- The Matrix age? That was yesterday - May 28, 2018
- From boycott to buycott – Digital Democracy as a 21st century tool for Sustainable Capitalism - March 28, 2018
- Interview with an Alumna: Ellen Klijnstra (KPMG) - December 11, 2017
- Data-driven capitalism: the Invisible Hand in 2017 - November 8, 2017
- Why care about Blockchain? [Longread] - October 16, 2017
- Business Night, building a legacy - April 12, 2017
- Interview: Sander Peels - February 20, 2017
Machine learning, blockchain, AI – there seems to be a new buzzword every week. Is the world moving so much faster than 10 years ago? Depends on who you believe. What’s for sure is: some of those buzzwords have long transcended the stadium of ‘cool ideas’. They are transforming our very reality, boosted by corporate adoption.
Maersk is cooperating with IBM for a world-wide rollout of blockchain technology to enable smart contracts; artificial intelligence (AI) has led to robots baking better hamburgers than the average student; and banks and governments are optimizing their fraud detection rates through predictive analytics. What’s next?
Student-economists of Tilburg University are faced with a tough choice these days: shouldn’t I take the step to the Master in Data Science instead of ‘just doing Economics’? With ceremonial pride, the Jheronimus Academy of Data Science witnessed its life light less than two years ago. And if we may believe the average CEO, CIO or CTO (or professor Eijffinger, for that sake), data science is the go-to-studies these days. There seems to be a reason to believe the statement.
Computer says ‘no’
The core focus of data science involves the investigation of, and experimentation with, the wide array of predictive analytics. Those includes evolving techniques such as deep learning, neural networks and clustering algorithm – and machine learning, indeed.
Machine learning (ML) is a subfield of computer science that gives computers the ability to learn without being explicitly programmed. It evolved from the study of pattern recognition and proves that algorithms can learn from and make predictions on data. And, as they begin to become more ‘intelligent’, these algorithms can overcome program instructions to make highly accurate, data-driven decisions. Now that’s worth something.
Machine learning is one of those buzzwords which was ranked high on Gartner’s Hype Cycle. But it’s become a serious innovation in the meantime; in 1990, IBM’s ML-driven computer called ‘Deep Blue’ managed to beat the world chess champion Kasparov in subsequent contests. And Google, alongside its chat-bots and newly released Google Assistant, is of course best-known for its sophisticated search engine which we all use on a daily basis. Its suggestions, often eerily right and sometimes rather surprising, are empowered by a complexity of algorithms. In Google Images, the search engine analyzes color and shape patterns and pair that with any existing schema data about the photograph, thus enabling it to understand what an image actually is. A picture is worth a thousand words, indeed.
We have long embraced such innovations from major tech firms such as IBM and Google, whose core business is to innovate on technology. But what’s the action on the business-to-consumer domain? How do high-tech innovations reach our consumers domain? The answer is: through all kinds of ways. But one of the coolest and high-impact ML applications must be Wall Street Journal’s (WSJ) recent paywall revolution.
Not your average paywall
As you may know, newspapers use online paywalls as an important means to generate revenues. Those occur in different shapes: as hard paywalls (you see nothing unless you pay; Financial Times), metered models (you get a number of articles for free; Financieel Dagblad) or freemium models (you get access to basics and pay for selected articles; Blendle).
WSJ has added a new option to the color palette: the adaptive paywall. They can flex their amount of sample articles based on reader engagement, while to the reader, it appears as a one-size-fits-all freemium model.
Over the last few years, the WSJ has operated a deliberately leaky paywall that has served as a sandbox of data collection and subscription sale experiments. Reader subscription intent is measured on three levels: they are cold, warm and hot. It houses an ML algorithm that measures reader activity across 60 variables such as visit frequency, recency, depth, favoured devices and preferred content types. This mix of variables adds up to a propensity-to-subscribe score that then helps inform how many sample stories users can access.
In other words, the activity you display as a reader determines the amount of sample articles you can read. Quite clever, eh?
From nice to necessary
It’s actually much more than a clever gimmick to WSJ. The problem with other subscription models is their inflexibility – they assume that every buyer has the same ‘tipping point’ to becoming a paying reader. This adaptive paywall differentiates between superficial ‘news scanners’ who are quickly looking for some information, and readers who are interested in specific in-depth articles in e.g. arts and culture. If one searches a large number of articles in the latter category within a short amount of time, the WSJ paywall is likely to ask you to become a paying member – while the former group may be free-wheeling for a longer time.
One of WSJ’s major revenue issues arose from the fact that people had all kinds of side-entrances to reaching the articles for free. Since the implementation of the ML technique, subscriptions have grown by over 350,000, mainly from students accepting ‘flash sale’ offers. The firm believes that this young reader engagement will pay off gradually, as they mature into life-long paying members.
That’s a time-consuming process. In the meantime, the algorithms continue rattling and rumbling to guide as many readers as possible past their own, unique tipping points.
So prof. Eijffinger may have been right after all: studying Data Science seems the best way to switch from the readers’ position into the shapers’ position, and to determine the news instead of merely consuming it.