Can machine learning or algorithms be used in an electoral process?

Guided by the “keystroke megaphone” principle, I decided to post this email reply I just sent.

An anonymous reader emailed me the following question:

Is it possible to use machine learning or algorithms in a campaign or in an electoral process?

I wrote the following reply:

I don’t have any personal experience with this, but I do know that all kinds of algorithms have been and are still used to influence elections, in the US and in the UK, and probably in other countries.

If you search online for “Cambridge Analytica”, you’ll find some of the quite immoral ways in which they used facebook and other social network data to influence public opinion.

(The Guardian has a great collection called The Cambridge Analytica Files.)

In short: With enough data and some clever “big data” analytics, it is possible to determine with quite some precision which groups to target with personalised advertising campaigns, in order to get whole cities or important states to vote for your candidate.

Usually this form of analytics is coupled with social media bots, that is to say, a large number of “fake” social media accounts that say some or other thing.

These efforts can be further boosted with purpose-created websites on legitimate-sounding domains.

The websites and fake accounts strengthen their combined message by citing each other.

One could go further and even apply psychology to craft these messages very carefully, so that they have an even higher success rate.

Folks forward these messages in their various forms via communication networks such as whatsapp (2 billion users this month) and facebook. The poisoned messages are further strengthened because new victims receive them from trusted sources, their social connections.

In an even shorter form: The human mind and its social context have been hacked and are being actively exploited.

Kind regards, Charl

Have I mentioned that I sometimes think about doing a startup that employs all of the above techniques and more For Good™, where with For Good™ I mean for all ideas that align with my perfect worldview?

P.S. Fake news is just the beginning, 2017. is quite relevant.

Contents