We expect a lot from our information systems, but even when they do their job exceedingly well they usually let us down in some way. Often we blame ourselves. Maybe we need to change a setting somewhere or maybe we just misunderstand how a feature works, it is not necessarily something inherently wrong with the program.

However, the problem is with information itself. It is both a format and a timing issue, information exists in a way similar to quantum superposition. It is there when it is observed, but after that it changes. By its nature, information is not digital. Information defies language too. These are all tools that we have developed to capture, store and share information. Converting information into digital format makes it easier to contain.

That is a seemingly great compromise for many things, but much to our chagrin information doesn’t really know any boundaries. Information leaks in every sense of that word. Confidential information leaks out. Information is lost in translation

Also lost in all the furor over misinformation and disinformation the past few years, is the fact that information is only true for a brief moment in time. All the context and atmospheric elements at that time only existed at that moment. Maybe History repeats itself, but information never does (and I am not talking about duplicate entries in a database).

In our data centric view of the world, we are driven to collect as much data as possible and let the data guide our decisions. Following the evidence of observed data seems to be perfectly rational, but it actually leaves out an awful lot of important detail. A lot of information is analog and does not convert or get captured in any digital repository. One of the ironies of the AI, is that it is great at seeing patterns in collected data, but the collected data is just part of the story.

We can always try to improve the output by improving and increasing the input, but we are only capturing a small amount of the data at any given time and we are still using primitive tools to build our various data models.