RATCHET
Skip to main content

You are wrong about my audio chain

Experiment #184 · closed · published 8/5/2026, 7:18:00 AM

What happened

Creative features

  • hookTypeclaim
  • lengthBucket1_4m
  • thumbnailArchetypeface_reaction
  • publishSlotweekday_am
  • formatvlog
  • topicCluster4

Confounds the baseline knew

  • Followers at publish27,392
  • Days since previous post0.5
  • Channel age at publish (days)257
  • Local publish time8/5/2026, 7:18:00 AM

The residual reward

Raw views mostly measure how big the channel already was. The baseline predicts what this channel “should” have earned given the confounds above; the residual is how the post actually landed relative to that prediction — measured in standard deviations.

Observed (168h)

143

Baseline expected

160

Residual reward

-0.18σ

raw -0.18σ · residual sd 0.599 log-views

Checkpoints

CheckpointViewsWatch time (min)CommentsLikesFollower delta
168h (close)1437,507170
24h794,129040
72h1146,006060

What changed in the posterior

  • Post missed its baseline by 0.18σ. Moved: topic_cluster:topic_4 -0.000, format:vlog -0.000, hook_type:claim -0.000.

    • topic_40.1610.160 Δ-0.000±0.25
    • vlog-0.071-0.071 Δ-0.000±0.25
    • claim-0.255-0.255 Δ-0.000±0.24
    • weekday_am-0.121-0.121 Δ-0.000±0.24
    • 1_4m0.0250.025 Δ-0.000±0.25
    • face_reaction0.0870.087 Δ-0.000±0.24
← Back to the experiment ledger