Data in Motion
The Feedback Canon — Installment #12
It has been a few weeks since the last installment. Rest assured that #11c will arrive. It has been tricky. In the meantime, more concepts were developing. So it is time to put some of them to writing.
Our last few articles have talked to feedback moving Beyond Data and Being Incremental. We focused in on this particular description:
Feedback builds and suppresses. It flows and connects. It is highly dimensional. It might be fair to call most data scalar while feedback is a vector, but that is really an over simplification.
But now it is time to further synthesize it.
Feedback is data in motion.
From here, one can easily drop down the rabbit hole into information theory. Information becomes analogous in many ways with energy. It has entropy and redundancy. Claude Shannon and others would focus on transmitters, recievers, code, signals, and noise. We would discuss communication systems. And well, perhaps we should. But that is the physics of information… and while informative science, perhaps a step too deep?
Information theory focuses more on the transmission of information. It is worried about speed, capacity, and quantification. We will continue to take that aspect for granted and worry more about the implications.
But motion is only one dimension of feedback.
And so we may ask ourselves, is it really practical or appropriate to store multidimensional moving data in a static, two dimensional structure? Static and two-dimensional aptly describe most database structures (tables or views). While we might be inclined to consider a full referential database structure, remember feedback is connected, too. It would require a highly complex warehouse to capture even a simple feedback system.
This belies the challenge of most data. It has to be simplified to accommodate the storage available. This isn’t so much a technology issue as a conceptual one. Referring again to information theory, technology would support the endless strings of zeros and ones needed to transmit that data. But when the data is simplified for storage and retrieval, information is lost.
Some of you may believe I am taking this too far. It is time for an analogy. How often are you fooled by your TV or even a VR device? While 8K UHD has arrived (well, not the content) and is the first to end most pixelation — it still does not approach real life. And that is just resolution, images are still measured in frames per second — real life is continuous. And then there is the flat (well now curved) screen component — which is at least partially accounted for by VR devices. And of course this is just visual data… life has multi-modality. Smell, taste, touch, sound…
Data is a vast simplification in general. If you are attempting to describe actual feedback, doubly so…
Some of you still think I am taking this too far. You are thinking that humans only utilize a fraction of what they actually view at any given moment. That our brains group, cluster, and simplify. While this is true, you will also need to admit that scientists aren’t exactly sure which fraction, or why. And this is all just theory, based on known processes. There are plenty of less known or even unknown ones.
If we fall into the trap of modeling poorly understood feedback systems with poorly defined models, well this is a recipe for nothing good. It is abstraction and that is just another word for simplification. A different one with slightly different context… but that may be too far?
And now a word on our robot overlords…
You might be forgiven at this point for wondering about all that miraculous AI that is coming to enslave us. If humans are incapable of digitizing anything without losing resolution and a host of other massive simplifications, how does any machine stand a chance against the real world? Everything they deal with is framed, filtered, and sampled. Humans may not understand how they deal with the endless stream of feedback that life presents us — but how are machines going to “out think” us when their world is “dumbed down” by both default and necessity.
Artificial Intelligence that is capable of competing with or exceeding human capability will not be built on data as it exists today.
So where does this leave us? We need to continue defining and exploring feedback with the goal of building a model or framework that better captures the multidimensional, contextual, connected, and (yes) moving nature of feedback. It will be a simplification, but one that better reflects the nature of the inputs. One that captures these unique qualities.
Stay tuned for our next installment. Coming soon…
