Thursday, February 05, 2004

Searches vs Time
Eurekster should create graphs like this on the fly... (from google zeitgeist)

The most popular search terms and sites on Eurekster vs lycos top 50

Popular Searches
NASA Catherine Bosley Jennifer Lopez Britney Spears Valentine's Day Blackjack IRS Hilary Duff Mars Rover KaZaA eurekster Atkins Diet feedster paris hilton string cheese incident friendster orkut martini recipes extreme segway kazaalite Popular Sites

Popular Sites
NASA - Mars Full Bloom - The Ultimate... Kazaa Media Desktop People Search: by SSN, la... Britney Spears :: The Off... Feedster :: RSS Search Engine The String Cheese Incident My quest for Friendship, ... The Social Software Weblo... NASA - Mars Origami Boulder Company -... Welkom bij Rijschool Brav... SPACE.com -- Mars Rover S... KaZaA Lite K++ Download Page Martini Recipes - World F... Clay Aiken - Official Web... Atkins Diet - lose weight... Catherine Bosley - Randi ... Orkut Buyukkokten's Home Page Official Lord of the Ring...

Its quite a bit different from the lycos top 50 from the week ending Jan 31 04. Eurekster shows what the "innovators" and early adopters are checking out. And its real time. Look out lycos top 50. You've just been out-predicted. Maybe?
my eurekster results...(again)

your friends top searches
eurekster extreme segway waypath social networking research HIGHLAND CAPITAL social software whizspark blog search engine "mark simpkins" Simon Bisson whizspark social network visualization search results clustered search engine watch bookmarklets eurekster blog search visibility report corante many whizspark blog seth godin snowden technologies your friends top Sites

your friends top Sites
Marc's Voice WhizSpark - Blog Search Engine Blog: Searc... matt jones | work & thoug... Simon Bisson Simon Bisson's Blog Susan Mernit's Blog: Navi... New Social-Network Mappin... Search Engine Watch: Tips... Tong Family Blog: Google ... Anders Jacobsen's blog Waypath Search Visibility Report ... rheingold's brainstorms The Social Software Weblo... WhizSpark - Blog Snowden Group - TECHNOLOG... NetIQ: Security Managemen... UK CAR REVIEWS :- AUDI A6... Welcome to the National O...

The terms are definitely appropriate based on my stated interests for building my network. However, I am probably the most active searcher in my network. Many of these terms were originated by me and then re-searched and revisited by my network. The majority of the ones that weren't originated by me aren't as relevant to my stated interests.

I'd love to publish these via an rss feed!
The Viral Marketing Blog Rates the Viralness of Eurekster

Here's the rating.

Coolness factor: 4/5 (At this point, any search engine that has the potential to be more relevant than Google is cool)
Pass it on ease: 5/5 (The invite incentive is huge, plus Eurekster has jumped on the toolbar bandwagon)
Marketing tie in: 5/5 (It’s already got paid results - plus tons of valueable information about searches / searchers)

Here's the post. He also mentions yours truly.

New tools for sharing information on Eurekster?

In the comments at Ben Hammersly's blog post about Eurekster, Mark Simpkins mentioned they'd like a workspace where people could share their

... this might be interesting but i want is some way to set up an area, say my research topic at the moment and share the notes there, and get input from other peoples thoughts, searches etc etc etc…and no more *ster’s …..mark.
Posted by: mark simpkins

A response from someone who seems official called Bill:

...In regard to your comments about sharing information related to your research topic, we are currently developing solutions that might meet your requirements. We would be happy to provide some of these features to you in beta form - email me for access at support@eurekster.com...

What's the deal. I want access to this! Or are they just referring to the most recent, popular sites page?

An excellent review on Eurekster.
Dan Keldsen, one of my eureksters, has written quite a thorough review of Eurekster. There's a lot here to absorb.

Here's the link. Here's the reblogprint:

Eurekster beta - Socially Transmitted Search?

A new Web social search engine, eurekster.com, powered by social networking technology, went live on January 21, 2004. This is an example of social software/social networking systems intended to join who you know with what you and they know -- or at least are looking for -- at least in theory.

Delphi Opinion:
The introductory message you receive upon signing up at eurekster.com states:

'eurekster learns what you and your network of contacts search for and find useful, and then provides you with more relevant search results and suggestions'

So does eurekster deliver?

In terms of 'Web experience,' eurekster has stuck close to the Google formula: a simple interface (low overhead, quick), sponsored results, and 'raw' results returned appear to be fairly high quality for a public Web engine.

The hoped for secret sauce behind the eureka experience, however, is supposed to lie in tying in the 'collective intelligence' (well, behavior at any rate) of those people in your 'social network.' Eurekster uses a form of collaborative profiling culled from the search subjects and behaviors of the your particular network group to filter results and establish relevance of specific documents.

The problems here (and the part of the blogging world attuned to social software is afire about the good/bad/ugly here) are that the benefits of a tightly-knit social network influencing your search results quickly fall apart as the network grows larger and therefore less homogenous or focused. In large networks, the suitability of leveraging what your friends/acquaintances are searching for is rapidly watered down.

The other side of the coin is that your network may not be nearly as tight as you had thought, in which case very low percentages of overlapping searches will happen, and you do not get the benefit of leveraging similarity of interest, at least in the short term.

Perhaps the biggest drawback, however, relates to 'social overhead' -- the search magic doesn't work unless you build a network for the special purpose of eurekster. For people who have already jumped into multiple social networking services (LinkedIn, Spoke Software, Ryze, Tribe.net, Google's orkut, etc.), the common pain of needing to re-build your networks in yet another service (YASNS = Yet Another Social Networking Service) are fairly high barriers to entry and adoption.

Bottom line: the tenuous 'social' moniker for eurekster seems to provide far too little benefit in improved accuracy, and smacks more of simple voyeurism rather than any real value to business searchers. The social network isn't exposed terribly well (only show what your friends recent searches, sites, top searches and top sites are, as well as 'insight' into what the wide world of eurekster searches are looking at), and doesn't have any immediate jump-off points to the sort of tribe/community facilities that other social networking services are already providing. In addition, the inability to fine-tune/filter search results via different 'social networks' you may be connected to degrades what could be a core service.

Eurekster's offer of yet another search button/toolbar to plug into your Explorer browser may or may not be of interest to some. The bar brings to mind that the case to be made for moving beyond the browser framework may be accelerated by the screenspace being lost to the proliferation of blogging and search buttons/toolbars.

In the end, perhaps eurekster's Internet focus blinded the developers to a key reality for the business searcher: what we really want is 'transparency.' We don't want to have to go to a special place to issue 'social network' searches as opposed to other searches, and if we're operating on a network (particularly inside an organization), why shouldn't a central search facility identify the important social networks and leverage them for us rather than forcing us to invite them into being?

First, I love the title of the email or atleast the allusion it makes "Socially Transmitted Search". Very clever.

On a serious note though, Dan talks about whether the "social filter" that is eurekster will deliver increased relevancy. If I had to make a ruling now, I'd agree with Dan, that the results aren't more relevant for me. However, I think its a bit too early. Firstly, my network is still small (14 people). Secondly, I am anxiously awaiting categorization, which has been confirmed, is coming.

The idea/thought that Dan brings to light in his question:

why shouldn't a central search facility identify the important social networks and leverage them for us rather than forcing us to invite them into being?

is what really sparked my imagination, though. It got me thinking about feedster again and how feedster enables me to do real-time research and have a conversation with people at the same time. Eurekster does this as well. I've made some new contacts, like Dan, that I've been able to have a conversation with. Without Eurekster, I wouldn't be able to continue this conversation. But this is attributable more to their contact system, than the search system.

But, what could complete this system would be a spider that identifies social networks that already exist in the blogosphere and visually displays them in a social network.

Good Stuff Dan!

If you'd like to learn more about DelphiGroup, go here to view some of their research and publications via guest access to their knowledgebase.

Google faster than Eurekster by a little bit.

I did a search for eurekster on google and eurekster on eurekster. To find out who is faster, I used this cool tool. (found via http://del.ici.ous)

T1 1.44Mbps 1.73 seconds

T1 1.44Mbps 1.83 seconds
Anders Jacobsen imagines a Eurekster/del.icio.us combination.

: "What comes next? Imagine social bookmarking site del.icio.us on drugs and merged with Eurekster-like search engine results enhancement functionality?! "

There are a host of social bookmarking sites: backflip.com, dudecheckthisout. Del.icio.us is a bit different because it uses bookmarklets, but they are all quite similar. There is also an open (sort-of directory) called zeal, operated by looksmart.

I'm not sure what I think about this concept. It would definitely be a feature to play with at eurekster. I think user categorized links should be just another chain in the algorithm that returns relevant results. Right now, its either a directory/bookmark manager or a search engine. When will these two merge in some way.
A few new Eureksters
A few people that have blogged about Eurekster that I've just invited to join my network. (found via waypath)

Marcus Zillman, Merrick Lozano, Niall Kennedy, WebSage

Already in my network:Andy Beal

I now have 13 people in my network and 17 invitations that haven't been responded to.
Small conversation about Eurekster going on.

Mike Perkowitz responded to my post. Thanks for the kudos, Mike. There is a bit of a conversation going on over there on Eurekster. Check it out.
What does SLI Systems do?

SLI Systems and Real Contacts are the brains behind Eurekster. So, what does SLI Systems bring to the table?

Yesterday, I pointed out that Eurekster has suggested search terms related to your original search terms. Taken from stuff.co.nz, here is what most of their business is LINK:

"He says most of SLI's business at the moment involves handling keyword suggestions. For example, if a user puts the word 'diabetes' into a search engine, SLI's software will suggest 'glucose' and 'insulin' as related search items. "

Here is a description of this Related Search service. Here is an example of it at excite.

I visited SLI's site and did a bit of digging too. Their business model is to resell their search services for corporate customers. Some of their better known customers include NBC, Quantas and Veritas. They bill themselves as "The Learning Search" (trademark).

Here's an excerpt from the site about it...

Q:What makes S.L.I. Search so great?

There are many search products available which will all rank search results differently depending on which algorithms are used. The algorithms can use link analysis or a statistical analysis of the frequency of the search terms weighted according to where the term appears or something else.

In the end, the real acid test of how good a search result is whether anyone clicks on it.

S.L.I. Systems is the only search technology that watches which results people click on and moves to the top the ones that are clicked on most. If for a particular search term the result that gets clicked on most is on the second page then it should be moved higher so people can access it more easily.

Here's a very powerful graph showing how it worked for NBC. The gist is that the users had to click on a much larger number of search results before they found what they were looking for, before SLI's system was used. It went from an average of the 12th result to the 4th result over time.

Very interesting! So, what does RealContacts do? That's for another post.

A side note: The eurekster blog is the only result for a search for "eurekster" on excite.

This page is powered by Blogger. Isn't yours?