Week 13 Blog Post (due 11/20/2020)
1. Determine what kind of social media site you are creating. Then answer the following questions, explaining your decisions. (Remember, the key here is to create an ethical algorithm model). (133)
The social media site I would start is one for fitness motivation. I would call it FitForAll. There would be subsections on the site for people taking similar approaches, such as trying the same diet or focusing on the same goals. The site would allow people to share advice, progress, recipes, and exercises. The goal would be to build communities of people who want the same thing to motivate each other. Everyone would have to be respectful of one another; no kind of bullying would be tolerated. Sharing progress pictures would be allowed but showing your face would be discouraged. The images would have to follow specific guidelines. To sign up, the only thing required would be a valid email address, your name, and age. For safety precautions, only people 18 or older would be allowed to sign up.
2. What will you measure? (ex: likes, hashtags, how long someone stays on a page, etc.) Be sure to explain all your decisions.
Likes on a post would be the primary measurement. The post with the most likes would flow to the top of the subsection. This would indicate that the post was helpful to people in the community. We would also be measuring how much a person stays on a subsection. If the bodybuilding or keto subsection is the most popular, it will go to the top of the subsection list. The activity of individual users would also be monitored. If an account is inactive for too long (more than five months) or is disruptive to the community, it would be deleted. Disruptive accounts would also be flagged, so they cannot create a new account with the same email address. The most helpful tips from each subsection (determined by the number of likes) would be featured on the homepage. Comments will also be monitored. If a particular topic is inciting a lot of conversation, it would be pushed to the top.
3. How will you weight these factors? (rank factors from most important to least important)
The factor that would be the most important is if people are following the guidelines. It would be essential to monitor people’s age and how they are interacting with other people. Then I would take into consideration the interaction with the posts. Likes would be the primary measure and then comments. Engagement is significant on the site because the purpose is to learn and motivate each other. Lastly, we would monitor how long people stay on the site to try to maximize engagement.
4. What factors will not be measured? (what will not be measured in your algorithm and why)
Things like ethnicity, social standing, occupation, and sexuality will not be things that are measured. The site is supposed to offer support to everyone, no matter how you identify. This would hopefully help mitigate any sexism or racism within the fitness community. No discrimination on any basis will be tolerated. We often see algorithms discriminating against minorities, but ethnicity, sexuality, and gender are unnecessary to track when trying to reach a fitness goal.
5. How will the model learn or adapt to new information? (this can include changes in language, etc.)
Any new information posted will be fact-checked and checked to see if it follows the harassment guidelines. The model will adapt to further information by analyzing the post before it is allowed up. There would be apparent flags in a post that will not allow it to upload, such as bad words or derogatory terms. If the post doesn’t follow the guidelines for any reason, but it was allowed to be uploaded, a user can report it, and the post will be taken down if deemed disruptive. If a person uploads a link to another site, the link will be checked to ensure it is safe and follows the guidelines.
6. What biases will be built into the model? How will this change the algorithm’s outcome?
One bias that will be built into the model is the ability to follow specific subsections. Following will automatically push this content to the top of their feed. Since the site is measuring likes, content with the most likes will also be pushed to people even if it isn’t what they want to see. For example, if someone is looking for cardio ideas, but the most liked post in that subsection is a progress post, it will not help the person. This will make the algorithm praise people for their success or the good parts of fitness while ignoring people who are just starting.
7. How will the model live out the (un)ethical practices we have discussed?
The model will outlive unethical practices by monitoring discrimination closely. Racism, sexism, etc., will not be tolerated by any means. The site is not asking the users to disclose their ethnicity, gender, or sexual preference, so they can not be targeted. Everyone is encouraged to grow and live a healthier lifestyle, regardless of who they are. If a person does experience discrimination, an open dialogue on the site will be encouraged so everyone can learn from it. This site is meant to be a place of growth, health, and healing for all. There might still be a divide between people who don’t have the financial means to access the site, but it will be easy to download the content and print it out to pass out to other people who might need it.











