Meta end to end encryption
technical program manager
Meta - Cross Integrity > Interactions Solutions. London (2022 - 2023)
the problem
Facebook and Instagram messages (Messages and Instagram Direct) had to move to an encryption communication model due to regulation.
The end to end encryption (E2EE) could create a potential increase in risky interactions.
We wanted to find ways to still keep the integrity levels high in our platforms after the end to end encryption.
There was not a single solution for every type of harm that can happen.
There was the need to have a well defined roadmap, with clear timelines, priorities and owners. I was hired to support this multiyear Program.
The contribution
Reduction 15% Risky Thread Prevalence. 20% efficiency in solution implementation
* Proposed and implemented the reutilisation of existing solutions, such as image banking of group thread images and for group thread names for different types of harms.
* Definition of a new integrity metric called Risky Thread Prevalence (RTP). Daily % risky 1 to 1 threads by platform, without reactions). Reduction of 15% on RTP after implemented enforcement solutions.
* Creation of automated real time dashboard reporting of key integrity metrics.
* Presenter to Monthly ExCo.
* Roadmap creation, definition of milestones and priorities based on harm severity and potential increase of risky interactions.
* Convinced 10 teams from the Org (Product Managers) to prioritise this Program into their 2022 deliverables based on E2EE timings, rollout and amount of people being cut from threads.
* Drove the implementation of features to detect harm and enforce solutions (based on a combination of tactical and strategic) by coordinating Engineering teams.
the learnings
When a large scale and complex program is in place, having everyone on the same page is key. Regular, clear and precise communication goes a long way.
When you understand the technical challenges of each feature (in this case, each harm type), you might find solutions that can solve similar issues in similar places. In this case, we reused existing solutions for other Harm areas that were facing similar challenges.
If I learnt something at Meta, is to create impact. To show it, you always need a metric to target. This led me to get more into data and analytics, and with the support of Data Scientist we created a new north star metric to target - Risky Thread Prevalence.
Identify as soon as possible the people that need to ratify your roadmap, prioritisation and approve your features. In this case, building trust and a great working model with legal and privacy allowed us to move along with no delays.