Presentation: David Faure Breaks The Law

• Event: Akademy 2014
• Location: Brno, Czech Republic
• Date: 7.09.2104
• References to sex acts: 1

Say what? @valoriez is surprised to hear what I am going to say about #KDE development in my talk tomorrow. #Akademy pic.twitter.com/FLPeJQYNHb — Paul Adams (@therealpadams) September 6, 2014

The Short Version

David Faure is a pleasure to work with, both within KDAB and KDE. He maintains three important character traits that, as engineers, we should all aspire to:

• He keeps a sense of fun about him at all times. This humour about his work makes him a joy to work with;
• He is constantly learning and improving, even with the wealth of experience he has behind him;
• He always has the actual end user in mind.

Despite these wonderful character traits, he has a dark side... He is a lawbreaker. So are many other KDE contributors.

In this presentation I talked about David Faure, Brooks' and Conway's laws and evaluated them in the context of KDE. Some rather fascinating, but alarming results are found!

Slides And Brief Notes

Slides 1 -- 6

David Faure: all-round legend and nice bloke of KDE. Fun to work with, thirst of knowledge and skill and, always thinks about the end user.

Slide 7

Positioning David Faure within the context of the wider KDE project and the people he works with.

Slides 11, 12

Introduce Brooks' and Conway's laws: it is incredible how many people identify as software engineers, but do not know what these laws are all about.

Slides 14 -- 16

How can we measure Brook's law in Free Software projects. We need to start by discussing how we measure anything at all! What are the sources for data in Free Software development.

Slide 17

How we model the community: nodes are contributors, edges show that the two contributors worked together. "Work" means they both edited the same resource in the observed timeframe. Edges get higher weights if the contributors work together in multiple resources.

From this we can find "cohesion", the average length of the shortest path between all pairs of nodes. The higher the cohesion, the more closely-knit the project is. If Brook's is correct then we should see cohesion drop with increased community size...

Slides 18 and 19

Yup, Brooks nailed it...

Slide 20

...or did he? KDE is clearly outperforming expectations here. Awesome. Well done KDE! Hint: This is where Conway's law comes into interest.

Slide 22

Wait, if we look at the data over time, something isn't right. Cohesion grows since 2004 until late 2007 (KDE 3.5, KDE 4.0?) but then cohesion declines...

Slide 23

Which just so happens to be the same pattern for funding of the eV, just offset by one year.

Slides 24 and 25

KDE has lost its way a little bit. Nothing disastrous yet, but the ship does need righting. As a start, being more like David and enjoying our work, continuing to learn and focusing on users, will get us headed the right way.