DARB 1000.

UX/UI design, interaction design, reflection, design pedagogy, user research, conversation design, chatbots

DARB 1000: the designer's assisted reflection bot

Developed through teaching and research, DARB lovingly generates custom reflection prompts to help people reflect on their work — providing a sweet, supportive, and silly space for all creatives of all kinds to build introspective muscle.

2022
UC Davis Department of Design

roles
+ instructor
+ researcher
+ designer/programmer

recognition
+ Outstanding Graduate Student Teaching Award (2022)
+ Savageau Award in Design (2022)

all designers need reflection — many designers don’t like writing.

DARB exists to make reflective writing more approachable by closing the gap between hands-on making and critical reflection — helping creative people access what they think, feel, and notice.

DARB's goals

Bring writing into the design studio.

DARB is here to support design students AND educators alike in serving as an off-the-shelf writing coach, ready to help integrate reflective writing into existing studio curricula. DARB is modular in structure and can be deployed with minimal effort to support students in reflecting on a particular assignment, on their experience in an entire course, or on other key issues in design.

Make critical thinking a little sweeter.

Self-reflection can be uncomfortable at times — DARB provides a warm, welcoming, boundary-affirming, and non-judgmental space for designers to reflect through writing. DARB’s generous personality and unfailing encouragement help break down some of the emotional barriers to self-exploration. DARB is also non-invasive, allowing participants to benefit from the emotional support of an encouraging character while also being able to reflect in private.

Motivate self-described "non-writers" to write.

DARB’s unpredictable phraseology and gregarious personality reward participants who take the time to connect to the character, and randomized text strings ensure that a participant’s experience is never the same twice.

Encourage exploratory, multi-scalar reflection.

DARB supports generative, divergent thinking as well as focusing, convergent thinking by helping designers focus their thoughts inward or outward.

process

teach course on creative reflection → analyze student reflections → program bot to promote the most useful reflections → user-test the bot for validation

DARB started out as a class — a course called Documentation for Designers that I developed and taught to inspire experimental self-reflection through media and writing.

The central premise of Documentation for Designers (as well as the rest of my MFA work at UC Davis) was this: what insights emerge when we turn design methods around on ourselves?

Documentation for Designers led students through a series of media-based inquiries and reflective exercises. Each exercise asked students to capture their creative process in a new way (through photos, audio, video, writing) and to reflect upon how the documentation process impacted their perceptions of their work. As students progressed through the class, they reflected on their work from all angles — gradually combining and recombining media and writing from earlier assignments into new, increasingly complex forms.

(More information on this course and the theory behind it is available on my public interactive thesis site.)

qualitative research

understanding what reflections worked best (and how)

Documentation for Designers wasn't just a class — it was a learning lab. I dug into multiple rounds of grounded, qualitative analysis of student assignments from the course to better understand the mechanics of what led to the richest reflections.

While my initial inquiry was intentionally open-ended, I particularly hoped to answer the following questions:

what helped students access meaningful reflective moments?
what reflective moments were most conducive to growth?

Sampling + data

I sampled the complete collection of written reflections that students turned in alongside each of their recipes. Documentation for Designers enrolled 18 students in the class, all of whom completed eight written reflections alongside their assignments, which meant that I analyzed 144 short pieces of reflective writing altogether.

Analysis protocol

I began with a largely grounded methodology: a round of close reading and open coding that stayed very close to the text. My second round of coding was largely utilitarian — bundling quotes by type (e.g. calling out when students gave direct feedback on an assignment, when students disclosed some type of emotion, when students had some type of realization, etc.). Subsequent rounds of coding were more thematic — subdividing existing code families (like direct commentary and insights) into more targeted groups (direct commentary on class assignments and insights about growth). In the final stages of analysis, I finally did allow my theoretical work to influence my coding as I started asking more pointed questions about specific types of reflections and insights that students were writing about, and on the specific pedagogical mechanisms that seemed to provoke them.

Coding schemes

In analyzing the data, I coded across a number of axes in order to track both the efficacy of the course in general, as well as more nuanced relationships between course content and student reflections. The following groups represent the major thematic axes of my final analysis.

Reflections and insights: moments where students demonstrated reflective activity, or had a realization (these things overlapped significantly, as many reflections led to insights — however, I felt it was important to track the thought processes that led to realizations alongside the realizations themselves)
Pedagogical mechanisms: since each assignment included a number of different activities and components, I subdivided all assignments into core pedagogical mechanisms — this allowed me to track how students responded to particular course activities with greater fidelity, and to notice how different aspects of my pedagogy influenced their reflections
Feelings: I kept track of content that read as emotionally laden in some way, and I loosely broke those moments down into positive, negative, and neutral categories
Commentary: I kept track of moments when students offered direct feedback on some aspect of the course or assignment (e.g. I liked xyz…)

findings

the importance of welcomeness, wideness, and words

I found that while it takes a lot going on behind-the-scenes to get to a good reflection, the good reflections can get people to all kinds of places.

Throughout analysis, I paid special attention to moments of insight, where students came upon new ideas or ways of seeing their own work. In tracking these moments against different variables and codes from throughout the course experience, I arrived at three major takeaways for the things that stimulated deep, productive reflection for students:

Safety & welcomeness are non-negotiable.

When it came to direct class commentary, students were overwhelmingly appreciative of the classroom culture of openness, and the course’s emphasis on compassion and emotional safety. Many students appreciated the repeated reminders to prioritize experimentation and play over outcome. The overall whimsical “vibe” of the class seemed to figure greatly into its success — more or less unanimously — and students found the experience to be refreshing, grounding, and affirming overall.

Useful reflections happen at two scales: process and practice.

In analysis, I began to notice a distinction between reflections on process (working an individual project) and reflections on practice (reflecting an entire body of work). This means that students jump between scales in their reflections, both making sense of an individual assignment (how it feels to use a camera in a particular way) and their entire body of work as designers (what they learn about their work through a particular style of photography). Intentionally harnessing the power of these multiscalar reflections could help students focus or widen the beam of their reflexive focus in ways that lead to the right kind of insights at the right time.

Writing is particularly helpful (even for "visual people").

Another interesting takeaway when it came to reflections that led to the richest insights was the runaway importance of word-based assignments. On the whole, assignments that incorporated some degree of play with words (whether through actual writing or activities like audio journaling, card sorts, and collages) generated the most reflective activity and the greatest density of insights overall.

conversation design

building the best parts of the course into a bot that would always be ready to serve up a designer's dose of documentation support

Not just any robot could get designers to write (and surprise themselves) — I needed to replicate the best parts of Documentation for Designers in a lean, interactive prototype buildable by someone who barely knows how to program (hi).

Enter DARB — a lo-fi, bespoke prompt-generator with a large, chaotic heart.

The goal was to create a standalone tool that students could continue to use to inspire reflection outside of the classroom. It was clear that this tool had to accomodate the three main takeaways from the preceding research, namely 1) the importance of warmth, welcomeness, and unconditional positive regard, 2) the fact that reflection can happen across multiple scales, and 3) the fact that that writing needs to be as rewarding and painless as possible.

To provide welcome, I leaned into character design and created DARB as an overly accomodating and clumsily erudite personality — genderless, bodiless, comprised only of punctuation, and yet cheerfully and rhetorically expressive. Enthusiastic and competent enough to be taken seriously, but in a way that is the opposite of intimidating.

To incentivise use (and re-use) despite a user's potential resistance to writing, I programmed DARB to serve up an ever-changing array of possible interjections and prompts via randomized text strings. Each time someone talks to DARB will yield a slightly different experience, rewarding those who choose to return.

To help users navigate reflection across different scales and target their writing to their own needs, I developed the following DARBitecture to elicit what type of reflection would be most useful to a person in the moment in order to serve a prompt that would either help them expand or focus their thinking.

The DARB program has two key features: a diagnostic section where participants self-evaluate their needs, then a writing section where DARB generates a semi-randomized custom prompt based on user inputs. (DARB is built in Twine — initially designed for interactive fiction writing.)

Introduction + conversation domain setting

DARB introduces themselves and gets to know the participant’s name and mood in order to offer a personalized experience. The participant also indicates whether they would like to focus their reflection on a specific project, or their entire practice.

Example: Right now, would you like assistance in reflecting on a specific project, or your creative practice in general?

Divergence <> convergence calibration

These are three questions to calculate the degree to which a writing prompt will provoke divergent reflection, or inward introspection. This resulting divergence score (on a scale of one to five) determines which prompts the participant will ultimately see — prompts are broken down into “buckets” according to each level of the scale.

Example: Today, are you looking for groundedness or excitement?

Adjective inputs

The participant receives five text input prompts (drawn at random) to solicit five adjectives that describe each of the following categories: inner child, creativity, style, self-image, and beliefs. Some combination of these adjectives will ultimately resurface in the writing prompt.

Example: As a child, your favorite outfit was very…

Prompt & freewriting section

DARB generates a writing prompt based on the above inputs and facilitates an in-browser writing session. After completion, the participant can choose to generate another prompt, or to exit the program.

Example (low divergence): Consider the following words: [$innerchild], [$creativity], and [$self]. Which of these words best describes the way you work? Which of these words least describes the way you work? Why? If it’s hard to decide, keep asking yourself why. Arrange all three words in order of importance, and journal about how you decided.

user testing & validation

putting DARB through their digital paces

It was time for other designers to meet DARB. I set up a series of user-tests to collect overall impressions of DARB and to identify areas to improve or iterate on further.

I kicked off a small study of unobserved user-tests followed by semi-structured interviews. My goals in this preliminary exploration were mostly to establish proof-of-concept for DARB’s character and conversation design, and to determine the overall feasibility of a chatbot writing coach.

Concept: are designers open to letting a chatbot facilitate reflective writing sessions? What are users’ experiences with the program like?
Character: what about DARB’s character enhances or impedes the writing or reflection process?

Recruitment & sample

I recruited a small number of participants via a single posting on a personal social media profile. Participants did not know in advance that they would be interacting with a chatbot — they only knew the project had something to do with reflection. I interviewed 10 participants — design students, practicing designers, and one musician. In order to be eligible, participants had to have some sort of active creative practice (though I did not specify any constraints regarding discipline). The average age of respondents was 25 years old — a little older than the average undergrad.

Activities + interview structure

Due to the intimate nature of journaling, I did not observe participants interacting with DARB live. After participants had time with DARB, I met with them live to ask questions about their experience. I primarily used these interviews to solicit feedback on DARB’s overall concept and structure, rather than on discrete UX/UI decisions (though those issues did end up surfacing in the interviews). Interviews were 30-40 minutes in length.

Sample interview questions

Off the top of your head, what stood out to you most about your experience with DARB?
How do you describe your everyday relationship to writing?
How do you feel about the writing you produced with DARB?
What were the main takeaways or surprises about your experience with DARB?

Skew: process vs. practice

Part of DARB’s programming is to ask whether a user wants to reflect on a specific project or their entire practice — this distinction affects the types of prompts they will eventually see. I recorded participants responses to this question in order to calibrate my findings accordingly. An overwhelming majority participants opted for a broad, practice-level reflection.

findings & next steps

a DARB for (almost) every occasion

Despite a few technical bugs, everyone got along with DARB. Early testing was supportive of the approach in general, opening up new possibilities for additional DARBs to be deployed to help people reflect in different kinds of settings, from classrooms to exhibitions.

The response to DARB’s character design was unanimously positive. DARB’s overall friendliness and warmth seemed to help many participants feel comfortable in their reflections. An overwhelming majority of participants spent more time than necessary in getting to know DARB through the optional trivia series and exploring different avenues of conversation. A number of participants also chose to generate more than one writing prompt.

The greatest takeaway from these tests was that many participants benefitted from the involvement of a character (rather than filling in a static form), but also benefitted from not being surveilled by another “real” human being. Many users commented that opening up to DARB was easier than opening up to a real person. The fact that DARB was only a bot seemed to be much more of an asset than a liability, as it let people write freely without fear of human judgement.

Since this initial prototype, additional versions of DARB have been created to support students in specific design classes and to guide visitors through my own MFA exhibition at the Manetti Shrem Museum at UC Davis.

User quotes: character design built rapport

In a class, it could be a fun way to replace a discussion post. Could also be a fun way to enjoy writing.
Really appreciated that there were certain indicators for taking pressure off — no pressure language. I liked when the bot said they were incapable of resentment.
They have a personality, but you also know that you're not talking to a real person, which means you can be more authentic. It's cool because you get the positive aspects of being human without judgement. Kind of like talking to a therapist.
My initial reaction to something like this is to picture it like an artificial customer service bot. Seems like you can game it, manipulate it. But this was different… There's little bits and pieces along the way that distracted me from that. Like the poem, and the option for music — it made it different than other bots.

User quotes: lo-fi approach had some limitations

The first half of the prototype (short questions) didn't quite feel connected to the back half of the prototype (the writing prompt). The adjectives that came up in the prompt seemed sort of vague.
Really liked how the instructions type themselves out — forces you to slow down.
In the adjective answer part — the fact that there was an adjective already in the text field made me biased toward providing a synonym to the pre-existing adjectives.

User quotes: writing experience led to new insights

Surprising to see my own words from one context put into a different context — made me think of those concepts in a completely different way.
Felt great about the writing. It was fun how there was an innocence to answering the early questions without knowing what was coming. When the word inputs came back in the prompts, it felt accurate to things I should be thinking about. Prompt was really nice and helpful — hadn't asked myself these questions in that wording.
Liked the way the choice adjectives work with the prompt. Still open-ended, but they kept you structured on the main point. The questions are really thoughtful. Some of them got too deep, like the jealousy one.

wanna meet DARB?

don't be shy — they love to chat

they love to chat!

more work

Public Service Sampler PackDesign

Headspace HighlightsDesign

The Communicating Pain ProjectDesign

doing something interesting? want a thought partner? need to consult on methods? how much do you like sticky notes? how about sharpies? more venn diagrams? need a pretty slide deck? how about a game?doing something interesting? want a thought partner? need to consult on methods? how much do you like sticky notes? how about sharpies? more venn diagrams? need a pretty slide deck? how about a game?doing something interesting? want a thought partner? need to consult on methods? how much do you like sticky notes? how about sharpies? more venn diagrams? need a pretty slide deck? how about a game?doing something interesting? want a thought partner? need to consult on methods? how much do you like sticky notes? how about sharpies? more venn diagrams? need a pretty slide deck? how about a game?doing something interesting? want a thought partner? need to consult on methods? how much do you like sticky notes? how about sharpies? more venn diagrams? need a pretty slide deck? how about a game?doing something interesting? want a thought partner? need to consult on methods? how much do you like sticky notes? how about sharpies? more venn diagrams? need a pretty slide deck? how about a game?

trace manuel.

I'd love to hear what you're thinking:
tm@tracemanuel.com
or send a booking inquiry