Creative Feedback Firewall
Let reactions inform the work without letting them program your creative judgment
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 4
- Confidence
- 95%
Dillon says feedback can shape a creator in two directions: criticism can derail them, while requests for more of a popular element can lead them to manufacture work they no longer enjoy. His answer is not to deny the validity of audience reactions but to stop consuming each reaction as a programming instruction. The filter uses three stronger signals: whether the work feels right to the creator, whether trusted people understand and enjoy it, and whether medium-specific behavior shows that audiences are choosing it. In comedy, Dillon mentions laughter and ticket buying; other fields need different measures. The mechanism protects creative intent while preserving evidence. A creator defines the standard, receives reactions without automatic obedience, triangulates taste, trusted judgment, and behavior, then revises only when the combined signal is meaningful.
Origin
Dillon explains that he limits feedback because both negative comments and audience requests can pull a creator away from what made the work enjoyable.
Core principles
- 01Every reaction is real to the person having it
- 02A reaction does not automatically become an instruction
- 03Trusted judgment and behavioral evidence outrank comment-by-comment demands
- 04Overfitting to praise can damage the work as easily as overreacting to criticism
How to run it
- 1
Set the intent
Before opening the feedback stream, state what the work is trying to do and what would make it successful. This creates a standard independent of the latest comment.
Pro tip Use one sentence for intent and two or three observable success signals.
Watch out An internal standard does not override factual accuracy or safety obligations.
- 2
Receive without obeying
Treat reactions as valid descriptions of individual experience, not automatic commands for the next piece. Group recurring observations instead of reacting one by one.
Pro tip Review feedback in batches rather than continuously.
Watch out Do not dismiss a correction merely because only one person noticed it.
- 3
Run the three-signal check
Compare your own craft judgment, the response of people you trust, and behavioral evidence from the audience. Look for convergence rather than letting one signal dominate.
Pro tip Choose metrics that represent real behavior, such as completion, return use, laughter, or purchases.
Watch out Popularity metrics can reward novelty or outrage without indicating quality.
- 4
Revise selectively
Change the work when the combined evidence identifies a clear weakness or opportunity. Preserve the elements that still serve the original intent, even if a vocal subset asks for something else.
Pro tip Write down the evidence behind each substantial change.
Watch out Refusing all change is rigidity, not creative independence.
In the wild
Listeners repeatedly ask a host to make every episode resemble one popular segment. The host logs the request, checks whether that format still feels creatively useful, asks two trusted collaborators what made the segment work, and compares completion and return-listener behavior across episodes. The evidence shows the segment works occasionally but becomes repetitive when overused, so it remains a recurring feature rather than the entire show.
→ The audience signal improves programming without replacing the creator's judgment.
Common mistakes
Treating every comment as a brief
Individual reactions conflict and can pull the work in incompatible directions. Aggregate the signal before changing course.
Overfitting to praise
Repeating what people liked can hollow out the work if the creator no longer finds it interesting or useful.
Calling all criticism noise
The firewall filters influence; it does not excuse ignoring corrections or repeated evidence that the work is failing its stated purpose.
Is it for you?
Best for
It is best for creators with enough audience response to risk overfitting their next work to individual reactions.
Not ideal for
It is not ideal as an excuse to ignore factual corrections, safety reports, contractual requirements, or consistent evidence of harm.
From the transcript
“I've tried to limit feedback because feedback can work in two ways.”
“you don't have to consume their reactions as a way to program your mind”
“Do people you trust and like enjoy it?”
From the episode
Tim Dillon (Comedian): "The Boomers Are A Selfish Generation!". "Gen Z Has Exposed Society's Scam!". "Bosses Are Being Threatened By Mental Illnesses!"