Debugging Your Beliefs

A few days ago I wrote this tweet:

Debugging is often about uncovering your implicit assumptions and verifying that they actually hold. — Jan Gosmann (@jgosmann), July 4, 2014

There is actually a lot to discover in this short statement and it even touches social issues. But for starters, let us take a look how bugs in computer programs come about.

How bugs are made

I think, there are at least two main reasons why bugs in computer programs occur. Sometimes the programmer just mistypes something. By double checking the code this is usually easy to find.

A second type of bug, however, is much harder to locate. It is the result of implicit assumptions made by the programmer. When programming you make every now and then an assumption how a programming language behaves or what a certain function call does without even being aware of this in that moment. As long as these assumptions are correct everything is fine. But when relying on a wrong assumption things might go awry.

I have a recent example of my own (which can be skipped over if you are not keen on programming itself). I wanted to create a list of size empty lists for which I wrote the following Python code:

[[]] * size

Can you spot my assumptions here? I am assuming a new list instance will be created when replicating [] and, thus, that all lists contained in the outer list will be different instances which can changed individually. However, a list containing a single list instance will be created first and then the reference to that same instance gets replicated. Thus, modifying one of the inner lists will modify all of them.

How bugs are found

How does one find such bugs? Essentially, it is necessary become aware of implicit beliefs and verify that they are correct. Once I realized what assumptions I was making in the example above it was easy to verify that the code does not produce independent list instances. Fixing the bug was then a matter of seconds. The hard part is to achieve awareness of implicit assumptions. It all comes down to the question: How do you become consciousness of something unconsciousness?

Sometimes another person can be of help. That person will have another set of presumptions and can therefore be able to spot such bugs more easily. This is a good reason to do code reviews and/or pair programming.

But what if you are on your own? In that case, I do not know.

Part of it is probably experience and practice. But I wonder: Are there methods to train becoming aware of implicit assumptions?

Why does it matter to non-programmers?

Implicit beliefs do not only exist in programming. They are part of our everyday life. Everyone has some assumptions about the persons he or she meets without even being aware of it. Nevertheless, they influence how we interact with each other. Teaching people about uncovering their implicit assumptions might have the potential to make the world a better and fairer place.