Steve Hilton - MachNation
Internet of Things (IoT) solutions give enterprises the ability to collect data for virtually all their assets, equipment, personnel and locations.
These data can include geo-location, climate information, movement/vibration, velocity, performance characteristics, medical vital statistics, video feeds and many others.
Devices as varied as sensors and machine to machine (M2M) computers stream the data--sometimes in real-time--to various applications and platforms that store, aggregate and analyze the data.
IoT data have value to enterprises. Capturing these data at a point in time allows enterprises to make decisions and undertake actions. Capturing these data over time allows enterprises to facilitate business process improvements, reduce errors and minimize business risks. Enterprise can use data to lower costs and increase revenues.
Automation is one of the hallmarks of IoT. Proponents of IoT analytics expound the virtues of their tools to analyze reams of data, find patterns, set-up algorithms and integrate with other systems to provide instantaneous feedback and direction to the devices being monitored. Real-time feedback lowers costs, reduces errors, increases the likelihood of compliance and offers myriad other benefits.
But enterprises are grappling with a fundamental question regarding these IoT data: are we comfortable excluding all human intervention from an IoT-enhanced business process and allowing machines to control end-point devices? Or are we incapable of giving up this control? Are we control freaks?
MachNation recently moderated a panel on the industrial Internet at IoT World in the U.S. On the panel were a mix of companies involved in the industrial Internet including platform and application companies Eigen Innovations, Evothings, Microsoft, PLAT.ONE, Prismtech and SeeControl; equipment manufacturer ARM; and communications operator Elisa.
We discussed at length the issue of automated control of equipment in the industrial setting--for example, controlling factory floor equipment, pipelines, heavy equipment in the field, and so on--without human involvement in the interpretation of IoT sensor-based data. We debated whether we as an industry were comfortable collecting all these IoT data from end-point devices, analyzing them against a set of algorithms and then using the results of the analyses to automatically control the equipment or things being monitored. We debated whether we were technologically advanced enough in computing, networking and analytics to replace the processing, interpretation and analytics skills of the human brain with machines.
After the event MachNation staff spent quite a bit of time thinking about this particular issue. There are several interesting angles to this question.
The technology--Most believe that we have the required processing, networking, application and analytics technologies to automate most processes and control most devices without human intervention. We can credibly create algorithms for most repetitive tasks and set-up alerts, alarms and automated responses for virtually any situation. Analytics tools are incredibly sophisticated and address the needs of enterprises as well as public sector organizations and academia. That being said, not all IoT implementations make use of the most advanced analytics, the most powerful computing fabric, the fastest networking and the most efficient applications. So while it is possible to deploy a completely automated IoT solution, often it is not financially practical.
Enterprise value--Would the deployment of fully automated IoT solutions yield significant value to enterprises? The removal of human latency and human error from manufacturing and business processes creates enterprise value. Latency and errors impute costs--both actual costs and potential costs from increases in risks--on enterprises. It is especially true that humans are prone to error when expected to complete repetitive tasks. Fortunately, machines do not get bored and can perform the same process almost an unlimited number of times. So it is conceivable that enterprises would find value in removing humans from the mind-numbing process of viewing sensor data, reviewing outputs and making adjustments to equipment as needed.
Automation--Humans have control issues. Contrary to any evidence presented to us, we still believe that our ability to process information and make correct decisions over long periods of time exceeds that of a machine. If we make a mistake, at least we can point to ourselves and say, "It was my fault." If the machine makes a mistake we somehow feel doubly at fault. First, we made a mistake because the machine failed. And second, we made a mistake in trusting the machine in the first place. In the industrial sector, workers are hesitant about relinquishing control of their equipment to an automated process. The lack of control is frightening. It also makes employees feel dispensable: if a process can be entirely automated, there is little reason to keep an employee employed.
Simply put, it is unlikely that humans will be excluded from all IoT-enabled business processes. For at least the next 5 years we anticipate hybrid models where processes are automated, but require human intervention if certain parameters are breached. We anticipate enterprises adopting more automation as comfort with analytics algorithms and computing frameworks improve.
Until then, a lot of enterprise will just have to live with their control issues.
Steve Hilton is managing director of MachNation, the only dedicated insight services firm for the IoT and IoE industries. MachNation specializes in understanding and predicting the IoT and IoE industries including developments in hardware, platforms, communication services, applications and deployment services. Steve has over 20 years experience providing guidance in the technology and communications sector. For more information, visit: www.machnation.com