Algorithms, Work and the European Directive – Interview with James Farrar and Sergi Cutillas

Interview with the authors of new report “Workers’ Recommendations on the draft EU Platform Work Directive” in podcast and text form

THE Platform Work Directive has reached a crucial stage. The European Parliament’s Employment Committee has backed a text which would strengthen the provisions for workers, while the Council of the EU failed to come to an agreement on its proposed changes, after the Czech Presidency of the Council sought to push the Directive in a pro-platform lobby direction. When the Council eventually does come to a decision, the Parliament and the Council may have to work out how to reconcile two texts moving in opposite directions.

Thus, there may still be an opportunity for workers’ to influence the final outcome of a Directive that will have a profound impact on platform workers’ rights in Europe and beyond. In that context, the recent intervention into the debate over the Directive by UK-based Worker Info Exchange (WIE) and Spanish-based The Observatory of Work, Algorithms & Society (TAS) is significant.

Workers’ Recommendations on the Draft EU Platform Work Directive” was published by WIE and TAS earlier this month. The report is the most comprehensive proposal for how the Directive should be improved to come from worker organisations in Europe so far.

To discuss the report, and the work of WIE and TAS more broadly, the Gig Economy Project spoke to its authors. James Farrar is the founder and director of WIE, as well as General Secretary of the App Drivers & Couriers Union (ADCU), which led the historic defeat of Uber at the UK Supreme Court last year. Sergi Cutillas is an economist and researcher, who is a member of TAS and has conducted research for taxi union Elité Taxi Barcelona.

In this podcast, we discuss:

01.30: How did the collaboration between WIE and TAS come about?

08:10: A presumption of employment for platform workers?

20:21: Algorithmic management and the need for a ‘human model’ in the platform work directive

38:52: The roots of WIE and TAS, and next steps

AN ABBREVIATED TEXT VERSION OF THIS PODCAST IS ALSO AVAILABLE BELOW

The Gig Economy Project: How did this collaboration come about between Worker Info Exchange and the Observatory of Work, Algorithms & Society?

James Farrar: We were introduced and asked to consider what the worker voice and input into the Directive could be, as the voice of the worker was pretty uncoordinated and maybe not as coherent as it could be. It’s been a great collaboration because we are working on a similar topic in different markets, but it’s the same platform dynamics that we are trying to get to grips with.

Sergi Cutillas: It’s almost the first research work that we have done at the Observatory. We have been building TAS for the last year, and we have been working on pre-litigation work in the sector of the taxi and the private hire vehicle, and we recently presented a lawsuit against Glovo, which is a competition law lawsuit, but this is the first research work and we are very happy. We knew of James Farrar’s work because I have been trying to find arguments to convince politicians and doing the economics part of different lawsuits and James’ work has been a reference in these years. So we hope this is the beginning of more collaborations and work together.

GEP: James, as someone working in the UK, why did you think the EU Platform Work Directive was something important to think about and work on?

JF: We have a common problem, and the regulatory response is going to be different in Europe than it is in Britain, but the first step is to understand that problem as best we can because there is so much hidden from us. Then try to think through what is the best, most ideal policy response, and if the platforms are operating the same in many jurisdictions, as they broadly are, then the policy response ought to be broadly the same as well. How policy ends up in legislation is a different story and obviously the process in Europe is very different. 

Here in Britain, I’m worried that we are actually going in the wrong direction. We said in the report that the EU Directive falls short in some key areas, but in Britain we haven’t even started to come to terms with the issues of algorithmic control and human oversight, which the law deals with, nor working time or the presumption of employment. And in fact the government is taking a knife to the weak digital rights protections that we have under GDPR. It is going to raise the threshold for access rights, our ability to get transparency in the explanation of digital processing, and also – really critically – to almost completely remove GDPR Article 22 protections against automated decision-making, which are key pillars of the EU Platform Work Directive. 

So it’s really useful to work with Sergi on the EU context because it gives us a lot of clarity about what needs to happen in Britain and hopefully it can happen here as well.

GEP: Let’s discuss the part of your paper focused on the presumption of employment. Should platform workers be employees and what should the basis be for triggering an employment classification? 

SC: Our initial thoughts were to ask for an automatic employment classification for all platform workers; that would be the most ambitious point of view. But we also understood that this was a very controversial issue and in this negotiation the Directive could be rejected altogether if this automatic classification was triggered. We also looked at the proposal of Elisabetta Gualmini, the European Parliament rapporteur, who proposed to move the criteria to the preamble of the Directive. That for some lawmakers was interpreted as a way to trigger this automatic presumption of employment. In the end our proposal has more or less remained as Gualmini put it, with this criteria described in the preamble. We thought it was good to support this proposal, because we think it improves the situation of workers now. Of course it could always be better, but we think it gives enough guarantees to workers for the coming period.

In the example of Spain there have been thousands of judicial sentences recognising the employment status of delivery workers, even before the Rider’s Law was passed. So it is kind of a common sense issue in a way. When one sees the conditions of control, the conditions of management, the coercion, etc, it is easy to see that this is a work relation not a business relation among equal partners. That was the most difficult part of the negotiation in the European Parliament, and I think we are happy with the vote in the Employment Committee in favour of this proposal.

GEP: James, the platform lobby say that the Gualmini proposal will lead to automatic re-classification of all 28 million platform workers in the EU and many of those workers won’t want that. How would you respond to that?

JF: When we started working on this, Sergi and I were looking at the platform lobbyist Move.EU’s position papers on the issue. They said that the presumption of employment burden had to lie with the worker because the platforms couldn’t sustain the pressure of having to bear the challenge of millions of workers around Europe. But when you turn that on its head you think, ‘this is crazy’, because are the platforms saying that those millions of workers should take on the individual burden themselves and challenge the platforms? The truth is that someone has to carry this burden, and it really has to be the platforms.

The other problem with this is that you end up in a negative cycle because how does the worker prove employment when the controls that indicate employment are concealed from you in the algorithm? A worker can’t prove that they are an employee because there’s no algorithmic transparency. So for those reasons – the precariousness of it and the fact that there is many workers compared to platforms and that the platforms are the ones holding the cards – means of course they should have the burden of proof. 

GEP: The second part of the report looks at algorithmic management, and you have a whole series of proposals for how to strengthen the platform work directive. What are the most crucial aspects of the Directive on algorithmic management?

JF: The crucial part is the transparency of understanding automated decision-making and having some consultation in it. The devil is in the detail of the level of information that is provided to you and how useful that can be. We tried to deal with that in the paper by referring to a standard of AI explainability that the Information Commissioner’s Office and the Turing Institute came up with. 

What’s missing from the Directive, and the issue I am really hot about at the moment, is regulation on dynamic pricing. This is a serious issue. The harms of dynamic pricing are already well understood. What you have is competing algorithms which are tacitly colluding to hold prices below where they would otherwise be. The clear risk is that you will have drivers who are gamed, you will have competing algorithms tacitly colluding to hold pay down, personalising pay by understanding who is going to accept what, and even if they deny that they do that the effect can be indirectly happening anyway by use of these technologies. For me, that harm in dynamic pricing is really serious.

GEP: Gualmini has said she believes the algorithmic management part of the Directive will be more important for workers going forward than the presumption of employment part. Could this Directive lead to a big change in algorithmic management?

SC: Both parts of the Directive are very important, but the algorithmic part which James has described is really important because this is getting really serious. Probably in a very short time we will have algorithms which are very complex, very difficult to understand, even by the programmers that created it, because the algorithm will have their own life and it will be very difficult to set limits on them so they are non-discriminatory and follow the laws. So if we don’t start setting legal standards and frameworks to deal with that then we will have a very serious problem very quickly.

This is a very important law, of course it is not enough, there are still some gaps, but it is a good start, and it is very important that workers across Europe, not only ones that are working in platforms but in general, understand that this is a very important issue. The technological companies move very quickly, and policy and laws and peoples’ consciousness has to catch-up very quickly.

JF: Already we can see that the technology is outpacing the Directive, and some of the gaps we have pointed too are things that the regulators need to think about, even if they are not covered. The reason why is we have had 10 years of platforms and in the early years they were fuelled by venture capital money, growing rapidly and causing lots of chaos. Those days are over, platforms have slowed down, they are maturing, platforms like Uber and Deliveroo are public companies they have to deliver profit and where can they turn to in order to increase their margins? It’s through algorithmic methodologies, because they don’t have a workforce, they don’t have any assets, they have nothing except these algorithms from which to conjure up money.

Uber in August sent a note to drivers in the UK to say, ‘you may remember a time when you were paid on time and distance, those days are over, we’ve come a long way, now we are using years worth of data’. Their explanation stops there, because if it is not based on time and distance, what is it based on? Nobody knows. It’s what they set it to be. We don’t have an understanding of how pay might be set going forward, and that opens things up to a lot of abuse.

Also, in the financial accounts, Uber’s margin had increased significantly, the ‘take rate’ had doubled in a year, but also the R&D budget had doubled in a year. And they told the FT that their biggest technology investment had been in this matching of pricing. And I’m afraid we as a trade union movement and the regulators need to keep pace with that, because those types of technologies have the potential to be seriously abusive to workers and markets, and we are barely scratching the surface.

GEP: As well as the European Parliament process around the Directive, there is also the European Council process, where they are working on their own amendments to the text. One of the things that was leaked about the proposals of the Czech Republic, which is the current presidency of the EU Council, is that they wanted to make the algorithmic management section no further than current GDPR standard. The Gig Economy Project did an article recently on a case study in Switzerland with an Uber driver trying to access his own data, and it was extremely difficult for an ordinary platform worker to use GDPR and get any meaningful information out of it. 

JF: In our report last year, ‘Managed by Bots‘, we chronicle all the abuses, obfuscation and delay to frustrate the process of worker access. But even when there is access there is also very limited understanding. Where this can become really important is at the point of dismissal. It goes like this: ‘I’ve been accused of fraudulent activity, but I don’t know what the fraudulent activity is, the allegation hasn’t been put to me, I haven’t had a chance to answer it.’ We have two cases currently at the Court of Amsterdam, we expect to get a ruling from the court of Appeal on the 20th of December. So it gives you an idea of how difficult it is and even when workers access their rights as fully as they can do they still are denied relevant information. 

GEP: Sergi, you have researched how platform algorithms work in the context of potential price-setting between different platforms operating in Spain. How important could legislative change be in empowering unions and workers in relation to algorithms?

SC: We are defending a human model against the potential of big capital with very powerful machines. It’s already difficult for workers to assert some bargaining power and defend their rights, and if we include this blackbox of algorithms and AI that is at the early stages, I think it will become even more difficult to defend our rights.

Human intervention needs to be there. The processes need to be between humans. Also, because machines are always used as a way to detach the employer and employee relation, but there is such a relationship.

Since the earliest stages of capitalism when we had workshops and small factories, the control was very direct. The boss was the boss. And then we started to have longer chains: Taylorism, Fordism, the thing became more complex and abstract. And with finance, even more. The management doesn’t necessarily have to be the same as the investors. And now we have reached a moment where it seems like things are happening through technology alone. So I think we need to fight for this clarity. Technology can improve our lives, but we must fight to make this human relationship clear.

One has to understand that these algorithms, at this moment, I think they are not very good. We always think about technology as very clever, but human beings are the best machines that exist. Because these machines discriminate, make mistakes, and produce all kinds of abuses, even if you program them not to do that. As this complexity increases, the number of problems will also multiply. This is a very important struggle, and I think politically we will be struggling in many areas to defend not only a social and just system but a human system.

GEP: The Worker Info Exchange and the Observatory of Work, Algorithms & Society are quite unique in Europe, in the sense that they are interested in data, algorithms and technology from the perspective of the working class and they both have close connections to trade unions. How did they come about?

JF: I was involved in a tribunal case against Uber. I brought evidence that I was paid £5.03 an hour, relative to the national minimum wage at the time which was around £7.50. Uber went to the press to suggest I manufactured my evidence. They came to court with every job that I’d ever done. My acceptance rate, my rejection rate, how much money I made, and I had none of that information. 

What they tried to argue in court and to the Select Committee at the UK Parliament was that I shouldn’t be paid minimum wage because I rejected X amount of my work. They were focused on ‘he is rejecting half of the work we send him, what’s he doing? He could be at home resting and all he had to do is just keep cancelling jobs to get the minimum wage.’ So they focused as a metric on what I rejected, but they didn’t focus on what I did, my actual production. And my production was at the higher end of productivity that Uber suggested to the Court. They said they expected a driver to work 40-60 hours, I was working 91 hours. 

It said to me that Uber had my data and was spinning it against me. I was luckily able to pierce through that and figure out what they were saying and rebut it. But it was just sheer luck.

Secondly, if you look at what the Supreme Court has said in the UK in respect to the Uber case last year, the message of the Supreme Court is that the statute book trumps the contract. So if you have a contract that creates this fictional world, it will be rejected by the court. What do you expect platforms to do now? If they can’t create a fictitious contract and impose it upon you, the best they can do now is to hide that control from you.

So those two insights for me were really powerful in shaping an understanding that if the burden was on the worker to contest that they are in an employment relationship, they will be doing it with their hands behind their back if you don’t access the data and understand its processing. 

GEP: Sergi, the Observatory is an alliance of different workers, you have the RidersXDerechos but also the traditional taxi drivers in Elité Taxi. What is the thinking behind the Observatory?

SC: What united us was a struggle against the platform economies, in the case of the taxis it is a bit different because they are from the outside. Taxi drivers have been accused of being obsolete, of not being technologically developed, etc etc. So the union Elité Taxi started with a strategy of mobilisation and defending the existing regulations, but at some point they realised that they had to make more sophisticated arguments. They had to start reasoning in terms of competition policy, competition law and the algorithms of the digital platforms.

So I started helping Elité Taxi with these tasks, we did research, working with lawyers, to understand what was happening in the economic area and also in the digital rights sphere. At the same time the taxis were supporting the RidersXDerechos struggle, so we collaborated in close initiatives and also lawsuits together. And at some point we decided we needed a vehicle to do things together. The Observatory has been the vehicle that allows us to go beyond the sectoral limits, and to link to organisations across Europe and the world. So far, it has been a successful initiative. This report is an example of how important it was to create the Observatory. And we are just at the beginning.

To sign up to the Gig Economy Project’s weekly newsletter, which provides up-to-date analysis and reports on everything that’s happening in the gig economy in Europe, leave your email here.

You decide if we close down BRAVE NEW EUROPE and the Gig Economy Project

We have set a goal of 20,000 GBP/Euros. If we do not reach this, we will continue as long as the funds last then close down the website.

Thus we are letting you decide. If you want the site to close down, then don’t donate. If you want them to survive and flourish then donate HERE

Be the first to comment

Leave a Reply

Your email address will not be published.


*