The digital economy (e.g. online shopping) is growing rapidly – already £50 billion in 2016 – this has been especially true during COVID, with a growth of online retail of over 30% this year alone. With new `try before you by’ clothing services such as ASOS and Amazon Prime Wardrobe, where any number of clothing items can be ordered for home delivery, and returned ‘for free’, the true environmental costs and impacts on workers are entirely hidden from the consumer. These services have heralded the growth of the platform economy, where an army of gig workers compete for highly variable rewards, and bear many of the infrastructure costs that once would’ve once belonged to an employer. In the flipgig project, we are looking at the role of digital services in this growing workplace and how these can be better designed to empower couriers to fight unfairness, challenge unfair models and algorithms in platform courier work, and develop new models that put fairness and sustainability at the core. In this talk we report on our fieldwork and give a voice to gig economy workers, identifying multiple forms of systemic and unintentional bias arising from being ‘behind the app’.
1. Behind the app: perspectives on fairness
and sustainability from the gig economy
Adrian Friday
Prof. of Computing and Sustainability, Lancaster University, UK
+ Oliver Bates and Carolynne Lord (Lancaster); Tom Cherrett, Fraser McLeod and
Andy Oakey (Transportation Research/ U. Southampton); Antonio Martinez-Sykora
Southampton Business School; Ben Kirman (York)
2. Thanks for the funding!
http://www.flipgig.org, FlipGig (2019-2021) EPSRC
grant agreement no. EP/S027726/1
FlipGig has collaborated on fieldwork with the
pilot project Switch-Gig funded by the EPSRC
Network+ Not-Equal (EP/R044929/1).
3. Why should
Computer
Science research
care?
• Our systems are at the scale they have impact on the world
• And the workers are implicated behind the app screen
• Clean interfaces and intuitive UX design hide ‘how this happens’ and ‘what
it’s like to work behind the app’
Image: unsplash/charles-deluvio-6g7K1idhMJw
4. Questions
How fair and sustainable is the gig
economy?
Do emergent properties and algorithmic
biases lead to unfairness and injustice?
Are there systemic factors that drive us
towards or away from environmental
sustainability?
How could these platforms offer more
socially, environmentally and financially
sustainable work?
5. Food Freedom
• Deliveroo's mission is to become
the definitive food company. It's
this ambitious goal that fuels
constant innovation within our
company; offering new and
exciting selection to our hungry
customers while being able to
offer our riders flexible work and
ongoing support to the
Australian restaurant industry.
[link]
https://foodscene.deliveroo.com.au/restaurant-profiles/australia-launches-food-freedom.html
7. What is the gig
economy?
1. Workers in the gig economy commonly take on
work that is mediated through digital platforms
2. Platforms match workers and clients in the
performance of short-term or individual tasks,
colloquially known as ‘gigs’ (Woodcock and
Graham, 2019)
3. Pay is linked to the number of ’gigs’ completed
Image: unsplash/mak-1uDgb-65_28
8. Why are gig
workers
attractive to
businesses?
• “Independent contractors” are effectively
self-employed and are not ‘company assets’
• Are ‘on demand’ labour
• Workers own their own infrastructure (e.g.
transport, maintenance costs, insurance)
• Pay no benefits such as holiday, sick pay,
maternity/paternity leave, pension
• Are self-employed for tax purposes
• Make a healthy commission (Deliveroo take
35%, (Shead, 2020))
• Plus get to monetise all that yummy data!
9. Digitisation • This form of working has expanded greatly
• Now an estimated 50 million gig platform
workers worldwide (Fairwork, 2020)
• An estimated 4.7 million in the UK (9.6%
of working-age adults)
• Increased 100% over the last three years
(TUC, 2019)
• This is more than just food, also
babysitting, graphic design and cleaning
services
• COVID has led to a digital transformation
with a 30% increase in online shopping
10. Dynamic human
infrastructure
• Then there’s all that ‘trip
generation’
• Different modes of transport
have different environmental
impacts
• Congest our cities in unknown
ways
• Cause possible pollution
directly and indirectly
• Or, could help ease these
impacts by ‘reshaping civic
infrastructure’
Images: unsplash/stephan-schmid-0MtjR0BvV6A/eggbank-4w3UZBFDacI
13. Who does gig work?
• This varies, but noted French authority Laetitia Dablanc’s survey of gig workers in Paris
(Dablanc 2020)
– More than ‘a convenient top up’, 73% full time gig workers
– 2% female; 14% French; 9% students
– 31% use a scooter (although this isn’t legal!)
– 16% use shared bikes
– 10% e-bikes
14. Enquiry 1 – Cycle
courier workshops
• Switch-Gig funded by the EPSRC
Network+ Not-Equal (EP/R044929/1) -
workshops in Manchester and York with 8
cycle couriers
• All participants were male
• 20-30 years of age
• 7 white
• 7 native English speakers
• 3 were students
• Several riders dropped out (including
two female riders) due to concerns over
COVID-19 and travelling to and
attending from Scotland
15. Why do these
people do gig?
• Love of cycling (50 miles a day is
common)
• Can do better than basic living wage
(some earn £15ph vs. £8.21)
• Mixture of money, adrenaline and
endorphins
• Contrast with Paris where easy access
to work for immigrants
Image: https://www.facebook.com/yorkcollective/
16. A day in the
life of a gig
worker
1
Commute
into
zone
2
Wait
for
job
3
Accept
job
4
Go
to
restaurant
5
Wait
for
food
6
Travel
to
customer
7
Deliver
17. Pay
Waiting &
travelling to the
next pickup =
Unpaid time
Delivering = Paid
time
Image: unsplash/matheus-bardemaker-RwoXb6lk7rA
18. Being successful
= taking risks?
• Tacit knowledge, second guessing the algorithms
• Going where other vehicles shouldn’t
• Taking risks daily (c.f. Gregory, 2020)
• Knowing which job not to accept
19. Work as a game?
• Unlike “real” independent
contractors, earnings depend on
the availability of sufficient work
• Pay and reward is linked to the
balance between urgency,
available work, available workers
• Peak and super peak times
• Refusing a job, can ‘encourage’
the reward to increase
20. Enquiry 2 -
Understanding
challenges
• Adapted the ‘Critical Incident Technique’ to reach
this hard to reach population
• CIT normally probes problems with the flow in
user interactions with a UI
• Here adapted to focus on challenges faced
‘behind the app’
• Snowball sampling, n=26 so far
• Responses broader than just the algorithm:
23. A variable
slice of a
variable
sized pie
• The amount of work is ‘on demand’ and shared
by those it’s offered to
• Theories, but lack of transparency about who
‘gets the gig’
• “Having free login for all riders means
sometimes I'm working but I don't earn very
much or even nothing at all. This was especially
a problem during lockdown. I would log in for
60+ hours but still barely make £100. Uber did
nothing to support riders during this time even
though they pushed for us to be classed as
keyworkers.” P3
Image: unsplash/alex-loup-aX_ljOOyWJY
24. Transparency and
accountability
• “Restaurant’s are disrespectful and can have huge waiting times
they have waiting times because the prioritise their own
customers over third party apps. Restaurant’s are disrespectful
due to the odd rotten egg in our chain and because support lines
don’t help them out. Customers can be aggressive when food is
late this isn’t our fault it’s the staff at the restaurants fault, I’ve
had pizza, pasta thrown in my face.” P9
25. Your future
as a beta
test
“I have worked with deliveroo for 3 years and
within my first year they terminated my contract
due to rejecting jobs which were too far and as I
was on an hourly rate of £6 plus £1 per delivery
and considered traveling these 1.5 miles plus
distances not worthwhile the time and risk.”
“now conscious not to get fired again so I have to
only reject jobs under dire circumstances
regardless if I'm getting paid less than £5 for
doing a 4 mile trip for instance or if the route is
unsafe for bikes” P26
26. Modality
and CO2e
externality
• ”Deliveroo systematically
deprioritised cyclists last year
and made it far harder for us
to be assigned orders or book
shifts to work at all (when
they used to have a booking
system)” P21
Image: unsplash/carl-campbell-NHv_7hIxJWQ
27. Customers
play too
• “I had a customer enter an incorrect
address which didn't exist. When I
called the customer to find out where I
should take the food he asked me to
deliver to an address outside my zone
of work on a high-rise housing estate I
don't know well and did not feel
comfortable entering. I told him I was
cancelling the order. I was really
frustrated by the loss of time as I was
not paid for this, and Deliveroo made
me return the food to the restaurant.”
P4
28. Beatings and
ratings
• “I was riding my motorbike down a
quiet street on a Friday night, 2 drunken
men tried to pull me off my motorbike
and were grabbing me. I managed to
ride off but the experience terrified me.
They told me the next day I would lose
all my bonuses for the week if I didn't
work my slots.” P14
• Female riders have experienced sexual
harassment from customers
29. Game over
• “I have worked during lockdown I am with
Deliveroo for over 2y We face bike thiefs
every week and police dont have any
action Deliveroo increased the drivers and
decreased our fees. The government
support its based in 2018 and i just started
in October so my help was 600£ for 3
months. To finish in great Deliveroo
suspended my account argument i was
taking long time. They ignored me, my
union and local MP when asked for
evidence of that.” P15
Image: unsplash/sigmund-By-tZImt0Ms
30. Burden of proof
• Little visibility or accountability
when unfairness occurs
• Algorithmic justice is swift
• Termination of an account results
in a loss of access to data helpful
to support a case
• “The order was then removed from my
app. I was told by Rider Support I would
still be paid for the order but needed to
send an email to request this. I then sent
an email to Deliveroo requesting to be
paid for this. They refused to pay because
I did not swipe "delivered". I was unable
to swipe "delivered" due to the order
being removed from my app at the time.
Deliveroo kept claiming that their "order
tracker" did not place my location
anywhere near the customer and this is
why they were refusing to pay. Luckily I
record every shift with Strava and I also
took a photo of the order, with location
services on.” P1
31. Sustainability?
IS THIS WORK SOCIALLY
SUSTAINABLE?
IS THIS ‘GOOD WORK’? IS THIS WORK ENVIRONMENTALLY
SUSTAINABLE?
32. Impacts of the technology
Disruptive digital
technologies are
changing when
and where people
work
The granularity of
what is paid work
The increased
‘outsourcing’ of
costs and ‘what’s
regarded as
overheads’
Minimising the
burden on the
employer to pay
the worker or
towards the
broader
infrastructure
Asks broad
questions about
what is acceptable
employment
33. Environmentally this is a complex question
Journey and
transport
modality,
could inc.
embodied
CO2
Carbon
intensity of
ingredients
and of
cooking it,
packaging
Length of
supply chains
and
transportation
modalities and
costs
Seasonality
of ‘diet
choice’ (hot
houses,
transport)
34. Gig workers’ voice
Gig economy workers are
independent actors,
subservient to the
‘platform’ – their voice is
missing from the design
of these systems
The platform is
continually changing
under them
No safety net
Little being done to
keep workers safe or
ensure that there is
sufficient work
The workers are not a
stakeholder
conspicuously visible to
the platform or customer,
except when things go
wrong!
35. Could we
help?
1. Ensure algorithms factor in fair pay and
conditions?
2. Transparency and attribution of hidden actors
(e.g. restaurant delays) or externalities
3. Better tools to keep workers safe, and reduce the
need for risk taking
4. Tools to enable reflection over true costs of gig
work for workers & policy setters
5. Provide better evidence sets for arguing for
workers rights
6. More primary data of ‘the ground truth’
experience, especially to help shape civic
infrastructures and policies
7. Promoting community and mutual support
36. Is this the
future of
work?
Call to action
Responsible
innovation in system
and UI design
Drones and
autonomous vehicles
37. • My thanks to Oliver Bates, Carolynne Lord (Lancaster), Ben
Kirman (York), Tom Cherrett, Toni Martinez-Sykora, Fraser
McLeod and Andy Oakey (Southampton)
• a.friday@Lancaster.ac.uk
• @gulliblefish
• http://flipgig.org and http://wp.lancs.ac.uk/sds
Image: unsplash/jon-tyson-hhq1Lxtuwd8