Making recommendation systems in apartment management platforms visible to domestic help: outcomes & output
Based on the learnings from our previous stagings (read about it here) we revised our approach in engaging with the domestic help. An important learning was that the workplace and work hours were not suitable spaces for conducting our research. We decided to engage with the domestic help outside of this context, possibly at her residence and after her work hours. Instead of having a single activity of introducing the artefact and observing the interaction with it, we broke it down into a series of activities that were designed to incrementally build up a shared understanding of rating and reviews for the participant as well as for the researchers. The activities aimed at aiding the imagination of the domestic help, to make visible the concept of rating in digital platforms and provoke her into thinking of the various ways in which platform discriminates against her.
Day 1: Informal meeting with the participant
We found another resident in the apartment complex who was willing to introduce us to their domestic help. But this time instead of talking to her at the apartment and/or interviewing her we met her after her work hours and had a simple conversation about our research goals, keeping details at a minimum. We sought her help in understanding how the apartment management platform could be better designed for her needs. The participant warmed up to the idea of working with us and we made sure to follow up the next day at her residence after work hours.
Day 2: First Activity — Mapping the Good and Bad
With the help of the participant we mapped out what a good and employer means to her. The mapping essentially encapsulated various stories of houses she had worked for over the years. We sketched these stories out on a large chart paper keenly noting down details of her emotions, the objects, the attitudes and words she used to describe certain behaviours of her employers. We followed it up with a similar exercise but for domestic help (good vs. bad). It was interesting to get an insight into her day-to-day interactions and this also worked towards building a camaraderie between the participant and researchers.
Day 3: The Rating Activity
Based on the stories we collected the previous day we designed an activity to demonstrate the idea of rating people, a concept our participant was unfamiliar with. Using boards and magnets to signify spaces (kitchen, hall and outside the apartment) and characters (the resident, the help and other miscellaneous characters) we staged scenarios derived from the stories she shared with us. At the end of each performance we would ask her rate the characters (1 to 5 stars), share with us the rationale behind the rating and then tell us how the character could improve their rating.
Day 4: Using Short Film as a Method to provoke Reflection
During our previous visits to the participants house we noticed she would religiously watch Tamil movies after she got back from work. This gave us the idea of using film as a medium to convey our politics and reasons for problematizing the current implementation of rating systems in these platforms. The film (which can be viewed here) is a dystopian imagination of what the life of a daily help would look like in 2030 if her concerns were not included in the design of apartment management platforms. The medium also increased the make-believe quality of our research artefact (the bangle) by allowing us to superimpose the visual display of rating using special effects to show what the rating looked like and how it gets updated.
The short film was centred around rating and divided into three parts with each section showing a progressive decline in the rating of the protagonist (a domestic help) due to circumstances within and beyond her control. We followed up each story with a short round of questions around what she felt about the situation and what she would want to change given the power to do so. By the end of the screening we saw a visible difference in how she had understood the rating system and what it meant for her future. She spoke more animatedly about what she felt about the system and how she wanted to now see it for herself. We consider this demand for transparency as a considerable progress in our research work.
Observations and Insights
Besides the demand for more transparency of what goes on in these platforms there were many other themes and ideas that we could glean from the conversations on the fourth day. We also expressed these insights in the form of a zine (shown as comic panels in this report) where the bangle is the domestic help’s sidekick and goes around saving the day!
The platform chose to quantify her work in ways that she did not find valuable. For example, she felt that it would have helped if, besides tracking her in/out time, the platform also measured how she went about compensating for the time lost in arriving late by doing her job more efficiently and quickly.
By accessing the platform the domestic help hoped that it would enable her to do the following -
- Take ownership of her availability for work (setting and removing slots)
- Be able to access to data points relevant to her — her rating and reviews
- Search for lucrative employment opportunities (houses that pay well)
- Search for safe and considerate places to work (a good household)
An apartment is not a single uniform entity. Various actors within it introduce and quell anxieties of the domestic help. The injustices imposed by one actor is sometimes corrected by another and the domestic help is able to negotiate her way through some tricky situations. For example, in one household there was a lady who would make the help do more work that she had agreed to but the lady’s husband would sometimes intervene and take the help’s side. In another household there was an old lady who would abuse the help while her daughter-in-law would step in and protect her. There’s a constant negotiation of responsibilities and expectations and it is worth noting that this makes it a lot more complex to assign a single rating to a household.
The platform shows no concern for the privacy of the personally identifiable information of the domestic help. She expressed how she was uncomfortable that her information was on a digital platform that could be access by anybody with a phone. She was particularly concerned about the security staff, often comprised of young men, having access to her photo, phone number and address since they probably remembered her 6 digit code.
The lack of an official contract and not being able to enforce it results in a frequent exploitation of domestic help where they are made to work extra hours and do tasks that they don’t get any additional pay. This could be a possible area of intervention for tech and labour unions.