9.10 “Networked Minds” Require a “New Economic Thinking”
187
All of the above implies the need of a “New Economic Thinking” (“Economics
2.0”), which enables us to organize our economy more efficiently and effectively
18
(see the next chapter and Appendix 9.2). I strongly believe that we are heading
towards a new kind of economy, not just because the current economy no longer
provides enough jobs in many areas of the world, but also because information
systems and social media are changing our interactions and opening up entirely new
opportunities. In particular, to cope with the increasing level of complexity of our
world, we need to catalyze collective intelligence. In fact, the principle of “networked
minds” makes it possible to bring the best ideas and skills together, and to leverage
the hugely diverse range of knowledge and expertise in our societies. Let us discuss
this now in more detail.
9.11 The Netflix Challenge
One of the most stunning examples of collective intelligence is the outcome of the
Netflix challenge.
19 The video streaming company Netflix was trying to predict what
movies their users would like to see most, based on their previous movie ratings, but
their predictions were frustratingly bad. So back in 2006, Netflix offered a prize of
$1 million dollars to any team that could improve their own predictions of individual
movie ratings by more than 10%. About 2000 teams participated in the challenge
and sent in 13,000 predictions. The sample data contained more than 100 million
ratings of almost 20,000 movies on a five-star scale, made by approximately 500,000
users. Netflix’ own algorithm produced an average error of about 1 star, but it took
three years to improve it by more than 10%.
In the end, the prize was won by a team going by the name of “BellKor’s Pragmatic
Chaos”. In the process, a number of really remarkable lessons were learned. First,
given that it was very difficult and time-consuming to improve the standard method
by just 10%, Big Data analytics wasn’t good at predicting people’s preferences and
behavior. Second, even a minor improvement of the algorithm by just 1% created a
significant difference in the top-10 movies predicted for each user. In other words,
the results were very sensitive to the method used (rather than stable). Third, no
single team was able to achieve a 10% improvement alone.
9.12 Diversity Wins, not the Best
The final breakthrough was made when the best-performing team decided to average
over their own predictions and those made by other teams that weren’t as good.
Surprisingly, collaborating with inferior teams boosted the prediction performance,
18 Helbing [16].
19 See https://en.wikipedia.org/wiki/Netflix_Prize and also the book by Scott E. Page (Ref. 245).
187
All of the above implies the need of a “New Economic Thinking” (“Economics
2.0”), which enables us to organize our economy more efficiently and effectively
18
(see the next chapter and Appendix 9.2). I strongly believe that we are heading
towards a new kind of economy, not just because the current economy no longer
provides enough jobs in many areas of the world, but also because information
systems and social media are changing our interactions and opening up entirely new
opportunities. In particular, to cope with the increasing level of complexity of our
world, we need to catalyze collective intelligence. In fact, the principle of “networked
minds” makes it possible to bring the best ideas and skills together, and to leverage
the hugely diverse range of knowledge and expertise in our societies. Let us discuss
this now in more detail.
9.11 The Netflix Challenge
One of the most stunning examples of collective intelligence is the outcome of the
Netflix challenge.
19 The video streaming company Netflix was trying to predict what
movies their users would like to see most, based on their previous movie ratings, but
their predictions were frustratingly bad. So back in 2006, Netflix offered a prize of
$1 million dollars to any team that could improve their own predictions of individual
movie ratings by more than 10%. About 2000 teams participated in the challenge
and sent in 13,000 predictions. The sample data contained more than 100 million
ratings of almost 20,000 movies on a five-star scale, made by approximately 500,000
users. Netflix’ own algorithm produced an average error of about 1 star, but it took
three years to improve it by more than 10%.
In the end, the prize was won by a team going by the name of “BellKor’s Pragmatic
Chaos”. In the process, a number of really remarkable lessons were learned. First,
given that it was very difficult and time-consuming to improve the standard method
by just 10%, Big Data analytics wasn’t good at predicting people’s preferences and
behavior. Second, even a minor improvement of the algorithm by just 1% created a
significant difference in the top-10 movies predicted for each user. In other words,
the results were very sensitive to the method used (rather than stable). Third, no
single team was able to achieve a 10% improvement alone.
9.12 Diversity Wins, not the Best
The final breakthrough was made when the best-performing team decided to average
over their own predictions and those made by other teams that weren’t as good.
Surprisingly, collaborating with inferior teams boosted the prediction performance,
18 Helbing [16].
19 See https://en.wikipedia.org/wiki/Netflix_Prize and also the book by Scott E. Page (Ref. 245).