Jun 10, 2014 | Posted by Ben Thompson
If we look at the framework of the scientific experiment and compare it to the framework of a kaizen we will find that their pragmatic schemes are practically identical. I hope that through comparing kaizen to the scientific method it will show how kaizen is the science of production:
In TPS it isn’t so much as a question that needs defining but a problem. But in order for this problem to be defined, in what process does it occur? how often? what is its impact on production? etc., we must first observe the process; otherwise we are just acting based on opinion.
This stage is arguably the one of utmost importance. From this point we establish a base line of how things are currently in the processes. This is that idea of job analysis and data collection I talked about in an earlier post.
“The idea behind job analysis and data collection is to time processes to gather data on how long the work within a process takes. We first take ten full cycle observations, to gather data on the process as a whole. We highlight the lowest repeatable time that this process can be completed in, because if it is repeatable within ten trials it is very likely it can be repeated continuously. If we take this repeatable time and subtract it from the greatest time it took to complete the process we find the fluctuation in the time it takes to complete the process. ” – Adventures in TPS Learning, Job Analysis and Data Collection! March 19, 2014
With our notes and timed observations on the process we can now proceed in planing and predicting the kaizen, much like how a scientist plans and formulates his experiment by developing a procedure to test his hypothesis. We use the information we know to infer predictions on our kaizen’s results. We develop a future state of how we foresee the process and then incrementally achieve it through kaizen.
Once more without performing the experiments or kaizens in a controlled measurable and reproducible manner we are just operating on opinion. So like the scientist we learn by testing changes in variables recording what we see, to establish the facts.
The Analysis is the grunt work, the number crunching of raw data and organization of the data in to clear concise terms such as table and graphs etc.
With the analysis of the data complete comes the interpretation of all of this data. The comparing of the data received to the initial data to see if the hypothesis has merit for implementation to get us closer to our desired future state or perhaps fail and cause us to rethink, reform and retest our assumptions based on our data.
The results of a kaizen are presented to production\management in the form of an A3. Consider it like a kaizen lab report, with a problem, data, analysis, results, kaizen strategy and next steps as sub headers. This stage is important to achieve consensus so we can move forward and implement the finding of the kaizen trail or simulation. It mixes hard facts with incremental instruction to get things done because unless the results are utilized what was the point of the kaizen.
Unlike the scientific method our retest is not done by our peers to corroborate evidence and publish in a scientific journal. Our retest is taking what we have done, learned, implemented and building on it to get ever closer to that true north future state. With waste-less processes running one piece continuous pull systems direct to the customer with no inventory needed. The point of the true north condition is not to achieve it but to get ever closer through kaizen. Just like how science will continuously get ever closer to understanding the universe without ever fully defining its awesome nature.