Posted 08 July 2021
Earlier this year, my surname appeared on a list at Pontins holiday parks. Published by a whistleblower, the list instructed staff to refuse bookings under my surname (among others) in an attempt to blacklist Gypsy and Traveller families from the site.
Thankfully, because Gypsies and Travellers are recognised as distinct racial groups in the UK, the Equality and Human Rights Commission were able to step in and put an end the blacklist.
I didn't want to go to Pontins anyway. But one place I did want to go was Instagram. I increasingly need an Instagram account to view food menus or opening times for local businesses, and so earlier this year I reluctantly signed up. Three times.
Each time, my account was immediately suspended with the message 'Your account has been disabled for violating our terms'. Perhaps I reminded their 'AI' of a difficult memory from training? A browser, locality, or email domain it doesn't trust?
Like Pontins, Instagram might be excluding me alongside others - but, bound by unknown variables in training data, and with no human label for us, I'd have no way to find those people.
So, to all the other members of group f1c6bc8ad92c49ee44a48e116b4b4582 (or whatever we're called), I'm sorry, no whistleblower will recognise us, and the Equality and Human Rights Commission will not come for us.
I'm probably as hurt by a lack of Instagram as I am by a lack of Pontins holidays - not at all - but these events had me thinking whether discrimination is only discrimination if the target has a name we recognise.
Prev: Gemini in tension