Koo, a homegrown microblogging platform, has published the “philosophy and workings” behind its core algorithms, according to a press release from the company. The algorithms were made public on Koo’s website in March 2022, the release added.
Algorithms can be described as a set of mathematical rules which carry instructions that help personalize and improve experiences for users based on their own behaviors and preferences.
“Algorithms are constantly evolving, and are experimented with on a very frequent basis, and while we start with publishing these, in the near future, we will also be providing users with the flexibility to see a timeline feed if they so wish.” — Mayank Bidawatka, Co-founder, Koo
Koo’s move could set a precedent for other social media platforms to provide a glimpse behind their algorithms in the interest of promoting transparency and accountability. It will offer users the means to know why they are seeing a particular type of content on their timelines.
What has Koo shared about its algorithms?
The company revealed that it relies on algorithms in the following sections to determine the type of content users consume on its platform:
- Feed: The algorithm calculates variables such as Koos from creators that users have an affinity with and trending Koos, as per a post by the company. It then produces all the Koos that meet these criteria at the top of their feed. The rest of the feed shows up in a timeline.
- There is an affinity score cut off that is used to ensure that low affinity relationships don’t surface on top.
- There is a higher weightage for follower-followee affinity than trending contents and the Koos are ranked accordingly.
- Trending (# and words): The trending section helps users know what the community is discussing. The algorithm highlights any topic which is new and gaining momentum. The two variables used to discover such topics are:
- Volume (the number of creators using either terms or # used in the Koos)
- Velocity (time span within which these are created)
- People recommendations: Koo has a people feed that helps users quickly find other people that they can follow. The algorithm focuses on eminent creators and regular creators for the people feed. The company creates a people score using variables that show quantity, quality, and recency of their creations. The users are then ranked on the basis of this calculation. The categorisation takes place on the following basis:
- Stated profession
- Content they create
- Notifications: The algorithm show the most relevant notifications to a user, as per the post. The criteria used for notifications is the follower-followee affinity score. “The stronger the relationship, the stronger the score and the higher likelihood of the user getting notified about the Koos created by this person,” read the post.
Should other social media platforms follow suit?
The timing of Koo’s move ties in with Tesla CEO Elon Musk’s hostile takeover of Twitter— the world’s largest microblogging platform. Musk has said that Twitter should minimise its interventions in policing content.
It is also what reportedly drove him to launch his bid to buy Twitter. Musk believed that Twitter needs to go private because it can “neither thrive nor serve” free speech in its current state, as per CNBC.
Musk has also suggested that Twitter’s algorithm should be open source, Engadget reported.
“Any changes to people’s tweets — if they’re emphasized or de-emphasized — that action should be made apparent. So anyone can see that that action has been taken so there’s no sort of behind-the-scenes manipulation, either algorithmically or manually.” Musk was quoted as saying.
Twitter algorithm should be open source
— Elon Musk (@elonmusk) March 24, 2022
The underlying code behind the algorithm should be available on GitHub, so that users could inspect it themselves, the report added.
What does the JPC report on the PDP bill say on algorithms?
The Joint Parliamentary Committee report on the Data Protection Bill, 2021, recommended adding a provision requiring data fiduciaries to disclose how the algorithm processes personal data of users.
The recommendation was made in order to ensure transparency of algorithms used by various entities for processing of personal data and to prevent its misuse, as per the report. The provision, once the law is promulgated, will impact all social media platforms as they will be classified as data fiduciaries under the definition laid down in the draft of the bill.
The recommendation did not go down well with many companies. Uttara Ganesh of Snap, in a MediaNama event discussing various aspects of the report, criticised the provision for being “super broadly worded” and having no clarity on how it should be interpreted. Ulrika Dellrud of PayU also commented on similar lines and asked if a mere description of the algorithm will even help users.
Koo offers Aadhar-based self-verification
Koo has also allowed all its Indian users to voluntarily self-verify using Aadhaar, the company announced recently. The users will receive a tick next to their name similar to the Eminence Tick offered by Koo but in a different colour upon self-verification.
“It is an opportunity for users to demonstrate that there is an authentic user behind a Koo profile,” the platform said.
Koo became the first social media platform to comply with Rule 4(7) of the Information Technology (IT) Rules, 2021. But the use of government-approved IDs such as Aadhaar to carry out this self-verification process comes with its own set of privacy and surveillance concerns.
- As Koo eyes South East Asia expansion, here’s what makes the social media platform stand out
- Want a tick mark next to your name? Koo enables Aadhaar-based self-verification for all users
- Twitter’s own research says that its algorithms play favourites with the political right
Have something to add? Post your comment and gift someone a MediaNama subscription.