The best book I've read recently that relates to work is "Thinking in Systems." I also specially bookmarked an MIT Systems Dynamics YouTube series, which I plan to study thoroughly when I have time. The most inspiring part within it is a section about how to change systems, with 12 leverage points, ranging from weakest to strongest:
12. Numbers: constants and parameters, such as subsidies, taxes, and standards
Only when the parameters start entering the range of one of the higher items on the list does it become a leverage point. In our daily work, many of the metrics we focus on are vanity or ineffective metrics. Sometimes, the designed metrics are overly cautious, leading many project managers (PMs) to concentrate on achieving short-term goals at the expense of long-term benefits—failing to see the forest for the trees. Therefore, these constants and parameters represent the weakest leverage points.
11. Buffers: the stability of inventory relative to its flow
For our non-physical startup company, there aren't any inventory-related issues, but buffers for us include:
Adequate capital reserves to ensure that the cash flow does not break (this is how most small companies fail); At least 1-2 backups for each position; More than 1 product line (this goes against the advice of many investors in startups who suggest focusing, but I actually think focus doesn't apply to all stages and all companies).
10. Stock-and-Flow Structures: Physical systems and their intersection points
For example:
If it is a financial statement, it is somewhat similar to a cash flow statement, designing the company's CCC cash conversion cycle (Cash Conversion Cycle) well. Advance planning of products and business pipelines. Early talent reserve. We often lament that we don't have enough good PMs, but we are reluctant to spend energy on recruiting and training PMs.
Like buffers, physical structures are crucial to the system, but they are rarely leverage points because changing the physical structure is usually not easy and takes effect slowly. Appropriate leverage points need to be designed from the beginning. Once the physical structure is established, to find leverage points, it requires understanding the limitations and bottlenecks of the system, maximizing their efficiency while avoiding large fluctuations or expansions beyond their capacity. For instance, many times the nature of the business and bargaining power determine
9. Delays: The speed at which the system responds to changes
Sometimes, we need to find ways to shorten delays and improve response speed. When reading Wang Dingjun's memoirs, he talked about the contrast between the Kuomintang and the Communist resistance:
"How to deal with such situations, the officers have never taught us, they pretended not to know, so I could only pretend it didn't happen. Later, when I was captured by the PLA in Tianjin and underwent re-education, I found that they held group meetings every night to exchange experiences, improve shortcomings, and solve problems. It was then that I realized that perhaps everywhere inside and outside Shenyang, there might be invisible PLA soldiers holding these review meetings every night. They shortened their sleep time for work, while the military and political officials of the Nationalist Government in Northeast China shortened their sleep time for pleasure. It is exactly as the saying goes: 'Some lie on firewood while others dance on stage, pitifully all are sleepless people.'"
Only when our iteration speed is fast enough and our learning curve steep enough do we have a chance to win in the final competition.
On the other hand, sometimes appropriate Delay can also serve as a protective function for the system. For example, the book gives an example of adjusting orders based on sales volume. The distributor will monitor and predict sales. For instance, if they notice an upward trend in sales, they will increase their orders accordingly to build up inventory and meet the potentially higher sales volume over the next 10 days at an elevated sales rate. This is because there are three time delays (perception delay, reaction delay, and delivery delay) that cause fluctuations in inventory levels.
However, if the response demonstration period is shortened, reducing the reaction delay from 3 days to 2 days to compensate for the perceived inventory gap, things will only get worse—meaning the inventory fluctuations will become more severe. This is what is referred to as "high leverage, but in the wrong direction."
If instead of shortening the reaction delay from 3 days to 2 days, the distributor extends it to 6 days, things will improve significantly, as shown in the figure below:
Iterations should be fast, but the premise is that the direction is correct. If the direction is uncertain, let the bullet fly for a while.
Sometimes, slow is fast.
8. Balancing Feedback Loops: Using balancing loops
7. Reinforcing Feedback Loops: Utilizing reinforcing loops
6. Information Flows: The structure of who has access to what information
The absence of information flows is one of the most common reasons for system dysfunction. Adding or restoring information can be a powerful intervention method and is often easier and less costly than rebuilding the physical infrastructure of the system. We maintain transparency internally while ensuring information security. Sometimes, granting access means greater empowerment and trust.
5. Rules: Incentives, punishments, constraints
As the CEO of a company, it’s even more important to establish reasonable and fair rules. Only when we strive to reshape the rules and understand how these changes influence people's behavior can we truly comprehend the power of rules. Besides setting the rules, leading by example in following them is also crucial. For grassroots startups like ours, we sometimes fall short, such as always making "exceptions" for ourselves. We should rely on rules rather than personal favors.
4. Self-Organization: The ability to add, change, or develop the structure of a system
Self-organization is the highest form of expression of a system's adaptability. A system that can evolve itself can adapt to various changes by altering itself in order to maintain survival. I once tried to persuade my colleagues at the company to learn AI, but with little effect. Later, Lily held several AI competitions, and everyone formed self-organizations. In this process, those who were already interested in AI and had accumulated knowledge stood out, allowing the company to give them more room to perform.
A large number of self-organizing computer models show that just based on a set of very simple rules, extremely complex and surprising patterns can be derived. Respecting diversity and encouraging experimentation may allow organizations to play a greater role.
3. Goals: The goals of the system
A participant in the system can clearly set, articulate, repeat, support, and adhere to new goals, thereby guiding the transformation of the system. I used to not understand the importance of vision and mission, but now I increasingly feel that they are the foundation of a business. When chaos appears in some daily work, vision and mission can better guide individual decision-making and behavior. For each specific organization, the O in OKR is also very important, and setting reasonable O is fundamental.
For example, I once asked a colleague from Alibaba about the difference in objectives between Pinduoduo and Alibaba. Alibaba's goal is "to make doing business easy everywhere," and since they once used GMV as a performance indicator, they have been wavering between serving small customers, brand customers, and consumers. On the other hand, Pinduoduo firmly serves its users. For instance, there was a merchant selling oranges who became a hit product, and Pinduoduo’s operations immediately solicited suppliers across the web for similar varieties, the same quality but at a lower price, redirecting user traffic to this new supplier. This better protected the users but indeed harmed the original orange supplier who created the hit product to some extent. However, the users benefited and stayed on the platform, making the consumer side of this two-sided network effect even stronger. Moreover, Pinduoduo once used the interest from suppliers' goods payments (rather than Alibaba's GMV) as the source of wages and bonuses for the category team, so many operators went to more remote corners to find suppliers willing to delay their payment for longer periods, thereby enhancing the diversity of the system and the bargaining power of the platform.
2. Paradigms: The mindset of a system's goals, structure, rules, delays, and parameters
Paradigms are the root cause of why a system becomes a system. Rooted in these paradigms, the system's goals and information flow, feedback, various stocks and flows, and everything about the system are generated. The system's goals, structure, rules, delays, and various parameters are all directly influenced by the paradigm.
Back to Liu Run's addition, subtraction, multiplication, and division:
The addition in the business world is cooperation at the same level (the upper-left quadrant). For example, how do you arrange your sales work? Assign 10 salespeople, each independently developing clients and bringing home their own harvest? If so, you are managing the company using "addition." Each salesperson contributes to the overall performance of the company in the same way and with the same weight.
However, some companies do things differently. For instance, Beike manages its company through multiplication. In the business world, multiplication means inter-dimensional cooperation (the quadrant at the lower left). Most real estate agencies have each salesperson working independently. But in Beike's view, this "scattered" style of "additive management" cannot accomplish great things. Why not try multiplication?
Beike divides the work of real estate agents into 10 roles. No one person can independently complete a sale; they have different responsibilities and then collaborate to complete a deal. Their contributions to this deal come from different dimensions (10 dimensions), and their weights are also different. This is inter-dimensional cooperation, this is multiplication in the business world. The best algorithms in the business world are basically all about multiplication.
This change from addition to multiplication is a shift in business paradigms.
In addition, we need to learn generative AI, which is also a change in production technology paradigms. Trump made an appearance at the 2024 Bitcoin Conference, embracing blockchain, which represents a change in financial paradigms. These are all very powerful levers.
1. Transcending Paradigms: Beyond Mindsets
Compared to changing paradigms, there is another leverage point at a higher level: freeing oneself from the control of any paradigm, maintaining flexibility, and realizing that having no paradigm is "true." Einstein said, everything should be made as simple as possible, but not simpler. We must also embrace the complexity of the world. The test of first-rate intelligence is the ability to hold two opposing ideas in the mind at the same time and still retain the ability to function: understanding paradigms but transcending them. (I'm just saying, I don't possess top-notch intelligence 🐶)
Breaking free from various rules and restrictions can expand one's learning domain, leading to a deeper understanding of jargon and the unique perspectives of each industry. Since systems are holistic, thinking must also transcend disciplines. Munger's so-called lollapalooza effect probably refers to this.
Maintain an open mind and sincerely respect different cultures, different voices, and different ideas.
Quoting words from the book:
Approach overall reality with the heart of a child. Only then can one enter a state of 'emptiness' (not-knowing), reaching the state of 'enlightenment' as described in Buddhism.
It is to let go into not-knowing, into what the Buddhists call enlightenment.