Councillors, your ward target pools are racist - they're also strangling your vote
By James Belchamber | Mon Aug 24 2020
Last week Rob Davidson said "algorithms don't fail people, people do" in his article about the A-Levels results crisis. But did you know that people can teach algorithms to be racist too (Twitter managed it in less than a day) It's with this in mind that we need to approach the algorithms within our own party.
Thought experiment: let's create a Lib Dem voter detector. We would feed it existing Lib Dem voters, identify things that are common between these voters, and then use this data to seek out new voters. We would enrich this data with existing data from our central database, which is where we upload details on people our activists have spoken to.
How could this go wrong?
Well, as an example: how many times have you heard, when working the streets as an activist, that we "Don't go to that area. That area is a Labour area"? If you ever dig into that, you'll find that a surprising amount of the time that means one thing: it's not a middle-class (and overwhelmingly white) area. This is one of the many ways we unconsciously bake racism into our campaigns.
Now, our detector doesn't understand what racism is - so, it can't realise when that racism goes from unconscious to conscious (remember: computers are not conscious!). It will just plough on ahead! Way past the point that a human would stop, and say "Wait a minute.. this seems wrong". This is an example of how algorithms not only reflect their teachers, but actually magnify their unconscious biases.
So, wait: did we just create a white people detector?
Well, our internal systems don't quite work like this - but I suspect the party has already created a white people detector. Councillors and candidates up and down the country trust our target pools to identify who we should speak to - but those pools are enriched by data from those councils and candidates that prioritise people in the target pools anyway.
This is an example of structural racism - nobody, individually, trying to be racist, but a system (perhaps seeded with data from a more racist previous generation) that nevertheless perpetuates racism.
I am sure people will arrive to critique this argument, and to defend the venerable algorithm. But I'm not going to ask you to scrap it - they have their uses. Instead, enrich it - in the months and years between "short" campaigns - by going and speaking to the people you never spoke to before, on the estates you never spoke to before. Home in on the hard conversations, step into the no-go estates.
Put away the target pools; there are Liberals, and liberally-minded people, in every street.
Go find them.