Twitter Makes Better Recommendations of Whom to Follow

Print Version
Share to a friend

Have you ever wondered why Twitter sometimes suggests that you follow Justin Bieber

and Kim Kardashian when you’re actually interested in people with a wee bit more substance,

like Margaret Atwood and Clayton Christensen?

Twitter thinks it has found a way to make accurate and useful recommendations of whom to follow.

Twitter will do it by tracking your Web browsing habits, if you let it, and will let you completely

opt out if you don’t.

Suggesting someone meaningful for you to follow when you first sign up for a Web site like

Twitter is a challenge for the site, because it knows so little about you.

This is called the Silo Effect.

 

When a person signs up for a new Web site or service, they come with no information about

their interests, even though other sites they have visited know their likes and dislikes.

Like grain on a farm kept in separate silos, people’s personal information remains siloed

on different services online.

For example, Amazon knows you like books about the future, YouTube knows you like surfing videos,

iTunes knows you listen to Snoop Dogg. But none of these services share that information with each other.

And they certainly don’t share it with Twitter.

Signing up for Twitter and asking them to suggest relevant people to follow is like going to a

grocery store and asking a total stranger in Aisle 12 to suggest your favorite foods to buy.

But Twitter might have just solved this problem and, more important, done so without any

sneaky antiprivacy practices.

When new users sign up for Twitter, the company will deliver a cookie to their computers that

will allow it to look at the sites they have visited over the previous 10 days that have ”Tweet”

buttons or Twitter widgets. Thus it will know about their previous browsing habits.

 

The privacy watchers out there might quickly hold up a red flag, but Twitter seems to

have anticipated this, announcing earlier Thursday that it had activated the Do Not Track

feature in the Web browser, which lets people opt out of cookies that collect personal information.

New users will also be given the option to expose that information by clicking an area called

“Tailor Twitter based on my recent Web site visits.”

In other words, Twitter will track your reading habits if you want, and let you completely opt out

if you don’t.

In a blog post on the company’s Web site, Twitter said that the information collected was not

being used for advertising purposes, and that Twitter had no plans to use it for

advertising in the future.

Othman Laraki, Twitter’s director of growth and international, also explained in the blog post

that with these latest updates the company hoped to create a more meaningful set of suggested

followers that reflected people’s interests. The new feature will be especially important for new users.

“To make it easier and faster for everyone to get started on Twitter, we’re beginning some experiments

with tailored suggestions in a number of countries around the world,” Mr. Laraki wrote in a blog post.

“By recognizing which accounts are frequently followed by people who visit popular sites,

we can recommend those accounts to others who have visited those sites within the last 10 days.”

So if someone goes to, say, The Atlantic, and to a San Francisco sailing club Web site, and then

signs up for Twitter, that user will no longer be told to follow Mr. Bieber and Ms. Kardashian,

unless, for some strange reason, he or she really wants to.

 

Source:http://bits.blogs.nytimes.com/2012/05/17/twitter-experiments-hopes-to-solve-the-silo-effect/