What does “separation of human work and machine work” mean?
Dear Gemba Coach,
I’ve been trying to learn more about Jidoka and I keep coming across references to “separation of human work and machine work” but I’m not sure what this means – any pointers?
Have you tried recently to log on a website you use infrequently? Puzzled at the password? Pressed the “forgot password” button? Waited for the e-mail? Got to the page eventually and struggled with coming up with a new password that you’d have a slight chance to remember without it being your kid’s birthday, not having the right number of characters or capitals or whatever, and finally logging on to do whatever you wanted to get done in the first place?
In this case, there is no separation between your work and the machine’s work – you need to fiddle with the website (plead? beg?) in order to get your own work done. Then, maybe you’ve got to print out a form out of that website. Mostly you hit the print button, et voilà – it comes out. Or there’s no paper in the tray, you have to change the cartridges, etc.
The central idea here is standardized work: routine tasks should be done in a standard sequence of steps that eliminate as much waste as possible to focus on the important part of the job – where quality is being built into the process.
In the lean perspective, the human is always leading the work, and the various tools support human work (even if the tool is a huge press or a computer system). In Western automation, humans are traditionally there to cope with parts of the process we can’t automate. Not so in lean – the entire process is seen from the person’s point of view, with support for every difficult action, in order to focus on the real value-adding work -- which is hard to do if you’ve got to fiddle with the tools all the time just to make them work.
Most of our daily lives are now deeply enmeshed with computer systems. Go to a restaurant and order, you’ll see the waiter fiddle with his tablet. Stand up and ask for the check, watch the waiter punch buttons, get it wrong, do it again, get the check out. For that matter, when you to the doctor they barely glance at you while they get the right files up on the screen, get on with tapping at the keyboard, and eventually, if it’s a good day, ask, “How are you today?”
This is probably a transitory phase as apps will get better and better, but most of the systems around us don’t allow separation of work between man and machine. In the restaurant case, for instance, that would mean a machine that prints the check the moment the order is closed so that the waiter can come and pick it up when he’s ready, and bring it to you.
The machine should do its work, produce its result, and have it there waiting for when you need it in your standardized work sequence of steps, so that you don’t hit a barrier in the flow of your own work.
I was on the shop floor recently in high-tech assembly where operators had to check every single part before performing a delicate operation – no separation of machine work and human work – they couldn’t trust the system to just give them parts to assemble without having to carefully check whether the part was okay or not. Then, they used an oven, and they knew they had to inspect the parts on the sides of the tray that often got overcooked, and so on.
In practice, eliminating waste out of work starts with examining abnormalities – when things don’t work out the way they should – much like with scientific thinking, we find new ideas by looking into where our models turn out to be wrong.
4 Starting Points
If the work is riddled with abnormalities, it’s going to be really hard to figure out the real waste. What machines, however, have going for them is that they’re really consistent. It’s a good place to start to clear the window. There are four starting points to investigate how well (or badly) the system is kept independent from the person using it:
- Upgrades: Systems have weak points that can only be discovered by using them, so regular automatic upgrades are the key to keeping the system available for users. Similarly, machines have parts that get more wear and tear than others. Studying these parts and implementing counters to know when we’re entering the danger zone before the problem occurs, and signaling the need for replacement (with enough of a buffer that we can do it when we’re ready to do so) is also a good place to start. Your car tells you when it’s about to run out of gas, with enough time to find a petrol station.
- Start-up: We experience many problems in getting the system or the machine up and running so that we finally get to do what we want to do. Back to the password example, I was impressed at Google to see that the person I met there had (1) a physical key to plug into the computer and (2) a very simple (easy to remember) password. The thinking is that if the computer gets attacked by malware, without the physical key, there is no access to the files, and if the computer is stolen with the key on it, without the password, no access to the files either. This system makes life much easier to start the machine. Similarly, with most machines, investing engineering effort in simplifying starting the machine (which happened with cars, for instance – I remember having to fiddle with starters and so on as a kid) separates your job and that of the machine.
- Delivery: How can the system or machine give you an answer that waits in a special place for you to use it, rather than keeping you waiting there for the operation to finish. The old-fashioned mailbox was great: you got to pick up your mail in your own time. Now, with the increase of home deliveries, you’ve got to be there to open doors, sign off, and so on. I hear that Amazon is trying something with giving people access to your place, so they can deliver without needing you there. Same thing with shopping – why all this fuss with queuing at the register? Again, Amazon is working on figuring out how to charge you for what you buy as you pick it up – we’ll see. In machine work, much the same – designing machines so that they self-eject parts and operators can pick them up when and as they need them.
- Poka-yokes: This is any device that alerts you ahead of time that something is not quite right, and you’d better stop now and check, rather than doing the job because something is off. Basically, the way to reduce the resources needed to control any process is to narrow down and visualize where control is really needed, and then progressively automate or secure these so that the person can get on with their work without having to inspect, confident that if something goes awry the machine or the system will warn them.
Each of these moments reveal how the system really works, and can open up our investigation. I realize this is often hard to do technically, but in the end, like many lean principles, it’s well worth the effort. Beyond mechanical systems and machines, the need to separate human work and system work is everywhere. Yesterday, I had to phone an administrative office because they sent me an invoice and there was nowhere in the long verbose letter explaining why I had to pay, who to write the check to, and – with two different addresses listed -- where to mail it.
This might sound silly, but as many of our jobs are now about controlling systems, look at the many unclear reports produced where a person needs to interpret somehow what the report says. Again, this is time taken away from doing the important job of moving value forward.
The same thinking applies here – how can we make more conclusive, self-explanatory reports that require less interpretation and that we, as users, can interpret as in “go left” or “go right” in our own time without having to fight with the system to get a clear-ish answer.
A key idea in lean is reusable learning: we don’t know what the solution should or will be, but we have a pretty good idea where to start and what direction to improve.
The starting point here is abnormality control. Let’s look at the normal flow of work of the person and spot where they get in trouble: hesitation, rework, waiting, and so on.
Then let’s focus on taking away all issues born from dealing with the machine or the system, to separate machine work from human work, and have all components of the human work good and ready for when the person needs them.
The vision here is to create purely enabling systems that enable us to perform the value-added part of the work without having to fight with tools and systems just to get the work done. Being able to trust that system work is a large component of mutual trust in a wider sense, and of satisfaction at work (as, conversely, systems that constantly break down on you is a major component of frustration and job disengagement). Separating human work from machine work is a critical component of the lean thinking system and the key to standardized work. As with the many aspects of jidoka, it’s a shame it gets such little attention in the literature. So many thanks for your question – this is indeed one of the deep secrets of making lean work.
Lean Lessons from Cobra Kai(zen) and the Karate Kid
The unexpected wake-up call of the modest perfection of the original Karate Kid movie was that we need to move beyond defending this or that method of work and look to highlight opportunities of improving things beyond monetization, says Michael Balle in this reflection on the meaning of this classic movie.
How Using Kanban Builds Trust
Kanban functions as a trust machine because everyone using it must understand what they have to do and why, says Michael Balle: "Our purpose here is to share our ideas on what we believe is important in lean thinking."
The Sanity of Just-in-Time
Path dependence is the worst enemy of smart resolution, argue the authors, who suggest greater "frame control" with enabling tools such as just-in-time to respect people on the frontline and respect the facts they share about what is happening to them. "Mastering the path as opposed to being led by it, means looking up frequently to reevaluate both destination and way as new information comes to light."