'Soft' Software Engineering: Challenges & Some Hopes for the Future

Invited talk presented at the SEmotion ICSE workshop, May 28th, 2019, Montreal, Canada.

Abstract: Within the general trend of considering more and more human and "softer" aspects in Software Engineering, there has recently been a strong push for studying the emotions of developers and engineers. In this talk, we look at some challenges when doing interdisciplinary research involving psychology, behavioural aspects, and affect/emotion of software engineers. Based on our experience from studies in behavioural software engineering, we also highlight some hopes for the future of "soft" software engineering.

Transcript

1. “Soft” Software Engineering: Challenges & Some Hopes for the Future SEmotion 2019 Robert Feldt
2. Challenge 1: Primacy of Tech Number of ‘soft’ SE papers on the rise
3. Challenge 1: Primacy of Tech Num‘soft’ SE papers on the rise (even w. normalised)
4. “Being more technical and mathematical is 'better' than focusing on 'soft' & human aspects” CHASE has been one of largest ICSE workshops for several years But someone must dare to say it: More tech/math/CS => more prestige, higher rate of acceptance, more paper awards, etc Challenge 1: Primacy of Tech
5. Challenge 1: Primacy of Tech Hard SE Soft SE “Key differentiator is Technology. Humans just in the way.” “Technology changes but humans is the core & will continue to be.” People-suck: “Make mistakes, cannot change, we work around them” Orgs-suck: “Orgs too slow to change” Change-is-hard: “People can slowly change but need tech to drive” Tech-FTW: “Tech so great, transforms the world” Tech-nerd: “Tech great for me, obviously also for others” Hardcore Soft SE: “Tech meaningless. Humans is key.” I’m an SE-Pluralist or SE-Nuanceist: “Ack the power of tech. But tech without human context & focus is rarely enough. Hybrid, multi-faceted studies & solutions needed.”
6. Think through your view on SE. Which knowledge & type of solutions will take us forward? Hopes 1: Primacy of Tech Each study: ask how contributes to your view. If not clear, how to fix it? Add more factors, consider more explanations? In SE community, how can you help promote a more dualist view (Tech&People!)
7. Challenge 2: Context & Nuance is king/queen! Cognition, Emotion, Motivation, … all inter-twined Psychology inherently complex! Many factors that inter-twine & co-depend! Need deep understanding of context
8. Challenge 2: Context & Nuance is king/queen! "The basic premise of SA is that the conditions of the situation are in the situation and therefore context as well as process should be under inquiry."
9. Challenge 2: Context & Nuance is king/queen!
10. Consider more factors and concepts. In addition to your main RQs, which contextual factors are critical? How can you collect data also on them? (Like culture & expectation variability in Linux civility discussion today) Hopes 2: Consider Context & Nuance Avoid single-message & single-concept papers. Find balance between clear / focused message & presentation and too simplistic study design. Consider & discuss alternative explanations. What else can explain this? What could have caused the engineer to act/write as she did?
11. Challenge 3: Forgetting active agents
12. Challenge 3: Forgetting active agents Human dev will/should game any “system” we set up Humans in SW dev are active and adapt! Solutions that affect their salary, career/status, situation, happiness will be “gamed”! “Gaming” signifies deliberate action but this also happens subconsciously. Anytime there is some “feedback loop” some “manipulation” likely happens.
13. If we act on this new knowledge I propose how will it affect the socio-technical system? Hopes 3: Considering humans as agents How can we introduce the solution to minimise adverse effects? And longer-term? How will they still “game” the solution?
14. Challenge 4: Solo investigators That we try to do it all without asking for help!
15. Challenge 4: Solo investigators We don’t involve social scientists or psychology researchers! Trend is positive here though, I hear!
16. Challenge 4: Solo investigators
17. How can you strengthen connections to social scientists, psychology researchers & X? Hopes 4: Inter-disciplinary Collaboration Are there funding calls favouring cross- collaboration? How can you respect the knowledge of others? Balance “tabula rasa” thinking with scientific collaboration!
18. Challenge 5: Splitting into silos Do we really need more / multiple workshops?
19. Challenge 5: Splitting into silos CognitiviSE 2020 How perception, thoughts, memory, & problem solving affects SE ICSE 2020 Workshop, Seoul, South Korea SEmotivation 2021 What really drives and motivates SW Engineers ICSE 2021 Workshop, Madrid, Spain TherapiSE 2021 Science-based treatments for SE problems of individuals & teams ICSE 2021 Workshop, Madrid, Spain
20. Be open and allowing when reviewing papers. Look for ways of broadening perspectives and accepting complementary views and concepts. Hopes 5: Unifying rather than dividing When is it time for an (open) journal of ‘Soft’ Software Engineering? (HASE!?) Submit also to the SE journals. Area is mature enough to warrant more journal publications.
21. Primacy of quantitative data & ‘objective’ methods More (SE) challenges but no time today… Replication crisis in (soft) SE Open Science with qualitative data Simplistic (statistical) analysis
22. Shameless plugs: Open Science
23. Lorenz attractor: https://www.researchgate.net/figure/Lorenz-attractor_fig1_309729952 Credit & Links to pictures Ambulance: https://unsplash.com/photos/Xruf17OrkwM Spy / agent: https://bit.ly/2QtSP2W Alex Hammond / Free Solo: https://bit.ly/2HF7NAd Silos: https://unsplash.com/photos/ZaVUNY5rHmY