Big data and in-sourcing January 30, 2013Posted by Cameron Shelley in : STV302 , comments closed
A recent piece in Wired makes an interesting point about the influence of information technology on corporate size and structure. Most attention on this topic has been focussed on the pressure that IT might create for companies to outsource and, therefore, to downsize. However, the arrival of the ‘net has created circumstances that permit big companies to get even bigger.
On the first point, we all know how IT and networking can lead to outsourcing. Once information can be represented in digital form, it can be quickly transmitted around the world via the ‘net. A good example might be how McDonald’s has outsourced the handling of requests at many of its drive-thru windows. Orders shouted at microphones in Honolulu are piped to call centers in California for taking (and up-selling) and then transmitted to the restaurant for the staff to assemble.
However, the article points out, the same technology can also create an economic pressure for companies to increase in size. In many cases, the data that companies might transmit away over the ‘net could be mined for valuable information. Companies can analyze the data to increase their efficiency and competitiveness in several ways:
But the rise of a different set of technologies—enterprise software to tie organizations together, along with sophisticated tools for measurement, analysis, and prediction—has conferred immense advantages on the Goliaths of American industry. In part that’s because they do more business, and more business means more numbers: transaction data that can be used to develop more refined credit models and consumer insights and to otherwise outsmart smaller competitors. In an earlier era, a regional bank might have had an advantage in making loans to local businesses on the basis of its intimate, on-the-ground knowledge of borrowers and the community. But nowadays, computer credit-scoring models calibrated on the outcomes of millions of prior loans can predict better than any human being whether a loan will be repaid.
Perhaps more important, though, is the way that data helps to manage workers, allowing these giant corporations to become significantly more efficient. Technology makes it easier to coordinate the activities of employees and to monitor in real time whether they’re doing what they’re told.
In brief, big companies mean big data, and big data can mined for profit.
This observation would be trivial when applied to big corporations such as Google or Facebook, whose business is exactly mining data from users that is of interest to advertisers. However, it seems less obvious when applied to other businesses.
I wonder if companies like McDonald’s would even reconsider some of their past outsourcing decisions. Facial and voice recognition software could allow the company to track the orders of drive-thru patrons, yielding information that could be mined for later use. Why give that away when it could be kept in-house? However, in order to capture and utilize data in this way, the company must grow as big as possible.
In his little speech, Five things we need to know about technological change, technology scholar Neil Postman claimed that technological innovation, including computerization, often seems to work in favor of large corporations. I think that this example would fit nicely with his views.
Are tedious captchas a good thing? April 26, 2010Posted by Cameron Shelley in : STV302 , comments closed
A recent New York Times article comments on how spammers may hire people to solve captchas on Websites. Say you were a spammer and wished to send people an email that originates on the New York Times (NYT) website. You know, “Your credit card number has won a contest; log in here to collect your prize from the New York Times”? The problem is that, in order to use the NYT email-an-article service, you must solve one of those little puzzles or CAPTCHAs.
(Photo courtesy of Ryan Staake; Wikimedia Commons)
These little puzzles are designed to be easy for people but difficult for computers. So, automating your spamming process is hard to do. Unless you can hire people to do the puzzles for you. Some spammers do just this, using impoverished people in India or Bangladesh to spend their days deciphering these little graphics, over and over again.
Many things could be said about this situation. On the one hand, such work seems degrading, an example of automation gone wrong resulting in the creation of mindless piecework through deskilling. On the other hand, if people are truly in such a poor situation that such work is economically attractive to them, then who is to say they should not do so? (Then there is the harm caused to the victims of the spam to consider.)
Interestingly, the article concludes by noting that the very tediousness of the CAPTCHA solving process impedes its economic sustainability:
That view was confirmed by an executive at one south Indian outsourcing company that advertises its captcha-solving prowess on a Web site. The executive, Dileep Paveri, said his firm had stopped offering the service because it was not very profitable.
His company, SBL, which is based in Cochin, got about $200 a month in revenue for each of the 10 employees it had hired to decipher the puzzles on behalf of a Sri Lankan client.
“We found that it’s not worth doing,” said Mr. Paveri, a manager in SBL’s business process outsourcing and graphics unit. Moreover, he added, “after some time, the productivity of people comes down because it’s a monotonous job. They lose their interest.”
You may dislike CAPTCHAs for being so boring. However, their very dullness may be one of their chief virtues, preventing them from becoming the foundation of an exploitive industry.
Outsourcing marking April 9, 2010Posted by Cameron Shelley in : STV302 , comments closed
The Chronicle of Higher Education posted an article about outsourcing the marking of assignments from Universities to private companies. These companies employ people with degrees in various fields whose job is to grade assignments according to a rubric provided by the professor who made up the assignment.
Some bonuses of the arrangement include: students get their assignments back earlier, comments are more thorough, the markers are experienced and can offer more instructive feedback than TAs at the university where the students are located. Of course, outsourcing may be cheaper than employing TAs locally.
Potential problems include a detachment between professors, TAs and students, and threat that fewer TAs would be hired locally, which would be hard on the graduate students who depend on TAs for funding.
An non-obvious benefit observed at universities using the outsourced service is a reduced dropout rate among undergrads:
Holly Suarez, an online instructor of sociology at West Hills, says retention in her class has improved since she first used Virtual-TA, two years ago, on weekly writing assignments. Before then, “I would probably lose half of my students,” says Ms. Suarez, who typically teaches 50 students per class.
Because Virtual-TA provides detailed comments about grammar, organization, and other writing errors in the papers, students have a framework for improvement that some instructors may not be able to provide, she says.
With dropout rates climbing here at Waterloo (and elsewhere), this point should arouse some interest locally.
What do you think about Waterloo outsourcing grading?