The Problem with Automation
Updated: Nov 10, 2020
Jeff Schuyler Wednesday October 21, 2020
The automated monitoring industry has grown by leaps and bounds since I first started in the business. Believe it or not, when I first started building systems in the late 90s and exploring internet based data display options for the Civil Engineering industry for construction monitoring, dam monitoring and bridge monitoring, I was swimming upstream trying to justify the approach. I even wrote and published a paper about "The Benefits of Automated Monitoring" (converted to a blog) as an argument for using it to augment human-based observational techniques of monitoring. The engineering consulting industry wasn't really buying it at that time - not in a big way. Part of this reticence was because of the industry business model, part due to contractors wanting to limit their liability, and part because "eyes-on" monitoring was being counseled by industry leaders like Ralph Peck and Karl Terzhagi (another blog).
I couldn't possibly imagine the changes that would take place with the explosion of technology. Now everything is about automated monitoring and automation and there are companies galore selling their approach in every industry. Many instrument manufacturers have jumped into the fray by offering some kind of cloud-based data service as part of their product offering. It's "new", it's "sexy" and it makes everything easy. That's the marketing pitch anyway. So I find myself in a different landscape possibly fighting a new battle to NOT let automated monitoring and automation take over and replace human observation and intuition.
Human Intuition is Worth Saving
Back before we could measure almost anything electronically we had to 'measure' things crudely or indirectly. We measured water level with a stick and flow with a bucket. We irrigated based on experience and minimizing "standing water". Sometimes when faced with a situation without any way to measure we had to 'feel' our way to a solution. I can remember a call from a customer that was standing in front of a control cabinet in the middle of a dark, stormy night. The system he was standing in front of was a combined security and batch loading system for tankering water - very complex with RFID tags and user ID tokens. It had malfunctioned with tankers on the pad and when he called I had a sinking feeling in the pit of my stomach that there was no way I would be able to solve the problem. He had no technical knowledge about the system and I dreaded trying to convey knowledge to him over a sketchy phone line in the middle of the night. The first thing that popped in my mind - while I gathered my wits - was to ask him to open the cabinet and tell me what he saw. He stood back and looked at the maze of wires and terminal blocks and flashing lights and saw something that looked out of place - a breaker that was not like or with the other groups of breakers in the cabinet. He had found the problem right out of the gate and we were up and running in a matter of minutes. OK - so it's only a breaker. But many people in the same situation would not have the presence of mind to see and feel that something was out of place.
Even me with my wiring diagrams would have faced a significant challenge to translate the symptoms he told me about (if he could describe them) to the appropriate circuit to identify the offending circuit breaker. He wasn't technical, but he was saavy in a way that comes with direct experience with making things work - in the middle of the night with production hanging in the balance. This is a combination of skill, experience and human intuition all working together. The human mind is great at taking in a lot of information simultaneously that it may not be able to consciously analyze or understand - and making a quick decision that is correct. A simple example of this is finding the mid-point of a line or center of a circle without measuring. I've seen people struggle for a long time to measure something in difficult situation when they could probably put their finger on the point they were trying to find. But they lacked the experience and thus the confidence to be able to make the call "intuitively". This is what we risk losing if we rely too heavily on instrumentation and associated algorithms to measure and calculate for us.
Automated Monitoring as Another Human Sense
When we started to connect wires to sensors and collect data at regular intervals without having to take a manual measurement it was a revelation. The reason is that regular periodic measurements remove the 'alias' associated with the process of a human having to take the measurement which they can never do as regularly or as infallibly. The device now becomes akin to an electrocardiogram for the heart showing all it's native rhythm and irregularities. Hopefully it shows the operator what they want or need to see, and provides crucial information to a computer that might be making decisions - but it also shows things that we didn't necessarily anticipate. A person experienced in evaluating the signal can see when there is something wrong even though they not immediately know what is wrong. And the signal or time series evaluated over longer and longer time periods (as opposed to instantaneously) conveys new information about the behavior of the thing being measured and the environment within which it lives - for those that are looking. In this sense data from long-term unbiased monitoring becomes a window into how things work, how they move, how they fail and how they are interacting with the world around them.
In this context the monitoring data enhances or complements the other things we can see and hear. It either reaffirms or contradicts what we think we know about a system or environment. Sometimes by adding a wired sensor as a secondary or backup form of measurement we can find out that a historic measurement technique we have been using for decades - is wrong. Sometimes it is not easy to convince an operator that their "tried and true" technique is wrong. But in the end the automated measurement - if deployed and configured correctly - will prove itself to be correct just because of its unbiased "professional" way of reporting the truth.
Integrating Humans and Technology
There is no way to turn back the tide of automation. It's part of human evolution now and there is much about it that is good and necessary. But just because we can do it doesn't mean we should flip the switch and automate everything. A modern passenger plane manufacturer might argue that there are too many systems and the planes are too complicated to turn over to "flying by stick". But a pilot could argue that by removing the human element and letting computers fly you take the "feel" of flying away.
Granted this is an extreme example. But perhaps a simpler passenger plane related example can illustrate the point. On a recent 5 hour flight the entertainment system on the plane didn't work which resulted in the screens mounted in the back of every seat blinking on and off for the entire flight. I could see that the problem was a power supply related issue as the system kept trying to start up and boot into the operating system, but would go dark at random times and start over. The whole main cabin was like a Christmas tree of blinking lights - for the entire 5 hour flight. Here's the point - nobody on the plane either knew how or was capable of turning the system OFF. They had to wait for the services of a specialized technician who could understand the system. This is a simple solution turned into a complicated process that is not well understood by average people. This is not where we want to go with automation.
The point is we need to find a way to use automated monitoring to educate, to enhance the "manual" method, and to manage complex processes all at the same time - before flipping the switch to "AUTO". A classically trained operator knows more about the process they are involved with than the machine that will replace them. Why not give the operator the choice about when and if to flip the switch to full automation? We don't want to raise our children to only stare at a screen with green, yellow and red indicators and push buttons in response. This scenario has been presented to me as necessary for water treatment plant operators for example, because there aren't enough young operators with the prerequisite experience coming up. The experience they are lacking is that experience gained without automation or automated monitoring. They don't know how things work, or how they sound or how they smell (or where the center is) because they never faced a situation without an electronic device telling them what to do. To my mind this is reverse human evolution because it stunts the very character traits that built our society.
I hope to add examples of how we can do this in future blogs.