
A curious baby. According to Prof. Laura E. Schulz, associate professor of cognitive science in the Brain and Cognitive Sciences department at MIT, babies' approach to learning about the world has much in common with the scientific method. In some sense, we would do better at debugging complex systems if we took a more childlike approach. Photo (cc) Creative Commons Attribution 2.0 by Kenny Louie courtesy Wikimedia.
To repair complex systems, many resort to "random twiddling and part replacement" (RTAPR) when they're under time and resource constraints. Sadly, RTAPR doesn't work very well. For example, consider a system that has six commercial off-the-shelf components. Let's suppose that it isn't working right. We decide to replace Module 2, which produces no change — the system continues to misbehave. Some might conclude that this proves that Module 2 is OK, but that conclusion might be mistaken. Suppose that the problem lies in the firmware of Module 2, which controls how it operates on the data it receives from Module 1. Since both of our Module 2 boxes contained the same firmware, the system behavior didn't change when we made the swap. A conclusion that Module 2 was not involved in the fault would therefore be incorrect.
A more careful approach can work better than RTAPR. Here are some guidelines that form the basis of what is usually called the scientific method.
- Perform no random experiments
- Random experiments, especially those involving system configuration changes, are unlikely to produce new knowledge. The more complicated the system, the less productive are random experiments.
- Keep excellent records
- Record the Random experiments, especially if
they involve system configuration
changes, are unlikely to
produce new knowledgedetails of all experiments and results. Typically, you won't refer to these notes until you're completely stumped, but that happens with alarming frequency for complex systems. So write the notes so as to make them clear in that kind of desperate situation. - Try to replicate unwanted behavior
- (a) If the unwanted behavior is reliably repeatable, observe the results of making a minimal change to the system. Any change in behavior can be revealing. (b) If the unwanted behavior isn't repeatable, try to find a system configuration that makes it repeatable, and then go to (a). In all such experiments, controlling the system's containing environment is essential.
- Base all attempts on hypotheses
- Because the input configuration for a complicated system is also complicated, proving that complicated systems work for all required inputs is difficult. Hypotheses about why the system isn't working are equally difficult to prove. Hypotheses can more readily be disproven than proven.
- Therefore, have a testable hypothesis in mind whenever you change the system configuration. Testable hypotheses are of this form (for example): "The fault might be A. If experiment B produces behavior C, then the fault cannot be A." Repeating this process gradually eliminates possibilities until only the truth remains.
- Fail forward
- Devise hypotheses and experiments that cause your investigation to "fail forward." That is, favor experiments that produce useful knowledge whatever the outcome of the experiment. If you make a change and the system starts working, that should help explain what was wrong. And if that same change causes some other result, that, too, should be enlightening information.
Adhering to these guidelines can be difficult, especially under pressure. If deviation is required, make note of it, and note how deviations affect your conclusions. First in this series Top
Next Issue
Are your projects always (or almost always) late and over budget? Are your project teams plagued by turnover, burnout, and high defect rates? Turn your culture around. Read 52 Tips for Leaders of Project-Oriented Organizations, filled with tips and techniques for organizational leaders. Order Now!
Your comments are welcome
Would you like to see your comments posted here? rbrenogMhuqCxAnbfLvzbner@ChacigAthhhYwzZDgxshoCanyon.comSend me your comments by email, or by Web form.About Point Lookout
Thank you for reading this article. I hope you enjoyed it and
found it useful, and that you'll consider recommending it to a friend.
This article in its entirety was written by a human being. No machine intelligence was involved in any way.
Point Lookout is a free weekly email newsletter. Browse the archive of past issues. Subscribe for free.
Support Point Lookout by joining the Friends of Point Lookout, as an individual or as an organization.
Do you face a complex interpersonal situation? Send it in, anonymously if you like, and I'll give you my two cents.
Related articles
More articles on Problem Solving and Creativity:
Dealing with Deadlock
- At times it seems that nothing works. Whenever we try to get moving, we encounter obstacles. If we try
to go around them, we find more obstacles. How do we get stuck? And how can we get unstuck?
Virtual Brainstorming: I
- When we need to brainstorm, meeting virtually carries a risk that our results might be problematic.
Here's Part I of some steps to take to reduce the risk.
Thirty Useful Questions
- Whether solving technical problems, creating plans, or puzzling through political tangles, asking the
right questions can be the key to finding useful approaches. An example: What questions would I like
to know the answers to?
Disproof of Concept
- Proof-of-concept studies of system designs usually try to devise solution options and discover the system's
operating constraints. But limitations can become clear too late. A different approach — disproof
of concept — can be a useful alternative.
Pseudo-Collaborations
- Most workplace collaborations produce results of value. But some collaborations — pseudo-collaborations
— are inherently incapable of producing value, due, in part, to performance management systems,
lack of authority, or lack of access to information.
See also Problem Solving and Creativity and Critical Thinking at Work for more related articles.
Forthcoming issues of Point Lookout
Coming December 13: Contrary Indicators of Psychological Safety: I
- To take the risks that learning and practicing new ways require, we all need a sense that trial-and-error approaches are safe. Organizations seeking to improve processes would do well to begin by assessing their level of psychological safety. Available here and by RSS on December 13.
And on December 20: Contrary Indicators of Psychological Safety: II
- When we begin using new tools or processes, we make mistakes. Practice is the cure, but practice can be scary if the grace period for early mistakes is too short. For teams adopting new methods, psychological safety is a fundamental component of success. Available here and by RSS on December 20.
Coaching services
I offer email and telephone coaching at both corporate and individual rates. Contact Rick for details at rbrenogMhuqCxAnbfLvzbner@ChacigAthhhYwzZDgxshoCanyon.com or (650) 787-6475, or toll-free in the continental US at (866) 378-5470.
Get the ebook!
Past issues of Point Lookout are available in six ebooks:
- Get 2001-2 in Geese Don't Land on Twigs (PDF, )
- Get 2003-4 in Why Dogs Wag (PDF, )
- Get 2005-6 in Loopy Things We Do (PDF, )
- Get 2007-8 in Things We Believe That Maybe Aren't So True (PDF, )
- Get 2009-10 in The Questions Not Asked (PDF, )
- Get all of the first twelve years (2001-2012) in The Collected Issues of Point Lookout (PDF, )
Are you a writer, editor or publisher on deadline? Are you looking for an article that will get people talking and get compliments flying your way? You can have 500-1000 words in your inbox in one hour. License any article from this Web site. More info
Follow Rick





Recommend this issue to a friend
Send an email message to a friend
rbrenogMhuqCxAnbfLvzbner@ChacigAthhhYwzZDgxshoCanyon.comSend a message to Rick
A Tip A Day feed
Point Lookout weekly feed
