A few weeks ago I had the great pleasure of attending the absolutely terrifying conference presented by Kevin Slavin from the MIT Media Lab, who discussed how algorithms and machines are progressively taking over our lives. He outlined a series of examples to support his theory.
It all began in 1994, when developers were obsessed with a game: battle of the algorithm. It all had to do with developing a program that would battle another program by anticipating what the other program was going to do. Slavin explains that what could have been a really cool game has since spiralled out of control.
The most well-known example is the stock exchange, where algorithms govern markets, acting on their own to make decisions that seemingly come out of thin air. Last year, algorithms went into a tailspin no less than 18,000 times, meaning as many micro-crashes within fractions of seconds where stock prices may drop a few cents before climbing back up. As long as everything goes well, that is.
But Slavin demonstrates that this new order doesn’t only apply to the stock exchange. Left up to an algorithm to determine book prices, the stimulating genetics manual, The Making of a Fly, hit the astonishing price of $23 million on Amazon.
An elevator manufacturer developed an algorithm to predict which floor the elevator was going to stop on. He even went so far as to remove the floor buttons, which had become redundant. The problem is that people panic when they get into an elevator without buttons, especially when there isn’t a button to stop the elevator in case of an emergency.
Over 60% of films viewed on Netflix are chosen based on a recommendation algorithm.
And now, algorithms are going one step further to estimate script potential for new movies and TV shows (a previous article recounts how Netflix made its production choices for House of Cards based on viewer data). In the end, Netflix will end up just like the 1994 game and the stock exchange, with production algorithms locked in loops with distribution algorithms and without any human oversight…
Would the future be so bleak if machines finally ran the world?
Slavin gives us reasons to hope. For over two years now, the best chess algorithm is better than the best human. A competition was organized; it was open to anyone who wanted to enter – humans, machines, groups of humans, groups of machines, combinations of humans and machines – and against all odds, two amateurs with three machines and a mediocre algorithm won. They beat the best algorithm in the world. Closer to everyday lives, Slavin states that the best way to “beat” Google’s search algorithm is still a human who can “hack” system logic with the help of some good software.
Slavin’s conclusion? We shouldn’t be afraid of machines, but rather, realize that they are only truly powerful when humans are involved. Citizens must be vigilant to stop the frenzy of autonomous machines running the world and put pressure on controlling bodies (like the SEC for the stock exchange or civil aviation authorities) to maintain the human factor in the equation.
Read more from Jean Pascal at A Nos Vies Numériques.
Tags: Algorithms, Big Data, Future Technologies, Kevin Slavin, MIT Media Lab, Netflix
In an article entitled “Giving Viewers What They Want”, The New York Times reported that Netflix created its first blockbuster television series, House of Cards, based on the tastes and viewing habits of its 33 million users. According to an independent study cited by TechCrunch, the series has already been watched by 10% of Netflix subscribers. This is a very good score, especially when you consider that the series will be accessible to both present and future Netflix subscribers.
Normally, producers and broadcasters will attempt to determine what audiences want based on past successes, awards, audience ratings and the bankability of actors and directors. Netflix, however, has made all of its production decisions for House of Cards based on the extensive amount of viewership data they have collected. And so, we are left with this paradox: if viewer data becomes the determining factor of what television shows or movies get made, it will inevitably lead to the standardization of TV programming.
That said, House of Cards is complex and dark and not what many would consider to be an “easy” series to watch. Nor is it one that production companies would readily sink more than $100 million into its creation. What remains to be seen is if this magic recipe is reproducible – something which will require viewers to keep a close eye on future productions from the house of Netfilx. After all, the history of filmed fiction shows that you should never rule chance out of the equation.
Read more from Jean Pascal at A Nos Vies Numériques.
Tags: Big Data, Connected TV, House of Cards, Netflix
“Watching TV” is pretty simple from the viewer’s standpoint, but not so much anymore for advertisers. On the contrary, it has become increasingly complex with regards to audience measurement.
Several overlapping layers of complexity come into play:
In order to see how new usage habits impact our viewing behaviour, we should logically take into account all of these elements when defining and/or measuring a program or channel’s audience. Unfortunately, this is impossible to do with existing measurement methods (panels and cookies) because they split up these uses in order to obtain true – albeit partial – data.
The best example of this narrow perspective is from the United States: bandwidth usage analysis shows that Netflix, an on-demand video service available by subscription on numerous platforms, uses close to 30% of available Internet bandwidth. This percentage is clearly huge, demonstrating a significant and repeated use, much to the detriment of traditional channels. This trend extends outside of the United States as the global explosion of multi-screen usage enables people to enjoy an enriched and social viewing experience. Also, as programs are becoming events in themselves, this multi-channel and multi-social experience is occurring before, during and after the show.
Current television and web audience measurements do not allow for these overlapping phenomena to be taken into account; Netflix doesn’t even rank among the top ten online video sites as determined by Comscore, an online audience measurement institute. This change in viewing behaviour is not yet coming up in the numbers, and the average number of hours spent watching “traditional” TV still remains higher than that of online viewing. But, if we look at the same data from a different perspective – say from a generational, perceived market value, growth or advertising impact point of view – a different reality, one that has not as yet discovered its performance indicators or instruments of measure, comes to light.
As it stands, only cable and telcos have access to real information, which remains somewhat hazy for advertisers who see their advertising performance diminish with each passing month. This can’t go on forever, but really – who’s going to rock the boat?
Tags: Comscore, Netflix, Online TV, Television ratings, TV advertising, TV viewing
Illustration: Dondy Razon
At a time when Google TV is searching for its own identity, Apple is set to re-launch its TV box, Netflix is increasing its audience and making more money and newcomers such as longtail.tv are coming out with new services, the question becomes more prevalent: what is the purpose of television connected to the Internet?
Obvious Uses
When it comes to operating, the “good” ISP boxes allow the user to record a program for later viewing. Some can even be operated at a distance, from a computer or smartphone. Replay televisions allow the user to watch a program after its original broadcast, generally for a period of seven days. Some channels even offer Internet prototypes during major sporting events that allow the user to rewind 15 minutes to catch what they might have missed. And you just have to look at how fast video stores are closing to be convinced of the success of online video rentals (both legally and illegally).
Enhanced Uses
The simple act of consuming online videos is improved by the ability to multitask. We watched the major Internet meetings while viewers’ tweets rapidly flashed on the screen. We’ve already seen this type of interaction with the live broadcast of SMS messages during talk shows. But when it comes to meetings, sports replays and reality TV, the future is bright. Thanks to connected televisions, we’re going to see a boom of different multimedia combinations that will allow users to switch from watching videos to texting to looking at or editing photos, one action after another or even simultaneously.
The Projection
Audiences will continue to spread out. At first, small channels will suffer from the direct competition. But it won’t be long before the larger outlets find it difficult to secure audiences for their advertisers. As a result, the film industry will lose its principal source of funding.
This dooms-day scenario is the same one that impacted music and media creation. Major players in audiovisual creation must be encouraged to invest – and fast – in the development of quality multimedia programs that utilize new technologies to enhance creation and industry development. Just like with music, new formats, viewing contracts and alternative financing must be developed beyond the present experimental approach.
But from a glass-half-full perspective, an enhanced and uninhibited art form should arise from all of this, a break from censorship that is obvious in the decision-making process of broadcast networks. It allows the average person to use connected TV much the way the Internet operates.
Final Note
To all my geek friends, it won’t be long before tablets will be super remote controls that transmit enhanced video streams while keeping apps and associated content at your fingertips. Could the next Apple TV look a like the iPad?
Read more from Jean Pascal at A Nos Vies Numériques.