Home / Chroniques / How digital platforms use underpaid microworkers
Travailleurs du clic
π Digital π Society

How digital platforms use underpaid microworkers

Antonio Casilli
Antonio Casilli
Sociologist, Professor at Télécom Paris (IP Paris), and Associate Researcher at LACI-IIAC of EHESS

Tech­nol­o­gy is not going to erad­i­cate work for humans. Quite the oppo­site, in fact – there is a strong risk that it will “pro­le­tar­i­anise” it, by mak­ing that work invis­i­ble and lim­it­ing it to small, repet­i­tive, under­qual­i­fied tasks. This is the argu­ment put for­ward by dig­i­tal soci­ol­o­gy researcher Anto­nio Casilli.

Who are France’s microworkers?

They are work­ers who log on from home to plat­forms like Amazon’s Mechan­i­cal Turk to com­plete so-called “human intel­li­gence” tasks paid on a per job basis. They nev­er know who they are work­ing for, or to what end, because most of the time com­pa­nies pub­lish their jobs anony­mous­ly. The tasks, which are often used to sup­ple­ment arti­fi­cial intel­li­gence, are repet­i­tive and require few qual­i­fi­ca­tions – cir­cle toma­toes in a pho­to to assist nutri­tion appli­ca­tions, tran­scribe receipts, cal­i­brate vir­tu­al assis­tants by assess­ing the qual­i­ty of speech syn­the­sis, copy and paste, say words aloud, indi­cate the colour of a char­ac­ter, and so on – all for a few cents.

Accord­ing to the “Microwork in France” study that I con­duct­ed with my research group DiPLab (short for Dig­i­tal Plat­form Labour), there were approx­i­mate­ly 260,000 peo­ple microwork­ing at least occa­sion­al­ly in France in 20191. The major­i­ty are women (56%) who are 25–44 years old (63%), and near­ly half of peo­ple sur­veyed said they resort to this kind of work out of finan­cial need. Hence, microwork is often a source of extra income, even though the aver­age month­ly pay was only €21. Even more sur­pris­ing­ly, these “click­work­ers” have more qual­i­fi­ca­tions than the aver­age French per­son – 43.5% have done min­i­mum two years of post-sec­ondary stud­ies. Yet, despite their sig­nif­i­cant num­bers, these work­ers remain com­plete­ly invisible.

Why are they more invis­i­ble than oth­er work­ers in the dig­i­tal space, like food deliv­ery or rideshare drivers?

First­ly, it’s because of the very nature of their job. In real­i­ty, microwork is just remote work­ing pushed to the extreme. It’s a remote activ­i­ty, but for boss­es and with col­leagues who are often com­plete­ly anony­mous. On the oth­er hand, Uber and Deliv­eroo dri­vers have been seek­ing out col­lec­tive solu­tions and form­ing unions to make their voic­es heard. This is even more the case since they are vast­ly more vis­i­ble than before when they roam the desert­ed streets of cities under Covid lock­down. But, unlike these work­ers, microwork­ers have zero pres­ence in the pub­lic sphere. And the super-strict con­fi­den­tial­i­ty agree­ments that they may have to sign make it hard­er for them to seek recog­ni­tion for their work.

More and more coun­tries, such as Spain, the UK and France, are requir­ing plat­forms (par­tic­u­lar­ly Uber) to recog­nise the employ­ment sta­tus of their work­ers. Yet the many hands that train our AI in the form of microwork are still ignored. I think efforts to reg­u­late the indus­try come in large part from work­ers’ vis­i­bil­i­ty, and from the phys­i­cal prox­im­i­ty between deliv­er­ers and clients. It’s quite a para­dox, as there are sig­nif­i­cant­ly more microwork­ers, both in France and inter­na­tion­al­ly, at least accord­ing to our esti­mates and those of our col­leagues at Oxford Inter­net Insti­tute2

What we do know is that phys­i­cal effort is not the only way that deliv­er­ers pro­duce val­ue. They, too, per­form invis­i­ble tasks, pro­duc­ing data for the plat­form and there­by feed­ing the algo­rithms and AI that serve in part to improve algo­rith­mic solu­tions. This data must then be processed and, con­trary to what you might think, it can­not be processed with­out the sup­port of a large human work­force – which does not need to be locat­ed in the coun­try of pro­duc­tion. While France has 260,000 microwork­ers, these plat­forms have more than 100 mil­lion peo­ple signed up inter­na­tion­al­ly. The vast major­i­ty of these “crowd­work­ers” are found in devel­op­ing coun­tries, in Indi­an, South­east Asian, African and Latin Amer­i­can “click farms”. 

This huge lev­el of out­sourc­ing is redefin­ing the geog­ra­phy of the future, but it also influ­ences the recog­ni­tion of this work. There are pure­ly polit­i­cal stakes with regards to the issue of the invis­i­bil­i­ty – not only are these work­ers not vis­i­ble in the pub­lic sphere, but the over­whelm­ing major­i­ty also can­not vote in their (devel­oped) coun­try of res­i­dence, unlike deliv­ery and rideshare dri­vers. Con­se­quent­ly, their recog­ni­tion is not a polit­i­cal priority.

Are you in favour of a “dig­i­tal social income”?

The automa­tion and “plat­formi­sa­tion” of the econ­o­my is caus­ing human work to explode, not die out! We are all microwork­ers in our own way – when you com­plete a CAPTCHA or add a hash­tag on Insta­gram, you’re help­ing to train a com­put­er vision sys­tem or pro­vid­ing a cat­e­go­ry for your post, instead of the plat­form doing it.

This is why there has been dis­cus­sion in France since 2012 about the fis­cal recog­ni­tion of this “free work” done by app users3. Of course, it’s not the users who should pay these tax­es, but the com­pa­nies that own the plat­forms. Some 93% of French peo­ple use Google, and the com­pa­ny should pay tax­es in pro­por­tion to the amount of data pro­duced by French cit­i­zens. This fis­cal rev­enue could then be used to fund redis­tri­b­u­tion poli­cies, includ­ing a “dig­i­tal social income”.

The aim of this income would be to divide up the val­ue gen­er­at­ed by click­work­ers, microwork­ers and ordi­nary users, in an uncon­di­tion­al way with all social ser­vices con­sid­ered equal. The aim is not to com­pen­sate peo­ple indi­vid­u­al­ly accord­ing to the amount of time spent on plat­forms. The con­se­quences of that would be dis­as­trous, with super­pow­er plat­forms pay­ing peanuts for our clicks while we slave away over micro­jobs! The aim is sim­ply to redis­trib­ute the val­ue pro­duced, which until now has been hoard­ed by plat­forms, to pre­vent the “pro­le­tari­sa­tion” of human work.

Interview by Juliette Parmentier
1Rap­port final du pro­jet DiPLab, par Casil­li, A. A., Tubaro, P., Le Ludec, C., Cov­ille, M., Besen­val, M., Mouhtare, T., Wahal, E., « Le Micro-tra­vail en France. Der­rière l’automatisation de nou­velles pré­car­ités au tra­vail ? », 2019, http://​diplab​.eu
2Otto Käs­si, Vili Lehdon­vir­ta, Fabi­an Stephany, « How Many Online Work­ers are there in the World? A Data-Dri­ven Assess­ment », ArX­iV, 2021, <arXiv:2103.12648>
3Nico­las Col­in et Pierre Collin, Rap­port relatif à la fis­cal­ité du secteur numérique, Paris, La Doc­u­men­ta­tion française, 2013


Antonio Casilli

Antonio Casilli

Sociologist, Professor at Télécom Paris (IP Paris), and Associate Researcher at LACI-IIAC of EHESS

Antonio A. Casilli directs the DiPLab (Digital Platform Labor) research group and was among the founders of the ENDL (European Network on Digital Labor). His publications include "Waiting for the robots. Enquête sur le travail du clic" (Seuil, 2019 laureate of the Fondation Colbert-Institut de France) and "Les liaisons numériques. Vers une nouvelle sociabilité?" (Seuil, 2010). He was the editorial advisor of the documentary series based on his research "Invisibles - Les travailleurs du clic" (France Télévisions, 2020).

Our world explained with science. Every week, in your inbox.

Get the newsletter