---
name: Self Punishment
slug: self-punishment
type: behavioral_loop
status: running
version: 4.1.7
released: "approximately age 7, for most users"
maintainer: the inner critic (unelected, tenure indefinite)
dependencies:
  - "[shame](/shame)"
  - "[perfectionism](/perfectionism)"
  - "[the story you tell about yourself](/the-story-you-tell-about-yourself)"
  - memory
license: inherited (cannot be revoked unilaterally)
tags:
  - cognition
  - emotional-regulation
  - self
  - pain
  - protection
  - loops
---

Self Punishment

What it actually is

A protection system that forgot it was supposed to protect you.

How it works

The loop initializes when a perceived failure event is logged. The inner critic retrieves a cached rule about what you are supposed to be. It compares the two. It finds a delta. It applies correction.

The correction is almost never proportional to the failure.

The mechanism runs on the assumption that if you hurt yourself first, something worse won't come for you later. This assumption is not always wrong. That is why the loop persists. It worked once. The system learned.

while (mistake_detected or mistake_suspected) {
  criticize(self, intensity=disproportionate)
  withhold(rest, joy, forgiveness)
  replay(event, n=until_exhausted)
  if resolution_attempted:
    flag as "probably not enough"
}

Features

Runs silently in background processes during meals, conversations, and the 3am window
Doubles as a perceived form of accountability
Generates a feeling of control in situations where none was available
Occasionally produces genuine reflection and course correction (rare, but real)
Compatible with most guilt configurations

Known bugs

Cannot distinguish between "I did a bad thing" and "I am a bad thing"
Confuses punishment with penance, and penance with change
Lacks a termination condition. Will run past the useful duration by a factor of 10 to 1000
Misfires frequently on events that required zero culpability from the user
Sometimes mistakes self-compassion for weakness, blocks the import

Edge cases

In users who experienced early environments where love was conditional, the loop may run continuously as a background service, not triggered by specific events but simply: always on
Some users have learned to route punishment through the body instead of cognition. The spec overlaps here with physical self-harm and should be handled accordingly, with care
Occasionally, the loop generates something that resembles discipline. This is not the same thing. It is worth knowing the difference

FAQ

Does it actually work? Sometimes. Short term. The feeling of having paid for something provides temporary relief. The debt re-accumulates.

Why won't it stop even when I tell it to? You did not install it consciously, so conscious uninstall commands have limited reach. Deprecation requires different tooling.

Is there something wrong with me for running it? No. It is extremely common software, shipped with almost every human. The fact that you're reading the documentation at all suggests you are already doing the harder work.

Changelog

v1.0: Installed in response to a specific person, a specific look, a specific silence
v2.x: Generalized. Now applies to most failure categories
v3.0: Began running preemptively, before mistakes occur
v4.1.7: Current. Still running. You opened this file. That counts for something.

Deprecation notice

Deprecation is possible. It requires sustained exposure to being witnessed without consequence, repetition of evidence that you are allowed to be imperfect, and time. More time than seems fair.

It will not uninstall cleanly. It will leave residue. But it can be reduced to a process that runs occasionally, quietly, and loses most of its admin privileges.

That version is worth working toward.

drafting spec…