Utrecht University organizes three private workshops about the platform economy. Platforms, trade unions, academics, governments and businesses are looking for solutions to dilemmas they have never really addressed together. The first session deals with ‘the algorithm as mysterious blackbox’.
Government, businesses and trade unions, the platform economy impacts all of them and they all hold their own opinions. A few dilemmas are recurring, especially because these parties rather avoid these common topics. Platform expert Martijn Arets of the Utrecht University is willing to bring about change.
He invites up to 40 stakeholders for three workshops about dilemmas regarding the platform economy. Arets: “They have their backgrounds in all kinds of trades. During the first session, trade unions like FNV, platforms as Roamler and Deliveroo and huge organizations like TNO and KPMG were present. Representatives of KPMG, the Dutch Ministry of Economics and Climate and the UvA shared their insights.”
During the workshops stakeholders get to know one another and start the dialogue. “The discussions are meant to be constructive”, emphasizes Arets. “Each and every one has an opinion, but stakeholders hardly ever talk with one another. Still they have to find a solution together in the end. It is my goal to convince them so.”
During the first discussion day, we dealt with the transparency of algorithms, or rather the lack thereof. Arets: “It is unclear how exactly platforms make the match between supply and demand. An algorithm is a mysterious blackbox for most of us and instills distrust.”
The urgency of transparency
Openness about algorithms is tricky for two reasons. First, the algorithm is an essential part of a platform; the app with the best matchmaking system for supply and demand wins. So, it is highly competition sensitive.
Secondly users often times do not want to know how algorithms work. An in-depth explanation is way to complicated and doesn’t contribute to trust in the system. Representatives of platforms also uttered that most them and their competitors are perfectly able to explain how their algorithms work in the case it would be asked. But practically users hardly ever call in to ask for such an explanation.
“Only when encountering trouble, things become urgent for platforms”, says Arets. He tells about Taxi app Uber, as an example, which is making prices soar as soon as it understands that you really can’t go without. Later on he mentions the Dutch trade union FNV, which in a court case against Deliveroo stated that deliverers are being disadvantaged when they often turn down gigs. “Deliveroo denied this. And that was how their discussion ended. It simply wasn’t possible to check and resulted in mistrust of the platform.”
KPMG and the algorithm accountant
It is smart to anticipate these kind of discussions, according to Arets. “Set out proactively with your explanations, because consumers are getting more and more aware of their responsibility and privacy.”
The best way of getting about was discussed in a presentation by, among others, Frank van Praat of KPMG. He explains that it all stands or falls with explainability.
This is to say that not all of your system has to be unveiled. Definitely not. It is about you telling which choices are made and the impact they have on the individual. On a babysitting platform it is a logical consequence that your rating drops when you showed up late for three times. But you have to be clear about that to the users.
Van Praat pleas for a kind of platform ‘insert’: a general explanation of the risks with a simple description of the choices an algorithm makes and their reasoning. This should be as understandable as possible; users should really read such an explanation.
Furthermore, he fancies the idea of an ‘algorithm accountant’, as long as the responsibility for trustworthy algorithms remains in the hands of the user. Van Praat: “The task of the accountant is mainly to check whether the user of the algorithm is working in trustworthy ways (i.e. conform the statements made in the insert).”
Regulatory pressure and innovation
Not everybody agrees with him on these statements. Controls cost both time and money. Platform entrepreneurs ask themselves if this regulatory pressure is proportional. Especially in cases it concerns app that work with less sensitive data. Parties are also afraid that too much control hinders innovation.
“There isn’t much experience yet in the field of algorithms and accountancy”, summarizes Arets. “It is of importance to think about the impact of algorithms and the risks of things going wrong. Based on these considerations it will be possible to decide which algorithms should subject to audits.”
“A second consideration is whether regulatory pressure should make a difference between digital and non-digital intermediaries”, he says. “Nothing should be done, just because it is possible to do so.”
This article was originally published in Dutch on ZiPconomy.
This post is also available in: Dutch