its and consumption patterns. This profile—which could include anything from the type of tea or liquor a person likes to consume to medical conditions and sexual orientation—allows marketers to customize ads, but it also offers deep insights into life events and changes. For example, when a woman begins buying vitamin supplements, larger quantities of skin lotion, hand sanitizers, and a larger purse or bag there is an extremely high likelihood she is pregnant. In addition, analytics software has become so sophisticated it is possible to estimate the delivery window within a few weeks.

Of course, using data to predict life events could have far-reaching consequences, particularly if family, friends, or a prospective employer become aware of a sensitive lifestyle or medical issue, such as an affinity for nude beaches or a diagnosis of HIV. Worse, the data may contain errors and present an inaccurate picture that could lead to an employer refusing to hire the person or the loss of a job. As a result, advertisers are attempting to get smarter—some would say sneakier—in the way they deliver ads. Increasingly, they are including coupons and ads that are completely random or irrelevant in order to appear as though they are not spying over a person’s shoulder.

Joseph Turian, president of consulting firm MetaOptimize, says that as organizations learn to use analytics and cultivate big data, insights that would have been unimaginable only a few years ago are moving into the mainstream. There are clear advantages for consumers—particularly those looking for discounts and deals—but advertisers need to avoid stepping over the line. “People like the idea of personalized searches and advertising,” says Turian. “Many already provide data willingly for discounts through rewards programs. But they want to be in control of their destiny.”

 

Cookies, tweets, and Dollars What makes the emerging field of data aggregation and analytics possible is a spate of online data-collection techniques that revolve around IP addresses, third-party cookies, and Web tools that track consumers as they click through Web sites and interact online. Internet service providers, Web sites,

Analytics software
has become
so sophisticated
it is now possible to
estimate a pregnant
customer’s delivery
window within
a few weeks.

and advertising networks sell this data to other companies, including data aggregators. Google, meanwhile, collects data from searches and through keywords in Gmail and You Tube while Facebook has unlimited access to the mother lode of information and messages that appear on its site. Finally, Twitter recently sold its multibillion tweet archive to a U.K. firm that reportedly has more than 1,000 companies lined up for the data.

Today’s data-collection system is largely based on an opt-out model that is nearly impossible to understand or manage, many privacy advocates contend. Consumers face the daunting task of trying to decipher lengthy and convoluted privacy policies that in some cases do not match actual practices, Cranor says. What is more, data collection firms often rely on loopholes and devious methods to circumvent cookie-blocking tools built into Web browsers and privacy tools such as Ghostery. In the end, users’ attempts to control tracking and personal data often ends up resembling a game of Whac-A-Mole, Nicholson says.

In fact, half of all Internet users recall the ads they view but only 12% correctly remember the disclosure tag-lines attached to ads, Cranor reports. When she studied usage patterns she found that the majority of participants mistakenly believe that ads pop up if they click on disclosure icons and taglines. AdChoices, the ta-gline most commonly used by online advertisers (it discloses sites’ advertising methods and allows consumers to click a button and opt out), was par-

Cloud Computing
New RaaS
Pricing
Model

the resources behind cloud computing services will soon be sold in increments of seconds, according to researchers from technion-Israel Institute of technology.

Providers of cloud computing have moved from renting servers on a monthly basis to renting virtual “server equivalents” for as little as an hour at a time. But even that is inefficient, say the researchers, who presented a paper, “the resource-as-a-Service (raaS) cloud,” at the recent uSenIX hotcloud ‘ 12 conference in Boston. Providers are moving toward pricing individual resources, such as memory, within a virtual machine, and changing prices in intervals of seconds, based on shifting demand. that trend from infrastructure-as-a-service to resource-as-a-service can save buyers money, earn more for providers, and make efficient use of hardware and energy.

“clients don’t need to buy
things they don’t need, hosts
don’t need to sell them things
they don’t need, and hosts can
accommodate more clients on a
server,” says Orna agmon Ben-
yehuda, a doctoral student and
co-author of the paper.
clients would use an
“economic agent” that makes
split-second decisions on
how much to spend on which
resources, and hosts would
allocate resources based on
how much a client was willing
to pay. cloud service providers’
software could also incorporate
economic agents to represent
their own business interests.
coauthor and doctoral
student Muli Ben-yehuda says
the trend demands a lot of
cloud computing researchers.
Software, for instance, will have
to adapt to use an ever-shifting
set of resources, and workloads
will need to be carefully
balanced. the challenge, he
says, is how to turn computing
into a commodity. “how do you
make computing something like
electricity? It’s there whenever
you want it, you can have as
much as you need, and the price
is set by the market.”
—Neil Savage