The Human Factor in Software Engineering

Today, it became official: I am supporter of the IREB International Requirements Engineering Board. I am member of the working group writing the syllabus for the Advanced Level Certification in Requirements Management.

I know, this is a bit off-topic because it has nothing to do with software development. But I like this idea and want to share it: We are now doing research towards the zero waste society in Europe. I am looking forwards to this. I know, it will take a while until we reach perfection. But meanwhile, I am doing my share of recycling and bring back some batteries to where they get recycled.

I want to recommend this article about "Twenty Dumb Things Organizations Do to Mess Up Their Relationship With Employees". When reading through the list, I had to smile several times, because I could remember real-life stories from my own work life as an employee. Frequently, I thought that our motivation would have been higher if the boss simply had done nothing and said nothing. We were intrinsically motivated, but management activities destroyed this motivation. Being a freelancer, I now do not need to deal with this type of obstacle. I do not need to find out how to overcome demotivation caused by the boss or by bad team work. Customers are almost always motivating. The fact that they need my results is motivating. And work is fun anyway.

However,... The article makes clear claims, but things are not so simple. For instance, Susan M. Heathfield recommends to treat all employees equally, and in the same time to not do it. When single persons make errors, do not solve the problem by defining new rules for everyone. In the same time, this is how process improvement works. You see errors been made, make a root cause analysis and solve it. But of course, the root cause can also be that single persons need a training or supervision. So, you do not need to punish the whole team for errors made by single persons, but then you do not treat them equally. Some will have more freedom than others. When I was a boss, someone complained that I do not treat the whole team equally. My quick response was: "You do not treat me equally, neither!"

Anyway, the more experience I have, the more I think that life is too complex to solve problems with single rules. In each situation, a fitting decision is needed. Rules can help define alternative solutions, but then intuition comes into the game because you need to predict what will happen when you act so or so.

I am very excited about the hitchBOT experiment: A robot will hitchhike its way from Canada's east coast to the west coast. Will people accept hitchBOT as a hitchhiker? Help it reload its batteries in their cigarette lighters? Will they have interesting conversations with the robot? We can follow hitchBOT's trip in its blog here. You can read more about hitchBOT and see its photo here. The trip starts July 27th!

Currently, I am rating a pile of student assignments. It is the fourth time I give this course. Each time, when reviewing the assignments, I identified misunderstandings of the topic and decided to tell the students more clearly what my understanding of scientific work is. Especially, I told them what plagiarism is and that I usually discover it and it will lead to a complete rejection of the assignment. Since that, I had no further case of plagiarism. Just "sluggish citations". I have just included an additional slide into my course material to discuss different ways of citing.

Now I wonder: The clearer I tell the students what I expect, the easier it is for them to do it. It is mainly a question of time invested, not of guessing my criteria. This means that the grades will improve. This time already, most grades lie on the top part of the scale. Similar effects I have observed in other courses where I learned better and better to prepare the students exactly to the type of questions I will ask in the exam.

While I think that this is fair play, however, seen from outside, it might look like I am loosing my teeth. That I do not thouroughly review assignments and do not see errors in the exams. The good grades might even look like I am trying to win students by making the course easy. Seen from outside. My students keep complaining how much work I demand from them. And this is true. However, those who work hard can be sure that they will be rewarded by good marks. And this is what I remember from my studies. There, we also had transparency about what we must learn, how it will be tested and what the criteria are.

I have just replied to a questionnaire about professional objectives. My answers must seem quite puzzling, but nevertheless, they are all true.

For instance, I answered with "very true" to "I like doing research just for gaining new knowledge" and "The results of my research must lead to products which improve peoples' lives". Well, yes, why not? Why not love doing basic research for the sake of knowledge and also practical research which really changes the world? This is no contradicting, but rather a problem with time restrictions.

And I said I am willing to take risks, and in the same time said I would like to have a secure professional future. This can both be true. For instance, I can take risks like the need to move to another country, but would like to have the security to have an income. I can risk to loose one month's work for an idea which will finally not work, but this is OK, when on the year's average, I can earn my living with the other projects.

I can also say that it is normal for me to work extra hours, but in the same time, I can wish it would be less.

In a previous questionnaire at one of my employers, I raised some (anonymous) attention by crossing "I would not apply at this company again" as well as "I would recommend this company as an employer to my friends". Why not? The fact that I do not fit in a company's culture does not mean that my friends would not fit. I had some persons in mind who would feel at home there.

I have just been correcting my students' homework about theory of constraints TOC. They had learned the network planning and critical path analysis CPA first, and now I see that I should explain the differences explicitly. This will lead to a philosophical excursion, but I am sure they will be able to follow.
TOC and CPA can be used for analyzing the same work processes, but their objectives are different:

• CPA optimizes the processe's duration. This is relevant when you produce something specifically for one customer and the duration between contracting and delivery are important. This can be a software project, but also producing an individually configured car.
• TOC optimizes the throughput. This is relevant when you produce a large number of identical products. In our exercise, we produce musical boxes. When we produce 5 music boxes per hour, we can sell this number. The customer does not mind how long the process takes. Therefore, parallelization of production steps is not as interesting as in CPA, because parallelization only saves calendar time, but not work time/ cost. Parallelization can even create cost in software engineering, e.g. when we start testing before the code is complete. We risk that code already tested is changed and must be retestet.

All together, we can summarize: Depending on which of the four corners of project manager's square (time, budget, scope or quality), you will use another method for analyzing the process, come to different results and take different measures.
For optimizing quality, of course, TOC and CPA are not made. Measuring and optimizing quality is another interesting topic but can be summarized by a simple "use standards".

In his key note talk at the Karlsruher Entwicklertage yesterday, prof. Michele Lanza talked of IDEs and software visualization. We were travelling through code represented as colourful cities in 3 dimensions (Code city). We learned that "software is beautiful". And that research is not like formula 1: "overpaid drivers running in circles, burning a lot of fuel". Research results like software visualization tools have practical relevance.
However, he was not so fond of empirical research. "Experiment is suicide", it is a waste of time to prove what we knew from the beginning. He said. I am not so happy with this, because having a proof is better than just "knowing". Researchers, too, are humans and can err. And sometimes, empirical research results can be surprising.

It happens amazingly often that you use an online system and even the standard use cases like "edit profile", "modify submission" or "delete submission" are evidently not tested. I won´t tell which conference system it is. But I am really puzzled that they evidently did not test the system. When I modified my submission, this created a new submission. So, I had to delete one of them. I got acknowledgement e-mails for all of these activities. While it is OK (although unusual) to ask me to confirm by e-mail my submission, it is strange that even when deleting a submission, I should confirm by e-mail that I will give this presentation. Ehm?