Blockchain and Manufacturing are worlds apart
Blockchain is, truly, a clever piece of technology.
It is decentralised and yet, every member in the network is able to have complete trust in the contents.
It is exceptionally secure, simply, because of the way it is designed.
But how well does it fit into the manufacturing scene?
Manufacturing is a very old field.
It has been there for much longer than the kinds of computer programmes that are taught in high schools these days.
It is also heavily restricted by the laws of Finance.
Due to the sheer amount of capital investment needed for a part of one line, Manufacturers are very careful about upgrading production lines.
You may find a twenty year old machine next to a state-of-the-art upgraded model making exactly the same product.
Modifying the production lines to be able to integrate with the tracking aspect of the Blockchain might be relatively easy for one of them.
But the old machines may struggle to cope with the technological and physical demands of maintaining the traceability throughout the chain.
After all, the entire Blockchain can only be as strong as its weakest link.
Turning Garbage into Fertiliser
There is a well-known saying "Garbage In Garbage Out", or GIGO as it is affectionately known.
If you start with garbage, you are very likely to end up with garbage.
But there is another way!
You could turn it into fertiliser.
The trick lies in filtering out what is useless and focusing on the good stuff.
This is what data science is about.
Imagine yourself standing in the middle of a busy intersection with cars honking their horns, drivers screaming insults at each other and an ambulance trying to squeeze through with its sirens on.
Now imagine that you are having a discussion about Einstein's equations.
All of the noise around you is going to drown out any creative thoughts that you might wish to exchange.
Big Data also has a lot of noise, which is a polite way of saying useless information.
If the garbage is not excluded before the analysis, you are going to end up with, you guessed it, garbage as the result.
The role of a data scientist is to question, identify and filter out the irrelevant pieces of the puzzle.
If you have carefully sifted through the garbage and been picky about what you choose, you might just be able to turn it into fertiliser!
Programming Certainty vs Artificial Intelligence Best Guess
In the world of traditional programming, every piece of code has a definitive outcome.
Artificial Intelligence (AI), on the other hand, is more of a "Best Guess" kind of code.
In a very loose sense, there is no certainty in AI but just a probability.
In technical terms, this would be called "confidence", "threshold", "standard deviation" or something similar.
Yes, there is a significant advantage of using a "Best Guess" method as you can have a possible solution even with incomplete information.
Hence, the presence of "I" for Intelligence in AI.
But it also comes with a disadvantage.
AI will most likely make guesses that we would consider to be incorrect for a certain number of repetitive tries until it gets it pretty much right.
In the case of solving a complex puzzle or playing a strategic game, for example, AI would simply need to keep trying until its way of guessing becomes better than how we humans think.
In the case of Manufacturing, the initial tries, and more importantly, the failures, could have an impact on the bottom line.
The same could be said about Transport, Supply Chain and Medical Services, just to name a few.
It could be even worse if the harm is not just contained to the profit but also extends to human beings.
It would be worthwhile keeping the possibility of a guess being incorrect and its effects in mind while deciding upon the extent and speed of introducing AI into existing processes.
In some of the fields, it might take quite a while for AI to become better at handling decisions than we are.
Until then, you would still need a human to interpret, question and validate the guess.
The Trap of Big Data
Big Data is the buzz-term that almost everyone seems to be talking about.
It is exactly what is sounds like.
There is a really big volume of data and you would like to do something about it.
But what can you actually do with a lot of data gathered from a factory environment?
OEE can be done without big data. In fact, in the scheme of things, this would be "Little Data".
If you are trying to find which products run the best on which production line, you could simply ask the line managers and the shift leaders.
If you are trying to compare processes across different factories, you are going to land in the Big Data Trap.
Statistics teaches young data scientists to be vary of the omitted variable bias.
Anything, absolutely "anything", that is not in the analysis will stuff up the interpretation of at least one factor that is in the analysis.
If you do not have the temperature at a crucial area of production, any adverse effects arising due to differences in temperature might be interpreted to be related to shifts.
If morning and afternoon shifts rotate but the night shift stays fairly stable, any negative effects due to cooler temperatures might be falsely attributed to the hard-working staff in the night shift.
This is the trap that you could land yourself in.
Believing the result of the Big Data Analysis to be the truth without questioning the missing data.
You can imagine the impact on the bottom line of the company by "blaming" production issues incorrectly on the night shift.
Big Data Analysis is a science.
And just like science, the interpretations and the models have to be allowed to improve.
Blindly accepting the conclusions as final would be the equivalent of still believing that the Sun is the centre of the Universe and never ever challenging it.
After all, the data in the 16th century was conclusive according to Copernicus to validate this theory.
It took a few centuries of a lot more data to even consider the alternative.
How long would you be stuck in the Big Data Trap?
Machine Learning and Machine Mistakes
Machine Learning is a very specific term used for self-learning algorithms.
But in order to learn, you have to make mistakes.
That's how a child learns to walk. It needs to fall repetitively before it knows how not to fall.
That's how we are still learning new things as adults.
That's what you are doing right now.
You are subconsciously comparing what you are reading with your perception of Machine Learning.
You are trying to map the deviation between new "data" and previous "belief".
Machines are no different.
After all, we are teaching them to learn the way we learn.
By making mistakes!
Over the decades, we have come to expect machines to be flawless, especially, if we are the end consumer.
We believe, or want to believe, that they do the same thing over and over again without fail.
But with Machine Learning, it is the exact opposite.
To reach the "learned" state, there might be a trail of mistakes along the way.
This is a paradox that we, as humans, must learn to accept!
Manufacturing isn't funky, it is efficient!
Every field has its beneficial traits.
Statisticians need to be data oriented, Web Designers need to be user experience oriented and Underwater Cave Divers need to be, well, spatially oriented.
Manufacturing is a different kettle of fish altogether.
Yes, there is a lot of data.
Yes, there are quite a lot of user interfaces.
And yes, you need to know which way is up!
But manufacturing is much more than that.
The primary goal in any factory is to produce the goods as efficiently as possible.
That includes minimising waste, making the most number of items in the given time and avoiding breakdowns of machines.
This may sound quite straightforward.
Aim to produce what you want to, when you want to and how you want to.
But more and more "funky" solutions are creeping into manufacturing.
Whether it is a button with rounded corners, a warning sign that quietly sits in the corner or animation that makes you dizzy.
The focus is shifting towards the funkiness of the solution and away from the efficiency of production.
Find it hard to believe?
Have a look at the last major project in your factory and give it a fair assessment. Was it funky or was it efficient?
What are IoT, IIoT and Industry 4.0?
IoT stands for the Internet of Things.
It is about connectivity between devices, people and systems over the Internet.
Now, you stick another "I" in front of it for "Industrial" and you get IIoT.
This simply implies that you connect devices in an industrial environment, such as manufacturing.
And what about Industry 4.0?
Well, this is the German Government's strategic initiative around manufacturing and connectivity [GTAI Industry 4.0].
The number "4.0" stands for IoT being the "fourth" industrial revolution.
So, in essence, IoT, IIoT and Industry 4.0 are more or less the same.
Whether you need the first "I" in "IIoT" is, really, a matter of taste and your definition of "Industrial".
Is IoT really that new?
The Internet of Things has gained a lot of traction in the recent years. But is it really that new?
The "things" that we might want to connect to, have been there for a while.
And the Internet, well, that was there since before the start of the century.
So, why the fuss about all of this now?
The answer lies in a few factors that have come together.
First, technology is cheap!
Gone are the days when companies had to worry about storage, memory or processing power.
Just look at your smartphone.
What are we upto now? Eight cores in a device that fits in your hand?
It was only about ten years ago that the first dual core processors started hitting the market.
Now, four times as many cores fit into much smaller space.
Oh, yes, that's another point.
Electronics have become more and more compact.
To make it even easier, connectivity is expected to be more readily available than air!
"The Internet is not working" is a phrase that ranks higher on the panic scale than "All flights for the day are cancelled due to bad weather".
So, things are smarter (not more than humans, at least for now), they are always connected (either that or the world is about to end) and technology seems to be cheaper than clean air (seriously, someone is making money by selling air).
And, voilà ... IoT is hotter than the melting point of Tungsten (it's pretty hot!).