Previous Chapter
Case Study
← IndexConclusions, Next Steps, Reflection

Chapter 8Conclusions, Next Steps, Reflection

This thesis set out to answer two main questions relating to flow in digital tools. Through a three-phase process the flow phenomenon and digital tool design were explored, a model was conceived and tested, and the finalised model was applied in a design project. This chapter will summarise findings and discuss the results.

8.1 How flow works

The first question posed was "How does flow work in digital tools?".

In a nutshell, the theory on flow learns that it occurs when your perceived skills and the perceived challenge are in perfect balance. This balance, combined with clear goals and immediate feedback, will cause emergent motivation.

Flow remains an elusive phenomenon that is by definition subjective and fleeting. Not only is it difficult to measure — most techniques will disrupt the very feeling they're trying to capture — but the theory on flow (digital and physical) also remains unavoidably vague about any applications of its ideas.

Flow theory has been developed with a focus on physical activities and they remain mostly unchallenged in the realm of digital tools. An interface that matches the user's current skill level with the appropriate amount of challenge, and allows them to develop their own proximal goals, will be well on its way to allow for flow to occur. But the exact pathways through which an interface causes flow experience are still unclear.

We must remember that the interface itself doesn't cause flow. Flow is an internal, subjective experience that stems from the user's own (creative) process that causes emergent motivation. The interface's job is to merely help this process along – to facilitate the user in reaching flow state. The user should be free to explore the interface on their own merits, with a level of granularity that is understandable yet challenging to them.

Within the confines of flow in digital tools, we have defined a tool's job as helping you do work. In its most abstract definition "doing work" can be defined as making decisions and carrying out those decisions. Over time you will become more experienced and thus able to make faster and more complex decisions. A flow-inducing digital tool should scale with you along the way, to allow you to make more and more complex decisions.

All in all, this is a tricky balance to get right, and there still aren't any concrete rules to adhere to when creating interfaces. This provides a clear opportunity for further research.

The practice of game design has already made a lot of strides in this area, from adjusting the difficulty of a game dynamically based on player performance (Zohaib, 2018) to relating multiple psychological concepts together to form a presence-involvement-flow framework (PIFF) in games (Takatalo et al., 2009). In my research I took some lessons from game design, relating play with flow, but these two worlds can be related to each other much more. For one, the similarities, differences, and applicability of the Flow Funnel to games and the PIFF to digital tools could be investigated.

The skills/challenge balance needs to be researched further. How do we measure a person's skills, factoring in the importance of perceived skill level? In a similar vein: how do we measure, concretely, the perceived amount of challenge an interface poses? Visual complexity will most likely factor into this, but it will not be the only metric. Task completion speed, which can already be measured (albeit in privacy-sensitive ways), might tell us more about how difficult it is to work with a tool.

In addition to measuring the characteristics of an interface, there is a need for better measurement techniques for flow state itself. In my research I attempted to use the Flow Short Scale (Kyriazos et al., 2018) as an objective verification of the qualitative interviews. This inclusion took a lot of time, participants often needed clarification on what the questions referred to, and all in all not much could be gleaned from the data.

It's difficult to envision how we might measure flow in objectively and quantitatively without disrupting flow. While some progress has been made using physiological measurement techniques such as EEG (Katahira et al., 2018) and GSR (Nacke and Lindley, 2008), these methods are far from perfect for measuring emotions and flow (Barrett, 2020). The search for new techniques must continue. Once we can measure flow as it happens instead of retrospectively, a whole slew of dynamic interface adjustment techniques could become possible.

8.2 Designing for flow

The second question posed was "How might we create interfaces that cause flow?".

As mentioned above, flow is experienced internally and is difficult to measure objectively. This makes it very difficult to gauge the "flowiness" of an interface. It's not easy to say which of two interfaces is the more flow-inducing one. Quantitative measurements — which often rely on tracking a user's every minute behaviour and are themselves a privacy nightmare — might provide reports of user engagement, but user emotion and user flow are hard to record — and even harder to predict up-front during the design process.

In absence of concrete implementations of flow, but with the possibilities of digital technology ever-increasing, many digital tools have focused on optimising for the metrics they can actually measure: user engagement, the speed with which they work, the amount of times a feature is used. This has a number of adverse effects. It foregoes expressiveness and creativity. It results in interfaces including buttons for the sake of buttons, without looking at the workflow they contribute to. Going even beyond digital tools, it results in software engineered to grasp our attention and keep it captive.

Our interfaces are directly influenced by our design process – our tools, in turn, shape us. The under-appreciation of a user's experience of flow leads to an underemphasis of causing flow in our digital tools. Thus, there is a clear need for a mental model of flow in digital tools, to aid designers in creating flow-inducing interfaces. The Flow Funnel aims to be that model.

The Flow Funnel is based upon the previously mentioned skill/challenge balance needed to reach flow state. Every application user has their own (perceived) skill level. An application's interface should match the user's skill level by adapting the amount of possibilities for action – simple interfaces for novice users, complex interfaces for expert users – and scale with the user every step of the way.

When designing a flow-inducing interface, we must be aware of this progression, and we must cater our interface to each of the skill levels – while still being familiar at all times. This will result in different interface paradigms and conventions being used at various levels.

  • Novice users act like Directors, telling the computer to do what they envision through a dialogue of interaction. They use few manual settings, but that doesn't mean they can't express themselves efficiently or creatively.
  • Proficient users assume the role of Makers, growing towards more and more independence from the computer and using informed experimentation to push what's possible.
  • Expert users, as Architects, have a deep familiarity with the interface and create their own systems within it, pushing the boundaries of the interface or even going beyond with their own templates or scripts.

The interface should smoothly scale its complexity up and down, ideally without switching modalities or concepts, as its various systems interlock.

The Flow Funnel is meant to be a mental model for designers, allowing them to think about their interface as not just a pretty & efficient manifestation of functionality, but rather as an ever-changing landscape of features through which users journey – each of them plotting out their own path and venturing further out from home. The Flow Funnel is meant to be used as a map of this landscape. It highlights which parts of the journey are unable to be discovered, or which paths might get congested from the amount of traffic they have to service.

Any map is an abstraction of reality, and the Flow Funnel is no different. It might even be argued that it is an over-abstraction.

For one, not all users will fall neatly into the director–maker–architect roles. Users who are experts in one domain might expect and demand similarly complex interfaces in domains they are less knowledgeable about.

Secondly, the model raises an interesting question around creation and creativity. When every user can use a similar interface to create something (whether that is an art piece or a spreadsheet) regardless of their skill level, does that reduce the value of works that were produced manually? To what extent is the creator "part of" the outcome? One might argue that novice user's creations are more like commissions from the computer, whereas proficient and expert users create something themselves. This question will become more and more prevalent as computers start becoming more and more creative, next to their analytical skills.

Finally, not every tool needs to be all-encompassing: there is still a place for quick utility tools that don't scale the entire skill-level spectrum. Some tools can be beginner-only, only can focus on experts, etcetera. But where and how does one draw the line?

Furthermore, questions around implementation of the model remain unanswered. Many different paradigms have been introduced for each skill level, but the exact way to integrate these paradigms into an application, and the ways they should flow into each other, is still up to the individual designer to figure out. Some of the existing trends in digital tools, discussed in the Literature Review, already point in the right direction. These could be examined further, placed into the Flow Funnel.

While some objective measurements can and should be used – tests of usability, accessibility, probability of use, and more – at the end of the day a designer needs to trust their intuition, listen to users with the greatest respect and empathy, and carefully select what features make sense and how they should be implemented.

It's not unusual for people to tell us that our app does way less than our competitors – to which we reply "that's right, you're welcome".

Jasper Hauser

Design is not an exact science because humans aren't rational creatures. While it may be difficult to produce, further theory around the "irrational" experience of interfaces, along with studies into the influence of models and tools on design processes would be greatly beneficial to the field.

What's important is listening to and understanding people's needs. We're driven by our needs, not by our reason. That might cause some products to look a bit strange, either from a creative- or a reasoned perspective. But in my experience, that balance is exactly what works best because you play to people's needs.

Jasper Hauser

All in all, the Flow Funnel still has a lot of questions surrounding it. Despite all of these unanswered questions, the proof of the pudding will be in the eating. So long as this model can help a designer think about their designs in a more cohesive fashion, so long as it helps them discover usage patterns they didn't cater to before, so long as it feels like a valuable addition to their toolkit...

It will have been worth all the effort.

Reflection 8.3 Final Thoughts

This thesis started from a fascination for the idea of an invisible interface. Many of my designs have been characterised by working around the user, as I believe users are not interested in my designs themselves – rather, they care about the end result, and my designs are just a means of reaching those ends. What would happen if I extrapolated that idea of invisible technology to be the entire basis of an app's interface? Could an interface be so advanced that it could predict every next action, so that you wouldn't ever have to switch modes and could just hit the "Next action" button continuously? This would provide you with the ultimate flow experience, as you'd never be distracted by the interface itself... right?

At the end of this thesis, it turns out that that idea of an invisible, all-knowing interface is missing the point entirely. Flow stems from the creative process — the creative struggle, even — of shaping something while using a tool. An invisible interface that works around you actually robs you from any flow at all, as it is the activity of interacting with the interface that creates the flow. Instead, the perfect interface is one that feels invisible – because it is so attuned to your workflow that you can trust exactly how it will behave.

Looking back at my process, the improvements I'd make can be summarised by the words "bigger and better": I would have included even more time for literature review, as I kept adding to it in later stages as I found out new things. I would have interviewed more users, as that would have validated the qualitative statements further. I definitely wanted to interview more experts, as those were some of the most profound conversations I've had.

And yet, I can't be much prouder of the project I've ran. Hardly ever did my motivation wane, as I looked forward to learning more about this topic every week. I believe it is this enthusiasm that let the process steer itself so well. As I was warned at the beginning of this project: "Flow is a bit of a honeypot for researchers", and yet somehow I managed to keep it within scope by sticking to my three-phase process and deciding to focus on developing a model early on.

And of that model, the Flow Funnel, I am the proudest of all. Not only of how tangible a result it is (one that I will carry in my designer's toolkit for years), but especially of how people other than myself saw truth in it, and could recognise themselves in its portrayed journey.

There are a lot of refinements to be made and questions to be answered. But on the path towards a comprehensive understanding of flow in digital tools, I see this as a very valuable first step.