Words matter. The metaphors we use can lead us down paths that lead to very different, very unintended outcomes.
Recently, I heard of a project idea to use Machine Learning to improve a company’s recruiting efficiency. Using candidates’ social profiles (Github, Stack Overflow, LinkedIn, Quora), the team wanted to predict candidates’ likelihood of getting hired - thus allowing HR to target candidates who wouldn’t have otherwise known of the company. This also meant establishing a baseline of data using current and past employees’ profiles.
Luckily, enough people spoke up about the legal and moral problems with this idea that it was nixed. Just as a sampler:
- Sending unsolicited email to people is Spamming - is that likely to make candidates interested in your company? Or is it more likely to annoy them and send you to the Junk folder?
- By using current employee data as the baseline, you’re entrenching current hiring biases and skewed demographics into future hiring
- Github, Stack Overflow, LinkedIn, etc. have Terms of Service that say you can’t scrape their website for user data. You’d be in violation and likely to get blacklisted.
- For users in many countries this is a violation of privacy laws (e.g. Europeans will be protected by GDPR, Americans by CAN-SPAM and other laws.) You’d be in violation and likely to get penalized.
- … and on and on.
At this point it’s worth asking, why did the original team think this was a good idea in the first place?
During the discussion, the dev team kept insisting they only want to help by “finding hidden talent” - which I think is interesting.
Finding talent is the wrong metaphor here. Using the “find” metaphor makes it a search problem - one that technology can solve. It’s a fun problem to work on too - throw some AI/ML in there, and you’ve got a cool exciting project on your hands. But it is not what recruiting should be about.
Attracting talent is probably a better metaphor for what they should be trying to do. A small change of words, but it leads you down a completely different path of activities. More difficult, more time consuming, but ultimately more productive activities. Some people recommended these very smart ideas that would yield better results:
- Instead of going outward, look inward at the last 5 years of interview data for insights
- Offer a trail of ‘breadcrumbs’ that attract candidates to your products and services. Make your product roadmap transparent, then enlist employees to actively engage outsiders in discussions, Pull Request reviews etc.
- Engage in passive (vs active) recruiting efforts that resonate with developers in an authentic way
- Experiment with Organizational Psychology to determine the broken parts of the recruiting process
The abuse of Big Data by companies, and the lack of diversity in tech hiring, are two well-known problems of our industry at this point. It was inevitable that they would overlap at some point. It’s still amazing to me how quickly we tend to reach for the high-tech solution for sociological problems, instead of fixing their systemic roots. If your company ever starts to think about ideas like the one above, please share the following talks internally and encourage people to reflect on their lessons:
Carina Zona: Consequences of an Insightful Algorithm
Adria Richards: Why Tech Companies Are Failing To Attract Female Engineers
Heidi Waterhouse: The Death of Data