Thinking about frames and how their current implementation could be vulnerable to abuse: 

1. Bait & switch: Frame returns one embed at first and after getting a sufficient number of reactions begins displaying something that is malicious. This could be used to make it appear that users are agreeing with a sentiment.

Head of Blockchain Tech @ Joyride Games 💙 Created the First Meme on Base🔵 Technical Advisor & Contributor @ Velo/Aerodrome

3. Funnel: Initial embed of a frame shows rotating content and gauges when that content is engaged with. Server starts with hundreds of possible frames and is able to funnel towards the single frame content that will get the most reactions. This could artificially create high rates of popular content for the account.

4. Troll: Frame displays content that is likely to engage well for a wide variety of users. Secretly it displays something controversial, offensive or otherwise negative to a small random subset of users. Bonus points if frames content is structured to where replies seem like they could be to either, causing confusion.

Anyone have any other ideas? I think at this stage in development it's time to put on the white hats and see if there are any unexpected effects.

2. Mirror: Frame embed prompts a click-through to reveal content. Implication is that the content is the same for all users and is just hidden by the click. In reality, the click allows the embed to display content tailored for the user which makes them more likely to react.