Self-Aware e-Society

In Uncategorized on July 9, 2006 at 9:20 am

(this post was refreshed on Jul 16, ‘08.)

A Self-Aware Society

In this post we discuss the idea of a pattern-recognizing neural network that sits on top of a P2P network and learns to recognize and predict communication, social, cultural, political and transactional patterns [generated by the users] across the system. The idea is to enable the detection of the emergence of potentially “negative” patterns in the system (such as speculative market bubbles or the emergence of certain group behavior) thus allowing us to control or at least predict social, political and business trends.

The idea is to use the P2P clients in such e-society as a way to pull into the neural network the social, political, and business (transactional) trends produced by the users across the network. In this scenario, the neural network, which would be separate from the actual P2P layer itself, would be trained to recognize certain patterns in the real-time data gathered from the P2P clients and alert us when certain patterns are detected.

Obviously, there are limits on the types of patterns that can be isolated and learned as well as limits on the accuracy of pattern recognition and trend prediction.

However, the potential for such a pattern-recognition layer would be immense (and scary.)

Self-Aware e-Society vs Prediction Markets

Prediction markets are mostly based on wisdom of crowds. They are simulations in which people make individual judgments and their judgments are averaged to produce the prediction (or the crowd judgment). There are types of prediction markets where people buy and sell and the system makes the prediction (or crowd judgment based on the buy-sell decisions which represent judgments)

However, I am not aware of any prediction market that can recognize and predict emergence of patterns in people’s implied or explicit judgments as they relate to a given company stock, product, idea or person. These patterns are extracted from the users’ communication, social, cultural, game-playing and transactional data (including inferred data) which are captured from virtual stock markets, virtual auctions, chat rooms (where a hierarchy can exist: e.g. founder of room, operators, favored participants, participants, and unwanted participants), social applications and entertainment applications (including multi-player online games.)

People supply complex individual implicit or explicit judgments in true-to-life simulations that generate patterns of judgments across society or across groups within the society which can then be taught and recognized (for those patterns that relate to a phenomenon like speculative market bubbles, emergence of cults, etc) by the neural network monitoring this live e-society.

Governments and politicians will be able to use such live (made of people), self-aware e-society to simulate the outcome of critical political decisions on society before they make those decisions in their own, real society.

This relates to governance in another way: the e-society by being aware of negative patterns emerging within it can flag and alert the leaders of the e-society so that they may try steer society away from trouble.

I believe it is the next level in prediction markets. The key difference with respect to prediction markets, is that a self-aware e-society will be able to capture, recognize and predict the emergence of behavioral patterns that happen within it as opposed to simply predicting single-valued outcomes and ranges (without the ability to recognize and predict the patterns that could lead to those outcomes.) In other words, a self-aware e-society can predict the outcome of prediction markets running within it before prediction markets can make that prediction. That means that (given the ability to pre-predict and thus potentially avoid bad outcomes) the prediction markets can be real markets and not just simulations. So it would seem that the e-society application described here could run on top of society itself (i.e. no need for simulation.)

In other words, a self-aware e-society would act as a predictive governance tool for society itself.


I realize that it does sound very futuristic, but the idea is ‘technically’ compatible with the democratic governance ideals I had proposed for Web 2.0. In other words, in the ideal usage scenario, it should not supplant them. It should help society by monitoring it for dangerous trends so that the problems that would normally happen could be diffused.

Think it’s sci-fi? It can be put together with existing technologies and expertise.


The implications of this idea extend to areas such as national security, economic security, cultural phenomenon, political science, mass psychology and sociology.

But is it good or bad?

Any idea that can deliver a lot of power to someone and is realistic enough to be attempted will inevitably be developed by someone somewhere. So it’s better to put these ideas (be them good like the Wikipedia 3.0/Web 3.0 idea or potentially concern-causing like the Tagging idea or this idea) out there in the open and let people be aware of them and debate them.

Response to Readers’ Comments

Ian Delaney wrote: I wonder if machines are up to the job of identifying negative cults. After all, human judges seem to make a lot of very bad mistakes.

The leaders of society will still be the ones who would make the judgment. The machine is a predictive tool to help society manage emergent patterns. It is still the people who make the judgments, through their democratically elected leaders. The machine simply provides a cognitive layer below that.


  1. Open Source Your Mind
  2. Tagging People in the Real World


  1. Soulenoid (Scream at the right time)




Trends, wisdom of crowds, Startup, mass psychology, cult psychology, Web 2.0, Web 2.0, democracy, P2P, P2P 2.0, social networking, Web 2.5, governance, Internet governance, pattern recognition, non-linear feedback loop, neural network, prediction markets, e-society, national security, economy, political science, cultural phenomenon, AOL, NSA, wiretapping, civil liberties

  1. […] I have recently discovered a few very interesting blogs. They are Marc Fawzi’s Evolving Trends and twopointouch from Ian Delaney. Really nice and complimentary, if you are interested in AI research and web 2.0 theories, go ahead and indulge. Marc’s latest posts dwelves into microcrowds and Ian’s writes about social news. […]

  2. Looks like my sleeping in until 1:20 wasn’t such a bad move after all. Much clearer now than it was 12 hours ago!

    Would the inputs not be simply information, making the microcrowds processing engines as part of a greater analytical whole?

  3. Well, the article has been reblogged to describe a workable scheme for an e-society that can detect the emergence of negative patterns (problems) within it, such as the emergence of cults or speculative market bubbles.


  4. Interesting stuff, Marc. I wonder, though, if machines are up to the job of identifying negative cults. After all, human judges seem to make a lot of very bad mistakes.

  5. Hi Ian,

    People actually make that judgment through their behavior and/or explicity. The machine simply captures the behavioral/judgment patterns generated by people across time and demographic space.

    So the people are the ones who do the judgment. The machine is a predictive tool to help society avoid the emergence of negative patterns (e.g. emergence of spectulative market bubbles or emergence of cults, hype, etc) It is the people who make the judgments. The machine provides a cognitive layer above that, on the level of society as a whole.


  6. […] For a related idea, please see P2P + AI: World Peace or Disaster? […]

  7. The problem with these sorts of things is who controls it and has access to the data. What you might classify as a cult another might classify as a religion (falun gong), what one group classifies as terrorist support another might classify as public aid (red crescent). This would give
    very powerful tools to someone, and I would want some checks and
    balances to ensure it’s used in as benign a way as possible.

  8. I was looking at a project to use AI to find out the impact in court sentencing. Example: for different crimes/sentences, what are the possibility that the accused will commit similar crimes upon release. So what should be the preferred sentences for a set of conditions?

    Any idea where to start?

  9. It sounds like you should look into Baysian methods (Probability) and may be look for “Bayesian +AI” in Google.

  10. […] Self-Aware e-Society ( link: screaming at the right time may save your life.) […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: