Eurekster Swicki Blog

Folks - it's about time we took down the beta shingle from our logo. Why? We launched the swicki platform two years ago with a clear vision, and slowly but surely we've grown the platform to fulfill that vision. Swickis now do what they are supposed to:

Allow anyone to create and maintain a vertical, topically focused search engine.

And we're not talking about site search here - we're talking build a web search engine that finds the most relevant content published by anyone, and place advertising alongside that content to produce a viable business for the swicki's owner. The new search engine business model quite distinct from the old publishing model.

Inspired by the elegance and success of blog platforms in enabling non-HTML-saavy users to create their own online publication, we knew we could empower non-Python-saavy users to own their own search engine. So that's what we did. The original swickibuilder allowed anyone to create a search engine in 3 easy steps, and that was good.

But when we looked at what the builders had created, we found that there were tweaks here and there that would vastly improve their swickis. We also saw that the work necessary for builders to maintan their community-driven swicki was too time consuming. What we had was a good, but obviously beta product. We listened to the feedback from our builders and members and we saw that we needed to focus on making sure that:

1. the swickis builders made were the best they could be
2. the swickis were always getting better, automatically

We validated many of our assumptions along the way, and learned plenty more. Big and little things, like:

• There is certainly a market for topical vertical search engines - the public has created over 100,000 of them, and over 9 million users searched them monthly.
• Improving search results by aggregating the wisdom of crowds around a topic is powerful, but needs to be harnessed and channeled properly. Otherwise, the search engine looses focus.
• Users are more likely to vote or comment on a search result when they see it will immediately impact the experience of a larger audience.
• The spider-based model for indexing web content breaks down when you want to find the freshest content because the spider has to crawl the site a hundred times a day overloading the site's servers.

We took what we learned and reinjected it back into the product, constantly evolving and improving it. So coming out of beta, it's not about any specific handful of features, it's about an evolutionary milestone. But here's a short list of my favorite features that we upgraded for this release:

Building
When you enter a URL in the "my site" field, we check to see if you have blog roll and add those sites to your favorites. We also grab any meta tags from your and your favorite sites and add them to the buzzcloud. You get more relevant sites and more search terms, automatically.

Maintaining focus and improving relevance
With our new relevance algorithm - codename Mobius - we constantly evaluate the search stream against the latest click and voting behavior and automatically favor results and buzzcloud terms that are both relevant and popular.

Freshness
We are now indexing your RSS feed so that your latest stories show up immediately in your swicki. If you entered additional favorite websites when you trained your swicki, we'll index their RSS feeds as well. We place a little timestamp with the publish date next to stories that we find this way.

Grab it!
We make it easy for you to post your swicki's buzzcloud on any number of blog as well as widget distribution platforms. The more searches you drive to your swicki, the better it gets.

Welcome to the new Eurekster swicki: it keeps getting better, and now out of beta. What are the most important features for you? We'd love to hear.