Friday, February 7, 2020

Self Harm via Too Much Information?


Over the summer, I had the opportunity to intern at a large job board site that had millions of users visiting it every month, and my goal was to show more relevant jobs to the user.

One idea was to allow the user to "dislike" and remove a job posting that they didn't like, and the confirmation read "we'll stop showing you jobs like this". During user research, some jobseekers were hesitant to use this feature because they didn't know what "jobs like this" entailed, and were afraid it would prune out jobs postings that they may have wanted to apply to. 


On the other hand, the usage data showed that users who used this feature saw more relevant jobs, indicated by the increase in the job click through rate & the number of applies.




Empowered by this data, I changed the confirmation to "This job will no longer appear in your search results" in the next iteration. It no longer mentioned anything about affecting other job postings similar to it, but I knew we would still stop showing similar jobs. Again the usage data showed increase in the positive metrics, but this time even in larger amount due to the change in the wording.

Now learning about the ethics of information transparency, it makes me think deeper if withholding information to encourage a certain behavior was unethical, even if the outcome was positive (connecting users with more relevant opportunities). The utilitarian argument would be that the ends justified the means, but the Turili-Floridi Information Transparency reading provides more nuance in this matter. 


                        

It states that this matter of transparency is not an ethical principle in itself, rather the disclosed or undisclosed information (confirmation wording) has an impact on other ethical principles. In this scenario, not disclosing the algorithmic implications of disliking a job may prevent accountability on my part of potential lost opportunities for that jobseeker. Perhaps I should've even shared the usage data that helped me justify the change of wording in the first place, reciprocating the information empowerment back to the users so that they could make an informed decision?

2 comments:

  1. Hi Kevin! I can definitely relate to this on a personal level back when I was applying to my first internship. I would just mass apply everywhere and occasionally filter out positions or companies that I didn't want to work at. In my opinion, I do not think that it is an ethical problem if the user is given the choice to dislike/remove a job or ones similar to it since it is their own personal decision. However, it would be more relatable to Floridi if you were, lets say, hiding/showing listings based on their clicks and applications without their approval. Overall, I thought your blog post dived into a unique topic and was interesting. The only thing I would say is to move your references to the readings further up and go into more detail instead of just briefly mentioning them at the end.

    ReplyDelete
  2. Hi Kevin! I really enjoyed reading your article, and could definitely relate to when I was earlier recruiting for finance internships. I would often filter out jobs based on location, company, roles, sponsorship requirements just to make it easier for me to apply to jobs I am interested in. You bring in an interesting point of view regarding the use of disclosed or undisclosed information on the company's side. To improve, I think you should integrate the readings more into the article in order to strengthen your argument on ethics. Maybe you could also bring in the ethics portion of the post earlier in the article as the introduction runs a little long. But overall great article!

    ReplyDelete

Note: Only a member of this blog may post a comment.