david j. garcia

Until recently, the only search engine I ever used was Google. It’s not that the other tools aren’t useful–I just haven’t needed them. If it’s on the internet, Google can probably find it.

But a recent experience finally convinced me that–in certain cases–there’s a better alternative.

Instead of the vast array of information Google uses to index and rank pages, Topsy.com currently uses only one: tweets. This leads to some specific strengths and weaknesses. And if you understand them, you’ll know when a search can benefit from Topsy and when you’re better off sticking with Google.

Topsy is the better choice:

• When you’re looking for ideas.
• When you need a recommendation.
• When you want to crowd-source, but don’t have access to the right crowd.
• When you’re curious about current conversations.
• When you want to know what others are using.

Google is the better choice:

• When you’re looking for a specific piece of information.
• When you need an obscure tidbit.
• When you want to get something in a hurry, and you know how to find it.
• When you don’t want to sort through the bias of current conversations.
• When it’s important to find something that’s built trust online over time.

I recently did a massive overhaul of my online presence, and I wanted a new theme for my blog. So I did an experiment: I searched for “wordpress theme” on both Google and Topsy. First Google:

As you can see, Google did a great job listing collections of themes. The problem? I’ve seen all the collections. I’ve spent a lot of time pouring over them. I didn’t need a collection; I needed a recommendation. Enter Topsy:

Now I have a variety of resources: whatever people liked enough to share with their networks and described using the words “WordPress Theme”. The first result is not a collection of themes–it’s a specific theme that lots of people recommended. Other results included several articles that helped me decide what I wanted in a theme. Then, on the second page, result #11:

Forty-nine times, someone has recommended this resource. After checking it out and modifying it, it’s what I ending up using.

In this example, Topsy provided a better way to find what I was looking for. Of course, it has some potential disadvantages:

• It’s biased toward sources that cater to the Twittersphere.
• It’s limited to things people are posting on Twitter. If no one’s posted something, it won’t show up.
• Tweets by influential users count more than those by the rest of us. Good or bad–you decide.
• It may still be possible to game the system.

Topsy’s co-founder, Gary Iwatani, had the following response to these concerns. Since he was kind enough to quickly and fully respond, he gets a point in my book for good blogger relations and his responses printed in full:

“It’s biased toward sources that cater to the Twittersphere.”

Twitter is sometimes discounted as a serious source of information due to its “what are you doing?” origins.  The reality is that Twitter is a communications platform.  In addition to 30M+ users, virtually every major news organization and top blog uses Twitter including the following:

NY Times:   http://topsy.com/twitter/nytimes

MSNBC:   http://topsy.com/twitter/msnbc

WSJ  http://topsy.com/twitter/wsj

REUTERS:   http://topsy.com/twitter/reuters

ABC:  http://topsy.com/twitter/abcnews

CBS:   http://topsy.com/twitter/cbsnews

Time Magazine:   http://topsy.com/twitter/time

The Huffington Post:   http://topsy.com/twitter/huffingtonpost

“It’s limited to things people are posting on Twitter. If no one’s posted something, it won’t show up.”

This is technically accurate but the wide adoption of Twitter combined provides very good coverage of search terms and the fact that Topsy generates search results mined from real-time conversation streams is a big advantage - not a disadvantage versus traditional search.   Traditional search algorithms based on document link structures are inherently slower to respond to relevance changes (such as breaking news).  Furthermore, there tends to be an old document bias in traditional search results as important documents of the past (which have accreted significant PageRank value) are still considered relevant today since fresher contemporary link citations from the social web are ignored.   (Eg. Google first page search results for “apple” still shows a result for a battery replacement program for the iBook G4 from 2003.)

“Tweets by influential users count more than those by the rest of us. Good or bad–you decide.”

This is a good thing as it maximizes the “signal to noise” ratio in the search results and eliminates spam.   Every author, regardless of influence, is indexed and discoverable in Topsy under their own author page.  Example:  http://topsy.com/twitter/davidjgarcia

“It may still be possible to game the system.”

No system is fool proof however Topsy’s influence algorithms are very robust and much harder to game than traditional search engines.

One point I found interesting–if you use Twitter, your author page is a neat way to find out how many people are tweeting the same links as you (example).

What do you think of Topsy? Will you use it? Have you had any success using it instead of another option?

(More about Topsy at TechCrunch)

——-

Like this post? Tweet it!