Experimental culture is a way of thinking; it’s about trying new things and learning together, solving complex software problems and creating value together. According to Terhi Aho, experimental culture in software organizations requires strong management support and psychological security.
Terhi Aho gave a talk on the culture of experimentation at work at ScanAgile 2023.
A culture of experimentation prevails in software organizations if an experimental approach is consistently applied, as Aho explained:
The fact that a unit in the organization carries out one or several pilot projects per year does not mean that there is still a culture of experimentation. But of course, every pilot helps to build that culture.
Working together, with a partner, in a small or larger group also brings a kind of security. However, it is good to be aware that the size of the group involved in the experiment can affect the speed with which the experiment can be carried out, Aho mentioned. Implementation requires that everyone understands what is being done. The larger the group, the more time needed for discussion, she said.
Joint experimentation and learning requires dialogue, Aho mentioned:
First, the challenges that need to be addressed are highlighted. Together we can come up with ideas on how to tackle these challenges. When a way is found that seems possible, it is jointly decided to try to solve it.
The schedule is determined and after the experiment we evaluate the result or effect, said Aho. The decision to experiment and then evaluate whether it is worth continuing removes the need for the experiment to succeed. If the idea doesn’t work, it’s not a shame for anyone, it’s just a learning experience, Aho noted.
Aho mentioned that management support ensures that experimentation is encouraged under conditions of uncertainty and that the necessary investment decisions are made to ensure the availability of time, money and skills. Management can also provide justification when work is questioned and remove barriers to experimentation, she said. Possible barriers in software organizations may include current processes and policies, lack of decision-making power, and opposition from individuals in leadership positions.
Psychological safety means that people dare to think out loud, reflect, ask for help and challenge, experiment, celebrate successes and share failures, Aho argues. Psychological security is built through action; it is not enough to say: “We allow experimentation”. What matters is how experimentation is encouraged in everyday life in software development and how the results of experimentation are perceived, as Aho explained:
In particular, what to do when the results of the experiment do not meet expectations. Is there a genuine willingness to learn or is it expected that all experiments will be successful?
A better understanding of the user case and user preferences will help focus development efforts on the most relevant features, Aho said. With experimentation, development is done in small batches and tested with customers, and the need for recoding is reduced.
Fast test cycles allow testing completely new solutions. Developers have the opportunity to build something that has never been seen before and act as pioneers, Aho concluded.
InfoQ interviewed Terhi Aho about experimentation in software development.
InfoQ: What role does psychological safety play in encouraging experimentation in software organizations?
Terhi Aho: In a climate of psychological safety, people actively experiment and think. New solutions and innovations are emerging.
If there is no psychological safety in everyday life, people will try to ensure that the outcome of the experiment meets expectations. In this case, they actually plan the implementation of the experiment so thoroughly that it is no longer an experiment, but a planned project. The final result may not be the best possible, but it is in line with expectations and therefore safe.
InfoQ: Why should we be prepared for experiment failure?
It’s me: It is dangerous if, at the beginning of the trial, there are already plans for further development and perhaps the launch of the software service being tested. In this case, the idea being tested is assumed to work for sure and is already invested in the future.
If the idea turns out not to work, planning ahead and preparing for the future will incur costs in any case. This will increase the total cost of the experiment and reduce the willingness to decide to terminate the experiment.
InfoP: What did you learn by performing the experiments?
It’s me: Experimentation can be applied to any type of organization, not just software. I feel like I learn something new every day through experimentation. Over the years I have learned to constantly experiment in small ways. Experimentation has become a kind of routine. As it became a habit, my attitude towards failure and messing up in general changed. I can honestly consider them learning situations.
When I make a mistake, it can be embarrassing at first, but the feeling quickly passes and I can focus on the lessons learned. This has helped me focus on moving forward, learn a lot about myself and my ways of doing things, and share the lessons I’ve learned with others. I claim to be easily adaptable and resilient thanks to a routine of trial and error.