Crowd Computing: going beyond the clouds?

Post date: Sep 28, 2011 6:47:47 AM

For many years now, we have seen a huge increase in the use of sophisticated systems that have enabled world-wide collaboration through our PCs and, more recently, through ubiquitous hand-held devices. The purposes are as numerous as they are varied: content sharing, whether through blogs or many well-known peer-to-peer (P2P) applications; collaborative computation, starting from the early SETI@home project (http://setiathome.berkeley.edu/ ) that was one of the first instances of large-scale grid computing. There are plenty other examples.

Content has changed also. It has moved from the classical relational databases kept by companies containing information about customers, providers, etc. Progressing to petabytes of data representing real social networks containing tastes, opinions, media and any other imaginable piece of information. Most importantly, people are gaining awareness of the power of collaborating through the network. We have recently seen riots, like those in the UK, or even national revolutions, like in Egypt, where the lack of centralized control over information played an important role. Technology is at the forefront of change.

The “crowd” is becoming aware of its power, and the next natural step is to enhance the tools and modalities for collaborative computing. Powerful devices, like smartphones and tablets, are able to carry out an impressive amount and diversity of computation. P2P computing has shown to be feasible and efficient. We have some examples such as Skype that show the model is valid and can challenge serious cloud-based competitors, such as Google Voice.

An interesting observation is that not only machines but also real people are “connected” to the network combining their computing and thinking capacities. Trends seem to be pointing to this model as gaining the position to complement (or perhaps substitute) cloud computing: connecting people and machines in a single network.

Nowadays, millions of people are asynchronously analyzing, synthesizing, providing opinions and labelling and transcribing data that can be automatically mined, indexed and even learned. And there is not much difference between this and classical computing: the “crowd” is working online, taking digital data as input and yielding digital data as output.

The main difference is that human brain-guided computation is able to perform tasks that computers struggle with, and do them at overwhelming speeds. Tagging a picture or a video based on their content or answering questions in natural language, are just a couple of examples. In other words, contradicting the classical fear based on the idea that machines will replace humans, the new paradigm is that machines collaborate with human. This permits overcoming the limitations of classical computing, by leveraging the potential that humans have to do some tasks naturally and more efficiently.

Srini Devadas, professor of electrical engineering and computer science at the MIT Computer Science and Artificial Intelligence Laboratory, believes that this so-called “crowd computing” will complement the cloud, offering a new infrastructure that will enable the world to become more "collectively intelligent". As an example, he mentions its potential use for mitigating the impact of natural disasters by collecting data and fostering collaboration from many people scattered geographically in order to plan evacuations, predict the impact of disasters, etc.

Crowd computing delivers the elasticity of cloud by leveraging peer-to-peer technologies. It also dispells concerns about loss of privacy, since a single cloud provider does not have a global view of anyone else’s data. It also presents a more economical solution compared to cloud computing: instead of paying a cloud provider for services, the contribution is made in-kind by becoming part of the computing system that offers computing power, storage capacity, data or knowledge. As a consequence, the concept of “cloud owner” is removed from the equation.

Some of the intriguing questions to be addressed include:

• Will we be able to crowdsource CPU hours in the future?
• Will the crowd carry sensors on their mobile devices to make the network more aware of environmental situations?
• Do new security questions arise?
• How should we deal with performance issues?
• How can we extract high-quality answers from data created by the crowd, which implies many small contributions from well-intentioned providers that may not be correct?

All in all, it is time for companies to start thinking about real-time and real-world crowd computing.