Let me just remind you, Gordon was the one behind the great project of Digital Memories, the “MyLifeBits” project. It is an actual “Memex” project brought to life with Microsoft Research and Gordon Bell as the soul and main author and driver of the project.
The actual reply here is to a recent CNN Interview with Gordon Bell regarding his new book, “Total Recall”, that summarizes the results of experiment Gordon Bell took part in for last few years, the MyLifeBits software project, and all the related research articles, experiments etc. done in this area.
Particular, the question that I would love to comment on is the following one:
CNN: Are you worried about losing your memory?
… Forgetting is not a feature, it’s a flaw. I don’t think forgetting is an important feature of human memory. I think it’s important to be able to remember things accurately.
In this reply, Gordon Bell told about opportunity to remember things accurately. For me, it is a slightly important topic that needs to be fully addressed in a more formal way like an article but for the purpose of this writing and with regard to readers of this blog here are few ideas that briefly comment on this suggestion.
I. In Physics, there is understanding that every process can be described, modeled precisely using Math and Physics formulas. However, it is also known that to be totally accurate everything should be known totally precisely. The opportunity to describe everything precisely in Physics comes down to quantum mechanics and here is a moment that there is an Uncertainty principle postulated by Heisenberg. To be short, the point is you’ll never know either precise position of electron or its precise momentum. Thus being showing something precisely, or, as Gordon Bell names “accurately” is technically impossible. This leads to question, if it is impossible to record some processes (and our life that we record with computer power is about series of different processes, after all) precisely, what options do we have?
II. My answer to this question comes from the so-called “General Semantics” approach, first introduced by Alfred Korzybski in his “Science and Sanity, an Introduction to Non-Aristotelian Systems and General Semantics” in 1948. It is that it is impossible to record things accurately. It is also that what we record always will be based on our opinion of the facts we record. And if use computers then we just build them with our opinions in minds. Thus what is important that a real digital memories system will have it’s own opinion that will be based on opinion of system’s author, or, if system will be intelligent enough, on opinion of its owner.
This answer is based on the following principles of General Semantics (let’s name process an event in this particular discussion):
- The map is not the territory itself – which means that any particular event is not equal to its representation, where territory is an event and map is representation
- Every particular event can have a human representation of it being objectified; but there are different object representations given by different parties;
- Every particular event has n of characteristics, or properties representing it where ; thus each event can have m of objects representing it where and each object highlights different sets of event properties;
- Every object representing a particular event is an abstraction of this event given by a specific point of view or need because of which some particular event properties are highlighted in this object;
- Every object can have its own object representation; because amount k of event properties showcased in particular object is countable but number n of event properties is uncountable; ; this leads to fact that amount of all possible abstractions of event uncountable.
One of the fundamental principles of people intentions is that if there are several opinions on something the most wise thing we can do is to find common things in these opinions in order to precisely describe that exact something. Thus the most interesting question for me is that actually, a recording system should be able to:
I. Be able to record situation of user from different perspectives/opinions
II. Be able to interact with other digital memories systems used by other people in order to build a better/more correct combined image of the situation
And here comes, that a simple recording system turns into a much more complex context-aware system.
The question comes how based on these ideas we could build a better digital memories system?