Twitch has different 'boosting' features available to some streamers, such as channel point boost challenges or the new Boost Train. While great in theory for discoverability, completed boost recommendations right now seem to be sent to completely random users.

This means that recommendations are not always seeking the audience that the streamer is hoping to reach, or is content viewers aren't looking for. For example, when checking my own Front Page for these boosted recommendations, I've mostly been recommended channels streaming in languages I do not speak, or in categories I've never engaged in as a viewer or streamer.

This randomness also puts marginalised streamers at higher risk of harrassment as a result of financial support from their communities. For example, when my community completed a Level 4 Boost Train, the only new chatters in my stream were trolls targeting me for being a trans person. Hype Trains are designed to be a moment of joy, not a time where we have to worry more about being harrassed.

Instead, Twitch should send recommendations from boosts out based on user data that Twitch already has (e.g. how they filter 'live channels we think you'll like' and other recommendations). Curating recommendations based on these aspects or other user data available make them more appropriate and meaningful for viewers and streamers alike.