[fa icon="calendar"] Dec 19, 2018 / by Ken Pulverman
As we close out 2018, sensors are now a dime a dozen, or pretty close. This pressure sensor (below) is $6.95 online, and I am certain if you buy in bulk they go down in price.
Sensors are so ubiquitous now that if you can dream up something to measure, you can measure it … with the exception of humans. There are 1,000s of companies that will happily line up to make you any sensor you want. In television terms, sensors have jumped the shark. Sorry Henry Winkler, you’ll likely go to your grave still owning this phrase.
In an earlier post, I referenced an angry mob that got really agitated at one of the year’s big industrial conferences because, they said, IoT isn’t the answer (it’s more of a trip to the hardware store, except these nuts and bolts produce petabytes of data and prompt tense conversations with your CFO about your AWS bill). Sadly, you don’t get a nice data lake. That sounds so very pleasant. You get more of a data swamp, with possibly some Facebook executives swimming in it. This may be the reason why one of our contemporaries, C3IoT, recently changed their name to C3.ai. Having helped companies put sensors into warp drive, it is now time to tame the data beast, and AI shows promise in doing so.
The siren call here is that another quick fix really just bit the dust. In my early career, someone once told me, “Begin with the end in mind.” As the angry mob found out, reams of data without the means to turn it into insight is not the end anyone was looking for.
Am I dogging sensors? Absolutely not. I think the ability to measure just about anything for 20 bucks and a wi-fi connection is completely amazing, and this capability will allow us to do incredible things, from driverless mobile offices to flying robo-taxis and all kinds of other things rooted in “The Jetsons.”
What I am, though, is questioning our ceaseless desire to believe that the next tech widget automagically transforms our businesses without the requisite new education and new tools to capture new insights and feed them back into our work.
Here is something that is actually far easier to put a sensor on: a human. If the machines are super reliable and you now try to measure 20 things it does versus two things it does, you don’t glean much more insight other than the knowledge that you need a new vision prescription staring at that screen of data.
As each manufacturer moves closer to its factory of the future, we are doing way more in our existing square footage. This means new recipes for producing new objects are being cooked up every day. To beat Deming’s assessment that 94% of the errors come from the system itself would mean that every operations leader and every industrial engineer would have to design the perfect process and instructions for every new widget a marketing bod dreamed up. This just isn’t possible, and the expectation is unrealistic.
Alternatively, you could design a system for producing a new object that is rapidly self-correcting. If the machines were always screwing up, lots of sensor data would be really helpful. The problem here, though, is that the machine is the most reliable instrument in the band. It just keeps on delivering its notes perfectly until we tell it to stop.
Us humans, though? We are the challenge. It’s not that we aren’t good at what we do. In fact, we are darn good at making sense of complexity. The challenge is that in a modern factory, we don’t play an instrument; we are the conductor and marketing dreams up a new sonata every other day.
Enough with the music analogy. The point is, human actors touch more of the process and more processes and, thus, introduce way more variability. The good news is that with mobile now going industrial, you can effectively have what used to be a super computer in everyone’s pocket and power industrial software to remind them what to do, collect data, provide feedback, and prompt collaboration to immediate resolve issues in flight. If you then combine this with understandable charts and graphs of the contributing human factors to key KPIs like OEE, you can get to root causes and really make some headway – no sensors required.
So boss, I have to admit the six sensors you gave me aren’t entirely senseless. We just don’t have the chops yet to make sense of what they are telling us, nor is this micro-tuning the lowest hanging fruit. I am liking where you are going, but my sense is we have to get our humans organized first.
I’ll hold out for day seven and save this box of sensors for a rainy day, when I have time to tinker and build some pivot tables.
About The Author
Ken is the CMO at Parsable.