Michael VealeOct 31, 2022

[New 📝] Focus so far on the EU’s proposed Platform Work Directive has mostly been on employees misclassified as self-employed, yet the draft contains powerful, broadly well-drafted protections on #algorithmicmanagement — new digital labour rights for individuals & collectives!

These will be be fought hard against by big platforms.

In our new pre-print, @RDBinns, Six Silberman + I analyse and propose specific tweaks to safeguard digital #workersrights.

📄: https://doi.org/jjr2 [🧵/n]

Show threadMichael VealeOct 31, 2022
Platform workers face challenges regardless of their employment status, including 🧾 unaccountable nonpayment; ❌ account suspension or closure without explanation/recourse; 🖥️ ineffective/unfair automated decisions; 💬 uncommunicative customers/admins. We focus on 4 main themes tackling these: Transparency; Human Oversight; Information & Consultation; Communication and Monitoring. [2/n]
Show threadMichael VealeOct 31, 2022
On 🧊 transparency, some provisions add little the GDPR, although unlike the GDPR the proposal requires the ‘main parameters’ and the ‘importance’ of important automated and semi automated decision systems. The EU has used the ‘main parameters’ framing a lot; in the paper we critique it, consider its interaction with the AI Act, and ask the legislator to think harder encouraging limits to model complexity that prevent accountability. [3/n]
Show threadMichael VealeOct 31, 2022
On 👀 human oversight and review — against what standard are automated decisions reviewed? If high, that’s really sneaky harmonisation of EU labour law; will never pass. Instead, our modest proposal focusses on injustices specific to algorithmic decisions. Individuals need rights to have decisions *restated* by a human, w/ separate, interpretable reasoning, supported by evidence, *substituting* original algorithmic logic, allowing decisions to be challenged. More on up & downsides in paper [4/n]
Show threadMichael VealeOct 31, 2022
On 🎙️ Information & Consultation, proposal provides for platform workers to have representatives. In reality, many platforms won't — the proposal needs to do more for when no representative exists to ensure platforms don’t bias procedures as we have seen them do before. There needs to be greater focus on consultation across languages and borders. And because we don’t really know how to consult platform workers at scale yet, information and consultation procedures need to be adaptive. [5/n]
Show threadMichael Veale
📞 Communication and Monitoring, new obligation on platforms to make a comms channel to allow workers to organise. This is hard, both to build a comms people want to use, and to do it without collecting personal data as proposal says! We evaluate the challenges faced by civic techs, the tensions in maintaining exception-free confidentiality (eg harassment), & pose a more workable way forward that helps workers choose comms tools they want to use, while flagging these to colleagues at scale. [6/n]
Oct 31, 2022 at 2:06pmWeb
Show threadMichael VealeOct 31, 2022

We also consider the tensions in monitoring: many provisions require impact to be assessed, or context about whether the worker is or is not working to be known, but with little data this becomes hard. It also may interact with fireworks in the world of general online tracking. This part needs some quite significant tweaking to get right.

Thanks, we hope it's useful & makes its way to legislators to contribute to Council, Parliament, trialogue, and regulators too! https://doi.org/10.31235/osf.io/7jyhe [7/n]