Thursday, February 4, 2010

Secure natural communications - how to proceed

After my initial posting last week, the most frequent question that I received was: how do I plan to proceed with this project? Is this something that we will build tomorrow? Is there some spec? Are these some wandering ideas?

Most of these questions came through email from my colleagues. This blog should be an active discussion community so please post your questions and comments to the blog – as several people have already done.

Reprise – and what is possible today

To reprise, the world should provide an infrastructure where: people can have a rich communications experience at will.

From a core technology perspective, almost all technologies required exist today in elemental form. There are numerous sensors that exist: often called by names such as microphones, cameras, displays, etc. These sensors can detect speech, movement, typing or provide images and videos. Properly configured, they provide the elements of natural human communications overcoming individual device conventions. With ample communications bandwidth we also overcome distance constraints. An infrastructure could both sense bits and interpret them – although not with perfection.

From this - in principle – one can take people’s utterances and build a management system that acts upon them. These can be primitive utterances that request basic communications. Or coded machine commands that request the performance of sophisticated compute function, rich applications, moving function to the cloud, etc.

What is not possible today

There is no infrastructure that provides secure natural communications. Not across the globe; not in a country; not in a city – not even on a campus. It would be considerably expensive to build and there is no known economic model to pay for this new infrastructure.

Goal and method

Here is my pitch. If it is possible to create such an infrastructure – and it would mimic how people naturally communicate – isn’t that compelling enough?

Hence my quest. I want to use a series of postings to express my views about the components of such an infrastructure. I need experts to add precision to this with their knowledge and views. After a while, we will develop enough clarity of viewpoint that we can prototype it on a small campus. If people like it – we will grow it from there.

Next several postings

So my first few postings will continue to discuss the high level concept of secure natural communications. I want to engage in discussion on the general idea. After that, I will elaborate in detail on the five constructs that I outlined last week. Provide a more complete description. What technologies exist? What more do we need to develop over time? And in discussion I will continue to need experts who have the domain knowledge to add precision.


  1. Can you talk more about the "management system"?

    I have two very different ideas about the possible management systems:

    1) Management System is embedded into the device used for communication. Based on the primitive commands it processes the request. (For example: I give a voice command to contact some person in my phonebook). But, with the proposed management system, the domain is not limited to the my phone. The management system will be intelligent enough to connect to Internet or any other means of communication wherever, the person is present and will transmit the data based on where the person is present. Example: A person is logged onto say gtalk, the management system will use the channel meant for Internet to contact this person. So, bulk of the processing will be done at the device end.
    2) The other possible management system would be the one at the Service Provider end. Here, we simply need to use whichever device we are using and simply say contact XYZ. Its the job of management system at SP end to figure out where the other person is and what is the most cost effective way to contact him. Example: When a user is present on gtalk and also using the phone, I can voice chat with him (cost effective)or call him up.

    After, your first post I had idea 2 in my mind. But, when you mention sensors and stuff I am confused :(

  2. Chiku,

    Fair point and it is worthwhile to clarify.

    Your logic already came to the correct conclusion that the management system needs to be in the infrastructure (Service Provider in your terminology).

    Parts of the function might be more efficiently handled by the local sensors - so having a distributed nature - with some delegated to the end devices should be part of an overall cohesive management system.


  3. Thanks for the clarification :)