【Shu-Ha-Ri of Prompting】What lies beyond "copying as-is": From tacit knowledge to explicit knowledge
Last time, I shared two reasons why copying someone else's prompt doesn't work, and how to zero out the context.
※Click here for the previous article⬇️
Starting by "copying as-is" from successful people is the starting point.
But what if you want to "add your own color" or "adapt it to your own environment"?
What do you do then? I will delve into that this time.
1️⃣ What is "Shu-Ha-Ri"?
Are you familiar with the term "Shu-Ha-Ri," which represents the stages of learning in traditional Japanese culture (especially in martial arts and the arts)?
I asked Grok to create an explanation, and it is as follows.
In a nutshell, it means "first imitate (Shu), then break (Ha), and finally make it your own (Ri)."
Shu (Protect): The stage of following the rules
Faithfully following the teachings and forms of a master to thoroughly master the basics. The beginner stage.
Ha (Break): The stage of breaking the rules
Once the basics are solid, analyze and critique the forms, and arrange or improve them in your own way. Intermediate to advanced level.
Ri (Leave): The stage of leaving the rules
Not bound by forms or rules, establishing your own unique state. The stage of an expert or master.
2️⃣ Interpreting experience and data through "Shu-Ha-Ri"
The other day, my follower Kaname-san used the term
"Shu-Ha-Ri" to discuss:
Ways of thinking about data design
Records and memories
Tacit knowledge and explicit knowledge
in an article they posted.
How to manage experience and knowledge. The outline is as follows.
※This is my interpretation, so if you want to know the exact details, please see the original article.
"Shu": Can be recorded by verbalizing it into procedure manuals.
"Ha": Recording is difficult. It can be verbalized, but it is meaningless unless compared with the person's own experience and memory; the need for context (information between records) arises.
"Ri": Not suitable for recording. The stage of wisdom where elements of sensation and experience are strong.
The wisdom of the "Ri" stage of an expert cannot be fully recorded.
However, by carefully verbalizing the "Shu" and "Ha" stages, you can shorten the time it takes for a successor to reach "Ri."
For that, design is important.
3️⃣ Another "Shu-Ha-Ri": First, faithfully follow the model of successful people
Actually, the concept of "Shu-Ha-Ri" applies directly to prompt utilization, which is also digital.
In a zero-context environment, the copied prompt worked as intended.
That is the completion of 'Shu' (protect/follow).
Only when the meaning of the framework is fully understood does the right to proceed to 'Ha' (break) arise. Conversely, adding your own flair before 'Shu' is complete is not 'Ha', but rather wandering off course. The 'failure of mixing in your own style' I wrote about in the previous article was caused by skipping this 'Shu' step.
4️⃣ Mixing your own style into prompts: From 'Ha' to 'Ri'
Once 'Shu' is complete, you start adding your own context. However, it is important not to change things blindly, but to be clear about which parts you are changing.
🍤 Step 1: Separate the variable parts from the framework
Within the prompt, identify the 'parts that should be changed according to your work or situation' and the 'framework parts that should not be touched'. This is the same concept as the [Enter content here] approach from last time.
🍤 Step 2: Change one part at a time
If you change multiple variable parts at once, you won't know which change affected the output. Change one part and verify, then change another and verify. This repetition deepens your understanding of the prompt.
🍤 Step 3: Keep a record of why you changed it
The verification results from Step 2 are important data. As you increment versions, keep a history of the reasons 'why you changed it this way.' This is important to avoid repeating the same mistakes. Even if it's a hassle, record it in the form of prohibitions, such as 'Avoid XX.'
This accumulation of trial and error leads to 'Ri' (transcendence)
. And it leads to converting tacit knowledge into explicit knowledge, which I will touch upon in the next chapter.
5️⃣ This has been a theme of mine since I started writing on note
The baby boomer generation is retiring. And there is a labor shortage. People are being replaced, and handovers are not being done sufficiently.
In my day, we were taught to 'steal by watching the backs of our seniors.' But in this day and age, we need a system for teaching. Since we weren't taught how to teach, we don't know how to do it.
Isn't this happening in your organization too? Actually, because of this, tasks that were previously handled by experienced staff are not being performed adequately. Such problems are starting to surface.
That is precisely why the importance of elevating tacit knowledge into explicit knowledge is increasing more than ever. However, until now, it has been difficult for us to verbalize the tacit knowledge of experience on our own.
Can we solve this by utilizing AI?
My original motivation for starting to write on note lies here.
And here, two 'Shuhari' paths intersect.
The 'Shuhari' of experience and data from Chapter 2.
The 'Shuhari' of prompts from Chapters 3 and 4.
Use AI to organize the 'Ha' and 'Ri' of experience, which were difficult with just humans and records. Improve the accuracy of the reproducibility of the means to organize with AI through the 'Shuhari' of prompts.
Currently, I am in the process of examining and verifying prompts that articulate human experiences and ways of thinking = elevate tacit knowledge into explicit knowledge. As shown in my previous and current articles, thanks to two of my followers, I was able to logically construct the background for this. I would like to take this opportunity to express my gratitude once again.
📌 Summary
In line with the concept of 'Shuhari' (follow, break, transcend):
As 'Shu' (follow), I will appropriately record daily information.
As 'Ha' (break) and 'Ri' (transcend), I will articulate my judgment criteria and ways of thinking as much as possible.
I intend to continue working on these.
By clarifying the overall way of thinking, I have reached one goal. And from here, it is also the starting line for elevating tacit knowledge into explicit knowledge. It is not yet in a state where I can share it with everyone, but I hope that a day will come when I can.
This ended up sounding like a series finale, haha.
