May 05

Both present interesting new search paradigms but will anyone really care enough for it to make a difference?


Grokker’s search concept is a perfect example of the notion that sometimes the best ideas simply don’t work. Bottom line is that a search takes too long and is entirely unintuitive.

Clusty’s concept actually reminds me of how eBay breaks down its searches. But the page has to reload just to see a subdirectory in that column? No thanks.

PS - You screwed up your Clusty link.

Dan Boland 09 May 05

It’s a problem when you aren’t sure how to phrase what you’re looking for. I always go into Google’s preferences and change the setting to display 100 results on a page to increase my chances of finding what I’m looking for on the first page. Honestly, I would set it to 500 or 1000 results on a page if they’d let me.

Justin Gardner 09 May 05

Agreed. Grokker is interesting, but it’s not there yet. People will search once and then probably be done with it. If Grokker really wants people to start using it, they might want to approach schools about doing demos for the next generation of searcher.

Clusty seems to be a nice extension for traditional keyword search. Categories are important and I think you’ll be seeing a lot more of this type of thing from the major search players soon. After all, you can only twist so many knobs on keyword search before it collapses under it’s on weight. Clusty offers explicit narrowing of search results, and people need that. There’s too much junk out there right now.


Will 09 May 05

I go beyond the first page of results all the time if I don’t find what I’m looking for on the first page…

That’s what searching is all about, LOOKING for what you want.


beto 09 May 05

Interesting concepts, lousy implementation (Grokker). Looks cute and might strike a chord with taxonomy enthusiasts, but that’s about it. Waiting for a Java plug-in to load (even on a DSL line) isn’t much of a confidence builder.

Clusty´┐Żs concept actually reminds me of how eBay breaks down its searches. But the page has to reload just to see a subdirectory in that column? No thanks.

Yeah I know, it is hard to accept that once you get addicted to Ajax… however I like Clusty’s concept of breaking down results by topic. Sounds like a much-needed filter for Google. A vanity search yielded result topics grouping many of the sites, people and places I’ve been to. Impressive.

Will 09 May 05

Clusty’s categorization is a nice feature though. I’ll be happy when Google adopts that functionality.

Allen Armstrong 09 May 05

Hardly anyone evers go beyond the first page of results,…

thats highly subjective… I think there are two types of searches. A specific search… I want to find out which browser support the text-shadow css tag. There’s a specific search and will be answered in a quite a few sites on the first page.

Then there is the research type of search. I want read reviews on a certian type of camera, or I want some papers on a medical problem for a paper for school… of course I’m going to look at more than one page of results and get as much info.

I think most searches are going to be the quick directed search… Like Steven said… Google is king at getting you the direct answer right away.

Now will Grokker or Clusty be “a problem” for google. Hardly… Grokker is too clunky of a UI to be useful. A little too much overhead and not intuitive enough for the casual user. Would be good for the research type of search.

Clusty would possibly be good for the previously not-mentioned third search… the VagueTM search… where you kinda know what you want to look for but dont have all the info and not exactly sure what to search for.

..ak 09 May 05

Both sites present interesting new search paradigms but will anyone really care enough for them to make a difference?

I think they will but it will take a while to catch on. Using search tools is nothing new and I’ve leard how to “guess” the search terms to get what I want. I tend to do a couple of searches to help me narrow down my terms. The speed of google makes failed searches easy to accept.

There is also the habit of people going to MSN and Yahoo that provide a search within their own properties. Since they own the content, you are bound to get the content you are looking for in the first page.

I would use Clusty for library research, document research, and narrow fields… much like Apple’s Spotlight.

William H. Harle Jr. 09 May 05

I’ve found when searching specific subjects for school, that clusty often comes up with little to no results for some searches in which i get exactly what im looking for in google. I like the idea of clustering my results in clusty, but im not sure its worth it if im not getting the same search results. Bad search results is why i stopped using dogpile.

mh 09 May 05

I don’t know if this is really a problem. What you’re describing (in the first paragraph of the main post) also describes the way most people choose their music (i.e., choose from the 15 songs being rotated on the radio, or from the front page of iTMS). Is it possible to *fool* people into being more discriminating?

Jeff Leombruno 09 May 05

Clusty’s biggest issue, as far as I am concerned, is that they don’t return results from the biggest search engine, Google.

In fact, on some searches, they are returning results from the smaller search engines, that don’t index the sites nearly as often, so the results might not be as correct.

If Clusty returned results from Google, MSN, and Yahoo as the main three SEs, I’d have found a new daily search engine.

It’ll be interesting to see how long it takes for the other SEs to pick up the clustering bit though….

Gabriel White 09 May 05

These paradigms are anything but new. Take a peek at the perenially interesting Atlas of Cyberspace ( to see how people have been doing these things for years (ooh, look, there’s Yahoo in 3D VRML; they were the days).

Tools like Clusty and Grokker (less so) are great for situations when you don’t really know what it is you’re looking for. I’m looking for stuff about Britney Spears (heavens knows why), and the search UI helps clarify my goal by suggesting different kinds of topic areas.

These kinds of search UIs are less useful for highly directed specific searching (i.e. when people already have a pretty good formulation of what they want to find).

jose 09 May 05

at a presentation at the university of michigan (us) last fall, a google researcher demoed a very early prototype of google’s clustered output. it’s the result of many research projects including infering the meaning of terms from known terms and their co-location within a result. why they haven’t just finished it off and integrated it i do not know, but it appears they’re working on it.