This session from the Global Peter Drucker Forum has a lot of gems about management in highly complex situation. Many good hints on leadership for knowledge workers.
Thierry Grange, President of the Strategic Board, Grenoble Business School; Special Advisor of the President of AACSB for Europe, introduced the session.
His favorite quote (from page 64 of Drucker’s book):
The purpose of the enterprise is to create value and there are therefore two functions that matter… Innovation and marketing create value; all the rest are costs.
– Peter Drucker
Bain and Company did big study in 2010 and have shown that “Complexity reduces revenue of companies more surely than the size of companies do.” Complexity uses more energy and resources than the problems due to the size of the company.
One solution is to keep the company simple and stupid. This KISS (Keep It Simple Stupid) solution does not work. Another solution is to slash complexity. To this Bain says “Slashing complexity to zero is not the right answer.” Henry Ford reduced complexity by only offering black cars, and that gave General Motors the opportunity to get ahead of them.
Good management requires no predictive power!
Peter Gomez, Professor Emeritus of Management, University of St. Gallen.
Leadership means motivating people, even if you do not have the answers.
– Paul Bulcke, CEO of Nestle
Good management practice is not based on getting more and more data, and getting more details, but it is about detecting patterns to guide actions. It is not about prediction of the future in the face of complexity, but about detecting potential future trends by pointing to the ascent of the knowledge worker.
We need a discipline that explains events and phenomena in terms of their direction and future state rather than in terms of cause — a calculus of potential, you might say, tather than one of probability. We need a philosophy of purpose, a logic of quality and ways to measure qualitative change. We need a methodology of potential and opportunity, of turning points and critical factors, of risk and uncertainty, constant and timing, “jump” and continuity.
– Peter Drucker, Landmarks of Tomorrow, 1957, p15
Nowhere in there does he say anything about ‘prediction’ of the future. Many people would like to predict the future, but complexity science says something about this
Economic theory is confined to describing kinds of patterns which will appear when certain general conditions are satisfied, but can rarely if ever derive from this knowledge any prediction of specific phenomena.
-Freidrich von Hayek
We will never use such an approach to predict anything specific.
Instead of predicting future earnings and valuations separately, we should try to assess the future course of the entire initially self-reinforcing but eventually self-defeating process.
Look at the feedback cycles of this process, and not try to predict it. On a little more positive note:
Financial markets exhibit pockets of predictability associated with pockets of order just like any complex system should.
We should not rely on prediction, but instead on “optimal simplification“. Even with Big Data, getting more and more data and smaller and smaller things will never work.
The situation is that the time required to do management is going up, but the amount of time we have is decreasing. We need simply to respond much quicker. How do we do this? Many times people oversimplify — just do something. When you simplify too much, you have improper segmentation of the problem, one dimensional action orientation, neglect of side effects, and authority behavior. If you don’t know what to do, just command it.
Consider Ashby’s Law from Cybernetics. “Only variety destroys variety.” If you have variety on one side, you need to have enough variety on the other side. You can reduce the complexity of driving by declaring to “drive on the right side”. Had “traffic lights” but a superior idea is “traffic circles”, people need to think more about whether to go ahead, but this helps it work better.
What does this mean for management: there are basic business logic, the traits in the environment, and the corporate culture. With those basics you reduce variety a lot. On the other side we need to invent something called “public value”.
A few guidelines for “optimal simplification”
- You have to be aware of how complex your situation is, because there are different approaches.
- You can try to redraw the boundaries of the system, in order to identify patterns.
- Find pockets of order in the complex situation
Managers can do without predictions of the future. They are futile anyway. However, you can detect patterns, and use that to guide behavior. Early movers, innovators, shape the future, so they don’t need prediction. I don’t expect complexity science to be able to predict things, but if they can provide some insight to recognize patterns, that would be very welcome.
Managing complexity by understanding multiplex networks
If you can not predict the outcome of your management actions, then you neither manage a system nor control it. You are subject to luck, fate, and external events.
From physics we know that if a system is very small you can control them, if very big you can control them, but managing complex systems is completely out of scope for controlling by physics principles. Mathatics and physics breaks down for complex systems because of interconnectedness, feedback loops, etc. Complex systems are evolutionary, but nobody knows the math of evolutionary systems. A little as a decade ago we had no idea of the backbone of complex system theory, and that is networks. We now have something like a theory of “co-evolving multiplex networks”.
Would view an organization as a dynamic network of information flows. Management does nothing but control, shape, and rearrange these flows. This network will restructure itself with and without management. It is not easy to tell a network how to rearrange.
Imagine a new CEO comes to a company. You know the org chart, but really don’t know the real communication flows. You want to know if you restructure things, what will happen? Information flow should be direct, unclustered, and somewhat redundant. Mapping this in a 400 person Australian insurance company by looking at phone logs and email logs. Watching this over time they can determine if the organization is getting more or less bureaucratic — something you want to avoid. If certain key people are removed, then communications breaks down.
PG: As a manager, you need to make so many decisions in a short time. You don’t have time to wait to collect all the data to do in-depth analysis. As a manager you need to act in the moment so you don’t miss the opportunity. If you can use this technique to reduce bureaucracy of the organization, that is good, but it is a very small piece.
ST: However we propose continuous monitoring. Instrument the network, and the results are available whenever you need them. Serious complexity science is not about predicting the future, but instead about phase transitions. Water can be liquid, steam, or ice. Complex systems have much more complex phases, but you want to know when these things change.
PG: Some of the examples mentioned here are just complicated system, and those are not hard to manage. A really complex system really changes in unpredictable ways. Look at a stock market: you need to consider what other people are thinking that other people are thinking of a stock. Market is dominated by the power law behavior … extremely large swings.
ST: It took a long time to understand water: hundreds of years. Complexity science is so young, we should not expect this to have dramatic results so quickly. Game theory has come a long ways, and can now provide useful results.
ST: (What about fractals?) Fractals always appear when you have non-linear systems. Fractals are a mathematical description that can be used as a tool.
ST: (Can organizations map their information flows today?) There are companies looking at communications plans, as well as “other people” looking at your telecomm connections.
PG: Still very early. Yet, these system where human activity is the deciding factor, will never be analyzable in the way that water/ice is. Never be able to make prediction about individuals and what they should do. We should instead rely on heuristics from other fields, like cybernetics. One heuristic is to always make choice that increase the number of future possibilities (except when choosing a spouse).
ST: (Will future employees ask to see such networks before joining a company?) Might be nice, but if you show me such a network today, I would not be able to read such a complex dynamic network.
PG: (Can Ashby’s Law be use to help management know if they are doing a good job?) Such an instrument might be useful to help guide investment. You should not be collecting data, data, data, but instead you should be looking for patterns.
ST: Data need to be put into a model that can help you distinguish those things that are causal and what is random. For example a MMOG with 500,000 people. Take this information and analyze it, you would be amaze at what you can predict.
PG: Look at the stock market. You can not predict what a single person can do, but if you consider a group of investors, you might be able to predict a general pattern of this group behavior. But on individual level, forget it.