Gig Economy Project – Accessing the algorithm to build union power: The case of Palermo

Interview with Fabio Pace of the NIdiL CGIL union in Palermo, which has used legal action to reveal the secrets of algorithmic management and strengthen the power of food delivery couriers.

Interview by Piero Valmassoi, EU policy expert in the platform economy and sustainable urban mobility.

On 20 June the Court of Palermo ruled that Spanish food delivery platform Glovo has to disclose how its algorithm allocates tasks to workers. 

The decision follows other recent court cases which have found that workers have algorithmic rights. In the Netherlands in April, a court found that Uber and Ola must provide workers with all information on their systems of automated decision-making. 

These developments have come in a broader context of concern about the abuses of algorithmic management practices (such as algorithmic wage discrimination). We have seen strikes of Wolt workers across Europe against ‘dynamic pricing’ and a legislative text for algorithmic transparency proposed in France by Communist party senator Pascal Savoldelli

The union NIdiL (Nuove Identità Lavorative) CGIL was one of three which jointly took Glovo to the Court of Palermo. Piero Valmassoi spoke to Fabio Pace, representative of NIdiL CGIL Palermo, about the importance of the Palermo judgment. 

Piero Valmassoi: What was the union’s motivation for pursuing the court case and why is it something important to you?

Fabio Pace: As NIdiL CGIL, we have been dealing with the subject of algorithmic management since the start of the platforms’ boom. Our objective is to access information about algorithmic management, a central element of the employment model of platforms, to demonstrate openly a series of aspects of the work organisation.

A decree of Mario Draghi’s government [from February 2021 to October 2022] had already imposed algorithmic transparency on digital platforms, but later it was partially reversed by another decree of the current right-wing government. This decree allows platforms not to reveal all the information about algorithmic management to protect commercial secrets. At the same time, it leaves us some margin of manoeuvre to demand transparency on the modalities of work organisation, through the tool of labour disputes for anti-union behaviour under Article 28 of the Workers’ Statute. 

This was the main reason to pursue this case: the more you access information about the platforms’ employment model, the clearer it becomes that workers are subordinate. 

PV: Can you give us some details on the judgement of the Court of Palermo? How will it work in practice? 

FP: Before I go into the details, I would light to shed some light on the general context.

More or less a year ago, NIdiL CGIL Palermo opened a channel of communication with Glovo, the leader of the Italian market.

In Palermo, we also had a smaller platform active locally called Social Food, which was guaranteeing to a number of its workers the possibility to earn decent money, but in February 2023 it was sold to Glovo. This implied no obligation to keep its workers, but some were absorbed by Glovo.

Glovo’s buyout of Social Food caused discontent among the workers, who were not able to enjoy the same working standards of the past, as the platform was hiring too many riders compared to the amount of work available: this was the leverage to open a channel of communication with the platform. Glovo’s manager listened to our concerns, but no action was taken, so we filed a labour dispute.

To answer your question, Glovo understood that workers and their representatives were able to attack them on different fronts and a couple of days after the court’s decision, they provided us with their policy on algorithmic management. 

PV: Do you consider this as a success for unions’ action on the issue of algorithmic management?

FP: Certainly, it is a step forward towards the determination of the nature of the work relation. We always come back to the falling point represented by the classification of workers, either or subordinated, which must match with the reality of the work relation.

PV: What did you discover from the information Glovo released on the modalities of tasks’ allocation via algorithmic management? 

FP: What is important to underline is that we as a union, thanks to a good relationship with riders, were already aware of most of the information that Glovo released. We needed this step to openly reveal the mechanism.

The mechanism consists of an algorithm that imposes certain standards on riders, binding them to continuity of work performance over time: the longer time you are available on the app, the more deliveries you will be allocated. And this mechanism implies certain elements of discrimination, as shifts’ booking opens every Monday (for the weekdays) and every Thursday (for the weekend) at 10am, and it is reserved to the riders that have a higher internal rating, while the others can only start to book working slots from 4PM, when the calendar is already quite full.

The internal rating according to which tasks are allocated is called the “excellence score” and it is composed of a series of parameters: the feedback of the customers, with negative feedback weighted on the total number of deliveries realised. 

Another criteria is the number of deliveries realised; according to the platform, the number is linked to the average of deliveries of all the riders in the same town in the previous 28 days: this average ca be spoiled by discrepancies, such as the rider from Palermo who had the world record of deliveries. This obviously impacts the possibility of receiving tasks for other riders, who are below this average. 

Additionally, Glovo considers the average time to complete a delivery, the number of deliveries realised by riders during peak hours – when the demand is higher – and the number of “no shows”, when a rider who gave his availability for a shift does not access the app. 

These are mechanisms that oblige the worker to perform tasks under the conditions set by the platform, which clearly shows a link of subordination.

PV: The Court of Palermo in April issued a similar judgement towards Uber Eats. Recently, Uber Eats has announced that it will cease its operation in Italy. In the meantime, have they respected the court’s judgement by opening their black box? And do you think that this judgement might have influenced their departure from Italy? 

FP:Uber Eats also provided their policy on personal data and on algorithmic management, which is more direct than the Glovo one: criteria for assigning tasks are linked mostly to clients’ rating and to work continuity over time. 

But this was not the reason why Uber Eats left Italy: they left because they did not successfully enter the Italian market, dominated by Glovo. The only competitor of Glovo is Just Eat, which is our main interlocutor in Italy, as they signed an agreement with the unions in 2021 which accepts that riders are subordinated, our main argument.

PV: In front of the opaqueness of the platforms about algorithmic management, did workers and unions put in place alternative strategies to get information about the task allocation system?

FP: Yes, we did that by going on the street, where workers are trying to build a relationship of trust between them and the union. As a trade union representative, you must put yourself into question, discuss with workers that do not necessarily agree with your demands and practices. It is hard work, but it pays back when you manage to put into place mechanisms of pressure on the platform – for example, through labour disputes or by sending hundreds and hundreds of emails – which in the end force platforms to deal with you and respond to your demands. 

On the other hand, you have workers who, thanks to this relation of trust, had helped the union to decipher certain aspects of the work, which were obscure to us as union representatives. The truth is that we do not really care about being officially recognised by the platform, or whether the worker is affiliated to the union. What we really care about is receiving meaningful answers from the platform, which can address the problems of the worker. 

PV: Have you discovered examples of dynamic pricing, where prices are hyper-personalised and context-based? 

FP: We have not ascertained that specifically, but we discovered that the same rider did the exact same delivery in different days, from the same restaurant to the same customer, earning a different rate. 

Glovo uses the tool of multipliers at their complete discretion: they can apply different multipliers to tasks done by different riders and at different hours, which results into different rates for the same work performance.

PV: Have you had experiences of riders who were blocked by the app, after being falsely accused of fraudulent activity by the platform? How does NIdiL CGIL deal with these unfair robo-firings?

FP: Yes, we have experienced every kind of problem that led to the unfair deactivation of riders from the platform. 

Due to the problem of the malfunctioning of facial recognition technology, which Glovo has now removed, we used to have at least three or four cases per week. This was one of the problems to which we responded by sending massive amounts of emails to the HR Manager of Glovo and threatening to undertake labour disputes. 

The reality is there is no person within the platform who is dedicated to address the individual problems of workers. There is a lack of real communication channel between workers and platform. There are some problems that are easier to fix than others (for example, a faulty recognition of personal documents or credit card number), but other cases are more complicated to address. 

PV: A fundamental question for gig workers is the possibility of accessing their data and use them to defend their interests. Do you think that unions in Italy have been actively organising workers around the issue of using data to prove certain aspects of the work relationship between platform and workers? What could be improved in their action? 

FP: As mentioned earlier, our strategy of labour disputes is based on this concept: we want to exploit the resources of platforms, the algorithm and their access to personal data, to use them as a tool to demonstrate the link of subordination. This was the main line of action until now and the recent judgement of the Court of Palermo shows that we are going in the right direction.

The Gig Economy Project is a BRAVE NEW EUROPE media network for gig workers in Europe. Click here to find out more and click here to get the weekly newsletter.

Thanks to many generous donors BRAVE NEW EUROPE  will be able to continue its work for  the rest of 2023 in a reduced form. What we need is a long term solution. So please consider making a monthly recurring donation. It need not be a vast amount as it accumulates in the course of the year. To donate please go to HERE.

Be the first to comment

Leave a Reply

Your email address will not be published.