The Technology Review posted the other day an article on the “10 Emerging Technologies of 2008“. Very promising technologies indeed, but the one that really caught my attention was the Probabilistic Chips currently studied by Krishna Palem.

The reason I find this particular research so interesting is, that

• it has the potential of extending current scaling laws - and particularly Moore’s Law
• it takes advantage of the principles of collective intelligence and the wisdom of crowds.

The theory of Probabilistic Chips

According to the article

Palem has developed a way for chips to use significantly less power in exchange for a small loss of precision. [...] chips could be designed to produce the correct answer sometimes, but only come close the rest of the time. Because the errors would be small, so would their effects: in essence, Palem believes that in computing, close enough is often good enough.

Current Scaling Laws

From the beginning of the twentieth century until today, a series of laws has emerged concerning the growth of computers and networks. They all negotiate the value of a network according to the technological leverage of the time.

• Sarnoff’s law, which was interested in the growth of radio and television networks (the value of the network is proportional to the number of actors)
• Metcalfe’s Law describing the value and the growth of small scale networks (the “value” or “power” of a network increases in proportion to the square of the number of nodes on the network).
• Reed’s Law describing the value of Group Forming Networks (the value of networks, that support the construction of communicating groups create value that scales exponentially with network size).
• Moore’s law handling the evolution and expansion of computer microchips (the number of transistors on a chip will double about every two years).

One can notice the evolution from Sarnoff’s Law to Reed’s Law.

While Sarnoff’s Law was suggesting, that the value of a broadcast station (television or radio) would increase proportionally to its audience it could not be applied to more complex networks, since the degree of interconnectivity was much higher.

So Metcalfe’s Law was an evolutionary step of Sarnoff’s Law, which better described the upcoming computer networks (ARPANET) in the 1960s. But Metcalfe’s Law could be easily be applied to small computer or telephone networks, but certainly not to huge networks like the Internet.

Reed’s Law emerged to describe the masive development of the web and more specifically of the social web.

Moore’s Law, on the other hand remains stable. Although it finds application on the expansion of computer microchip technology for more than 40 years, its validity is lately at stake. The reason is, that silicon transistors are becoming smaller and smaller and therefore less reliable.

But that’s where the probabilistic chips come in play, to keep the microprocessor technology rapidly evolving.

With probabilistic chips, tiny microprocessors may be designed in such way, that the individual parts might be imperfect, but collectively bring perfect results.

Collective Intelligence

The probabilistic chip technology actually takes advantage of the theory of collective intelligence.

As supported by the Condorcet Jury Theorem [pdf], the probability of a correct answer by a majority of the group increases toward 100% as the size of the group increases. The validity of the Theorem is based on the hypothesis, that the answers provided by each individual are not random - but instead there is a more than 50% probability to be correct.

The Condorcet Jury Theorem might be criticized when we are dealing with humans, but machines can be programmed to work in such a manner.

It seems to me, that this is how cultural revolutions emerge: combining disciplines, which at first seem irrelevant to bring forth innovative ideas and technologies.

Probabilistic chips, can change the scenery of energy consumption, mobile technologies and microprocessor development; and all that by applying an almost 200 old theory to a completely different research field.

A post in the ReadWriteWeb a couple of days ago, guided me to a very interesting document. A summary of Project10X’s Semantic Wave 2008 Report (available here).
I just finished reading the report, which provides some very insightful information about web 3.0. The semantic web will transform the web from an information-centric to a knowledge-centric system, by developing 4 fundamental attributes:

1. Knowledge

The web is a fragmented place. Knowledge is scattered in all its corners, sometimes locked in operating systems and complex algorithms. The semantic web, will pursue to change this. It will facilitate technologies, which will extract knowledge and

will enable communities to create, curate, and share knowledge in human readable and machine executable forms.

2. Transparency

Information will evolve in knowledge, through its encoding in a semantic form, which will be transparent and accessible at any time to any machine. Knowledge was previously stored either in human readable or in machine readable form. In the semantic web, it will be stored transparently, so that users and machines will be able to read the same piece of data. In that way, it will be possible for data to be used, validated and combined with other data. This will allow

a system to “learn” to do things that the system designer did not anticipate.

3. Connectivity

To overcome the limitations and restrictions of OS platforms, the semantic web will encourage a real time usage of automated and semi-automated methods, of interaction between man and machine:

Web-tops; platforms spanning multiple OSs connected over the internet
Mash-ups; two or more data sources or works combined to become a new data source or work
Context-aware mobility; dynamic composition and personalization of services across devices, networks, locations, and user circumstances and
Semantic service oriented architectures; using machine-interpretable descriptions of policies and services o automate discovery, negotiation, adaptation, composition invocation, and monitoring of web services.

4. Technology

The key of the sematic web is the usage of technologies, which represent meanings and knowledge seperately from content, in order to be interpretable from humans and machines. Such representations will range from pattern recognition, analogy and reasoning with uncertains to deep linguistics and causality.

The integration of social Web and semantic technologies in Web 3.0 allows new synergy that lowers the cost of data and knowledge creation, and raises the computational value of gathering.

The semantic technologies, which will power Web 3.0 will concentrate on:

• Semantic user experience (how the user comprehends things)

• Semantic social computing (how users communicate and collaborate)

• Semantic applications and things (how products and behaviors can be seen empirically and objectively)

• Semantic infrastructure (interobjective network-centric systems and ecosystems)

• Semantic development (how meanings and systems can share what they know)

Epilogue

The report refers also to semantic technology markets and other interesting points. It explains the the information I shortly mentioned above very well and I definetely suggest you to read it.

The 4 attributes I listed above gave me the impression, that they are the key traits, which will revolutionize the online experience. Where, the emergent behavior of the whole system will bring user interaction in new levels. I believe, that the development of services such as data portability and openID are steps to this direction. But, to a certain degree, it is a personal preference.

I’m interested to see, which attributes of the semantic web do you find most revolutionary.

I had a post the other day about Facebook and one of its rivals “Kaioo” and the discussion came to the subject of data portability.
M@ri@nn@ expressed her skepticism on the project. This short video (found at Partcls.blog) sumarizes the concept of data portability very well. Enjoy.

Sorry, I had to remove the video because it really messed up with the look of the blog…

Get involved in data portability.

True communication had always language as a barrier. Any discourse above the fundamental, pantomimic level requires a common language. Although the worldwide web has brought us all much closer tearing down any sense of space, this basic obstacle in the exchange of information still remains.

Sure there is a multitude of ‘bots’, which can translate any web content, but lets face it: “Automated computer translation is decades away from producing acceptable content for most forms of communication.”

In the steps of Wikipedia, the Worldwide Lexicon calls on the collective knowledge to bring online communication in new realms. The translation of web content is in the hands of the users.

Now the project brings a new plugin, which makes translation much easier. It justs adds a small pencil in the page and by clicking on it you can start translating the site’s content.

Wisdom of crowds

The potential of abuse is quite obvious. By installing this plugin (either in your site or blog), you give everyone the opportunity to translate your content in any way he/she wishes. Still, the worldwidelexicon relies on the wisdom of crowds to aggregate information correctly. Anyone comfortable with two or more languages can choose any web page and translate its content. Meanwhile other users are also free to rate your translation and even redo it or edit it. An emergent microbehavior where many agents work to develop a macrobehavior develops.

Conclusion

Trusting the collective intelligence is, in a sense, an objective issue. It depends on the personal experiences, views and morals of each one of use. Some believe in collaborative projects like wikipedia, where no central control guides the development process. Others are looking forward to projects like the Google Knol, where they see a trustworthy source, capable of collecting high-quality data. But I’m asking you: do you believe in the wisdom of crowds?

Additional resources on worldwide lexicon;

VentureBeat

O’ReillyNet

In a previous post about aggregation of information I noted the role of the administrator, as someone who will moderate the discourse. Teut Weidemann in his interview remarked that “…total control will destroy a community pretty fast”. So indirect control actually means letting the community takes its own course and interfere only when things really go out of hand.

Learning from software

But indirect control has one more implication. Consider the StarLogo software. It is a software which demonstrates the development of bottom-up systems but also obeys to a centralized control; namely the commands of the programmer.

But the programmer does not define what will StarLogo generate. Instead he/she gives the micromotives (actions) and lets the software evolve a macrobehavior of its own. The programmer does not control the process any more.

Do you remember Gearheads, the computer game with the windup toys on a chess board? It uses the same principle. You released the toys and then it was up to them how the game would develop (sorry I couldn’t find a proper link to the game; help would be appreciated).

To summarize

So what do such software tell us? That indirect control - apart from the obvious role already mentioned - is about providing tools to the community. Tools, which will give the group the necessary input in order to nourish a healthy discussion. Now its up to you to decide the form these tools can take.

It may be questions, polls or surveys. It may be the way you moderate comments (or comment on others - they are part of your network too). Let your imagination go wild. ‘Navigating’ your community correctly, giving the right input, will draw the attention of a diverse and independent group; and in the end of the process (although you can suggest that this process sometimes never ends) the subject of discussion will be examined under many perspectives and with great objectivity.

I noted earlier the paradox between a group of independent individuals and the social web: the social aspect of human interaction, sets influence between the group members as a standard. The structure of the network itself can assist in avoiding this paradox.

Decentralization and the collective intelligence

The notion of collective intelligence takes decentralization as a prerequisite. When the community consists of a diverse and independent set of people, who all work together on the same problem, then no central control is required. The community can evolve on its own, powered only by the initiatives of its contributors.

In parallel, self organization fosters specialization – specialization of interests, of attention, of labor. Each member of the group can concentrate on a specific subject, according to its interests and knowledge. Apart from the fact, that specialization increases efficiency and productivity it also encourages diversity of opinions. Dividing a subject in subcategories and appointing users to work on them, will cultivate concentration and increment of knowledge.

The strength of self organization

The strength of self organization is that it allows people to be close to the actual problem and coordinate their activities, while each one concentrates on a special topic. But remember: the purpose of the community is to aggregate information; to take the specialized, local knowledge and make them collectively and globally useful.

The emergence of self organization

Just like Google’s search engine collects local information of millions of websites to make the search quicker or like the linux operating system collects user submitted information and bugs for improvement, so a community needs an administrator to take over this responsibility.

In both Google’s and Linux’s cases – and actually in every emergent system – we are talking about a network, which has the ability to adapt to new data. In typical emergent systems, this happens with a higher-level intelligence, which is aggregating the local knowledge to adjust the global system to fresh information. It transforms the microbehavior of the agents to a macrobehavior. Perhaps, this view finds an application when we are dealing with software (Google’s search robots), but in social structures the term ‘higher intelligence’ sounds exaggerated.

In the social web, groups do not need someone who will make the final decision. Instead they require an administrator who will moderate the whole process, who will indirectly control the community.

Values and self-organization is an interesting entry about the subject. Also Alex Iskold wrote an article about the digg effect and its self-organized structure.

“In the 1950s the social psychologist Solomon Asch conducted a famous experiment that highlighted the fragility of the person in a mass society when he is confronted with the contrary opinion of a majority, and the tendency to conform even if this means to go against the person’s basic perceptions.” (Solomon Asch; Opinions and social pressure)

From diversity to independence

Either by willingness to respect the informational signals shared by group members of a particular belief or by fear of dislike, people usually do not speak out their opinion. The informational and social influence can have a great impact in the decision of an individual.

An attribute that directly derives from diversity, is independence. Especially, when we are dealing with a group of diverse degrees of expertise, independence is necessary for the less experienced to not hesitate in expressing his/her views or questions.

I don’t see independence as a a form of isolation, but instead as a relative freedom from the influence of others. Having independent individuals in a group, prohibits the correlation of mistakes people make and verifies that new information will come to light. The misjudgment of a person, regardless of her knowledge degree, will not change the collective thought. Apart from that, autonomous individuals will bring diverse perspectives to the group.

Naturally I’m not implying, that with independence comes rationality and impartiality, but surely, under the correct circumstances, irrationality will not influence the rest of the group.

Independence in emergent systems and social networks

An emergent system can easily develop independence, since the constituting agents are ignorant of the overall state of the network. They are carrying out a simple, repetitive task guided by a pattern recognition system. But accomplishing independence in a social network can be a very difficult task.

In a previous post about social representations, I noted how strongly the private and public sphere can change people’s perspective. It seems as if influence were inescapable.

People are primarily social beings; they are constantly trying to expand their knowledge and perception by interacting with each other. The social web is – as the term implies – based on this function. It is founded on the principle of interaction between individuals and not between computers. Still though, collective decisions are most likely to be good ones as long as they are made by people with diverse opinions reaching independent conclusions, relying mainly on private information.

The question that rationally follows, is how can a community overcome this paradox. Creating a network, which consists of diverse, independent individuals seems not adequate enough to produce correctly aggregated information.