The new head of Google search, Prabhakar Raghavan, has been at the forefront of search related research. A focus of his research includes co-authoring research on link analysis as far back as 2000 as well as the Yahoo Research on trust and distrust in links.
Prabhakar Raghavan has a PhD, has authored books and research papers, is a member and leader of various computer science organizations and so on.
But what we in the search community are interested in is who is this person and how will he impact search?
What is of more interest to us in the SEO community is getting a picture of who the new head of search is because wherever he leads we’re going to have to follow along.
That’s the focus of this article.
It may be useful to understand who is in charge at Google Search if your business involves search marketing.
Prabhakar Raghavan was formerly at Yahoo Labs before coming to Google. Yahoo Labs gave us such innovations such as TrustRank and research into predicting how site visitors will scan a web page or bail, which is useful for increasing engagement.
Discovering the kinds of research that interested him helps paint a portrait of the person who is now in charge of Google Search.
For example, one of his research papers from 2000 is called, Graph Structure in the Web. It’s concerned with how links can provide useful information for search engines.
That’s interesting because it demonstrates his knowledge of links and link spam.
The abstract states:
“The study of the Web as a graph is not only fascinating in its own right, but also yields valuable insight into Web algorithms for crawling, searching and community discovery, and the sociological phenomena which characterize its evolution.”
Other research he co-authored looked at user behavior analysis and predicting what they would do.
A research paper he published in 2012 explored determining a user’s next action. The paper is called, Are Web Users Really Markovian?
What’s interesting about this research is that he questions basic assumptions of PageRank that haven’t been studied before and demonstrates that algorithms like PageRank have not taken into account realistic models of user behavior.
Here’s what it says:
“Classic Web algorithms such as PageRank use this model. Some prior work… offers weak evidence in support of users’ behavior being Markovian.
We know of no prior work that has examined whether Web users’ behavior is in fact truly Markovian, thereby justifying the assumption implicit in PageRank and other algorithms.”
What makes this fascinating is that it reveals a curious mind unraveling not just overlooked flaws in commonly accepted prior work but outlining the possibilities for information retrieval in a way that better serves users by actually trying to understand users.
I think if you look at the research that he’s been involved in, a common thread is trying to understand how users behave when they look at search results or when they click on a link.
Understanding users in order to better serve them seems to be an underlying concern.
Here’s another example of his research. It’s a paper from 2011 called, Optimizing Two-Dimensional Search Results Presentation
This is research into how users scan web pages and search results. The research paper presents product and image searches as an example of how search engines were presenting data to users in a way that may not be optimized for how users scan web pages.
He showed how search engines ordered results from the most relevant in the top left and proceeding in a left to right and top to bottom manner, with the least relevant images or products lower down on the page and more to the right.
Raghavan’s research showed that users actually scanned in a pattern that resembled a triangle, with the wide portion of the triangle at the top. He also noted that there are elements of randomness into how users scanned.
In 1995 he co-authored a book with Rajeev Motwani called, Randomized Algorithms.
A review of the book in American Scientist stated:
“The techniques described by Rajeev Motwani and Prabhakar Raghavan are wide-ranging and powerful, so this book is an important one. As far as I have been able to find out this is the only book on the entire subject … this excellent volume does us proud!”
Raghavan also co-authored Introduction to Information Retrieval, a copy of which is available online.
Although published in 2008, the information contained in this book is a great starting point for those who wish to attain a foundation of solid information about how search engines work, information that is free of common SEO myths and assumptions.
It’s not an easy read but as long as you can look up terms like “Markov chains” to understand the contours of what it is, anyone can get a working idea of how search engines work which in turn will make you a better SEO because you’ll be able sniff out fake SEO ideas.
Of particular interest is the section on Link Analysis which may be useful to those who want to learn straight from the scientists like Raghavan who were instrumental in development in many of the ideas that are a part of modern information retrieval.
Prabhakar Raghavan is someone who is deeply involved in link analysis, user behavior modeling and out of the box thinking about how search engines can do better.