First the usual disclosures. Google is an investor in Fon. Twitter has no relationship with Fon, but I know Ev Williams a little and some investors like Chris Sacca well. Personally I think that neither investments nor friendship will taint the objectivity of this post, but disclosing is better than non disclosing.

Now the post.

Google knows that Twitter is both an opportunity and a threat. Twitter is a threat because it is instant search – compared to Google’s crawled (slow) search – but also because, in many cases, Twitter yields better results. For example, what is the point of collecting a 3-year link history for a fashion brand and give search results if the most relevant information about that fashion brand may be that 5 minutes ago a hugely followed celebrity says she’s crazy about it. Twitter is now big enough to move the needle in the real world quickly enough for Google to miss it.

Now here is the opportunity:

Google is great at creating hierarchies of information. The original idea of Google, which comes from science, is that “he/she who is linked to the most must be saying the most relevant things”. Now what is missing is that same analysis but cranked on Twitter data.

How would I refine search in Twitter? To me there are two measures of Twitter relevance. One is how many followers a persons has, and the other one how frequently this person is retweeted. In my opinion, the opportunity for Google is to use its computing power to come up with very relevant, instant answers to problems using Google results, Twitter results and when using Twitter ranking according to followers and retweets. To that, it should add PINGED results, namely results whose location on the web has been volunteered as in blogs or news. To see a ping search engine check out Technorati in which I am an investor.

Google created a meritocracy on the internet. If you have a high Google ranking, what you say, for example in your blog, matters more in search. Well, I think Google is the company most equipped to blend crawled search with real time search by combining Google results, Twitter results and PINGED results (results that were not crawled but picked up from recent blog and news pings).

Here’s a small experiment I did called “unfolding news” that shows the beginning of a twitter+news+blog results. This is what you get when you search for “ash cloud” in unfolding news. This is what you get when you search for “ash cloud” in Google. Unfolding news is an experiment that does not use Google results, but I find it more useful than Google for learning about things that are happening or “unfolding”. Google could do a much better job blending “old web” and “new web” sources.

Conclusion: instead of wasting time with Buzz emulating or copying Twitter, Google could complement Twitter with what Twitter needs most, making sense out of searching something in Twitter.

Reblog this post [with Zemanta]

Follow Martin Varsavsky on Twitter:

No Comments

Martin Giorgetti on May 17, 2010  · 

I totally agree with your entire post.

I would add that Google could also use the whuffie bank reputation system, instead of building a new one, to refine search in Twitter.

3.0 rating

Manuel Jiménez on May 17, 2010  · 

The only flaw to your otherwise thoughtful post is how to define the value of different twitter accounts.
Is Ashton Kutcher more influent than, say, Russell Crowe?
Trust in twitter is build around personal choice, not everyone (fact is… does anyone?) follow people because of their twitting value. Most people follow friends, acquaintances or just famous people.
Then you have lists… It’s too complex for giving away popularity by mere follower count (which you can buy) or RT numbers (which you can also buy).
I hope Google takes another thinking step and surprises us with an astonishing view of twitter massive information. Otherwise twitter will be another pagerank = valueless

3.0 rating

Osvaldo Osorio on May 17, 2010  · 

Uhmm…. Buzz was a great experiment for Google. I think there is more hidden potencial in this App. But it’s your proposal sound as a good idea: complement the big and huge information generated every second and measure tendencies with more variables, etc.

3.0 rating

Ouriel Ohayon on May 17, 2010  · 

while doing i observed that a strong level of relevance was most of the time brought by users that have a ratio follower/following superior to 2. The number of followers alone is just not enough and can be easily gamed. So this + RT factor would be indeed a great line

3.0 rating

Martin Varsavsky on May 17, 2010  · 

agreed Ouriel that is also a measure of Twitter relevance, it’s trial and error work but if anyone can make sense of the Twitter data is Google

leovernazza on May 17, 2010  · 

Both companies are already doing what you suggest.

Twitter has announced their “resonance” during Chirp. They take into account many other things to discover the tweet relevance (i.e. avatar clicks, hashtag reuse after tweet, link clicks, etc).

Also, I am not sure that the followers number are a good indicator. If all your followers has too many people, they might not be following you, really. I would prefer “less but good” followers. It is not clear to me, that the followeres qty simplification is good enough (when you actually can collect much more information).

And to conclude, Google is also doing what you suggest. But you have to click on “latest results” in the results menu. Here is the link for your “ash cloud” search:


3.0 rating

Martin Varsavsky on May 18, 2010  · 

If both companies are already doing what I suggest I guess I am on to something, but the key here is not the idea, it’s the execution of it.

leovernazza on May 18, 2010  · 

I just commented the post in case I was missing something… if you are saying there is a huge potential in “real time search”, well, yes, I agree…

This post with internal Twitter’s documents shows many insights about their strategy & opportunities:


3.0 rating

carlos on May 18, 2010  · 

Hi Martin,

You should take a look at the latest version of kikin … I believe that we might already have solved the problem you are describing above.

First, I would argue that YOUR feed/social graph and affiliations matters the most to you [we call this “favorites”.] So, if you are searching for/interested in “Barack Obama” at Google, Bing, Amazon, etc, kikin delivers relevant items relating to “Barack Obama” from YOUR Twitter, FB, Google Reader, etc. feeds and other relevant sources … we call this “favorites”. Second, not only do we allow you to get YOUR content from Twitter, FB, etc. at points of relevance (and more importantly “moments of intent”) — which by definition requires pointed searching of those feeds — but when available, we also deliver “relevant surprises”, i.e. deadly relevant content from your favorite sources (e.g. the general Twitter feed), or from other sources that you might not even have as a favorite (e.g. Yelp! or Citysearch for restaurants; Shopping.Com for commerce; etc.). And we present this content in a rich experience and make it easy to consume, e.g. no need to click away to watch a YouTube video … just watch it right here! Third and finally, we allow you to share your favorites/relevant surprises within your social graph … “ask and tell” is a key element of social networking!

Sure, is only in quiet beta, but we are constantly improving the service:
1. We recently expanded beyond search to 15 targeted sites (Amazon, IMDB, etc.) … and we will soon be releasing it to all ComScore 1000 sites.
2. We are constantly adding content … Citysearch and Yelp! recently came on board to provide local content in the U.S. .. we have 25 more in the pipeline … and we will soon issue an open content/commerce/services API through which any and all could join kikin.
3. Our quiet beta started in Germany, and expanded to Brazil … kikin is now entrenched in U.S. (our main focus for this year) and will soon be coming to you wherever you are.

Please check out and send us your feedback. Un abrazo …


3.0 rating

carlos on May 18, 2010  · 

P.S. 500+K users are loving kikin in quiet beta … we are executing, dude!

3.0 rating

cheap nike sb dunks on June 7, 2010  · 

I like shopping and is my favorite place to shop. Everything nike dunks shoes and Vibram FiveFingers you can imagine is offered

3.0 rating

Leave a Comment

Español / English

Subscribe to e-mail bulletin:
Recent Tweets