{"id":20066437,"url":"https://github.com/apparebit/penal-colony","last_synced_at":"2026-03-19T13:24:08.572Z","repository":{"id":159502574,"uuid":"592158588","full_name":"apparebit/penal-colony","owner":"apparebit","description":"Letters from the Stochastic Penal Colony 🏝","archived":false,"fork":false,"pushed_at":"2023-07-10T10:10:00.000Z","size":69211,"stargazers_count":1,"open_issues_count":0,"forks_count":0,"subscribers_count":2,"default_branch":"boss","last_synced_at":"2025-09-09T09:29:59.315Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"TeX","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/apparebit.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2023-01-23T04:09:08.000Z","updated_at":"2023-04-21T16:09:27.000Z","dependencies_parsed_at":null,"dependency_job_id":"d0dc3347-04e4-4525-9ef9-2350bb3e9ee5","html_url":"https://github.com/apparebit/penal-colony","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/apparebit/penal-colony","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/apparebit%2Fpenal-colony","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/apparebit%2Fpenal-colony/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/apparebit%2Fpenal-colony/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/apparebit%2Fpenal-colony/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/apparebit","download_url":"https://codeload.github.com/apparebit/penal-colony/tar.gz/refs/heads/boss","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/apparebit%2Fpenal-colony/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":30108647,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-03-05T03:32:43.378Z","status":"ssl_error","status_checked_at":"2026-03-05T03:32:22.667Z","response_time":93,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.6:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2024-11-13T13:58:21.082Z","updated_at":"2026-03-05T03:33:10.660Z","avatar_url":"https://github.com/apparebit.png","language":"TeX","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Letters from the Stochastic Penal Colony 🏝\n\n[Paper](https://github.com/apparebit/penal-colony/raw/boss/penal-colony-v1.1.pdf)\n(v1.1, PDF) by Robert Grimm, Independent Investigator, Brooklyn, NY, USA.\n\n\n## Abstract\n\nThis paper serves as pointed critique of algorithmic practice outside the\ncriminal injustice system. Far too many interventions including social media's\ncontent moderation are excessively punitive, often resulting in the figurative\ndeath of users through permanent account suspension. First, based on my own\nexperiences and grounded in procedural justice, this paper starts by exploring\nthe many ways policy and automated enforcement turn punitive on the example of\nOpenAI's DALL•E 2. Second, it illustrates how even best-practices policy turns\npunitive performance on the example of pre-Musk Twitter. Third, a comprehensive\nsurvey of non-Chinese social media demonstrates the pervasiveness of excessively\npunitive content moderation. It also tests the limits of their accountability,\nnotably by projecting the likely impact of the European Union's Digital Services\nAct and by correlating data released by Facebook, Google, and the National\nCenter for Missing and Exploited Children. Fourth, to illustrate the limits of\nalgorithmic content moderation, this paper presents a successful strategy for\nsubverting DALL•E's aggressive automated censor, which inadvertently also\nunleashed grotesquely racist imagery. Fifth, this paper proposes a new\nintellectual property regime specifically for AI. It re-combines proven\nelements from copyright and patent law, resulting in a framework that balances\nthe interests of those who invest in state-of-the-art AI and everyone else.\nFinally, this paper concludes by pointing towards harm reduction as a mindset\nfor, possibly maybe, making life in this digital penal colony at least somewhat\nbearable—because, I fear, we are stuck in it.\n\n\n## Findings\n\nHighlights of the paper’s findings include:\n\n  * Content moderation by all surveyed social media is punitive and excessively\n    so. Social media are on the best way to creating a new underclass of people\n    without a voice on these platforms.\n  * Content moderation by all surveyed social media runs against the public\n    interest. Particularly prohibitions against misinformation are extremely\n    chilling given pervasive failures by medical experts during the pandemic.\n  * Transparency reports by all surveyed social media besides Reddit and YouTube\n    suffer from significant data quality issues.\n  * Transparency disclosures by Meta are so ridden by data quality issues to be\n    wholly untrustworthy and meaningless. Unfortunately, that is the case for\n    Meta’s data disclosures to researchers and customers as well.\n  * As demonstrated on OpenAI’s DALL•E 2, algorithmic censors based on large\n    language models are vulnerable to a new kind of attack strategy that is hard\n    if not impossible to mitigate.\n  * As demonstrated on ChatGPT, large language models can significantly simplify\n    and shorten the experiments necessary for that attack, raising significant\n    doubts about the efficacy of AI-based content moderation.\n  * OpenAI’s DALL•E 2 produces deeply racist images without being prompted to do\n    so, most likely due to a naive diversity mitigation.\n\nThe paper explores regulatory responses to this sorry state of content\nmoderation and transparency reporting but rejects them as too punitive. Instead\nit points towards more subversive, harm-reducing approaches to dismantling the\nstochastic penal colony. It also proposes a new intellectual property regime for\nAI that remixes existing, proven copyright and patent provisions to ensure that\nall of society benefits from this amazing new technology.\n\n\n## Source Code and Supplements\n\nSource code and supplements for the paper “Letters from the Stochastic Penal\nColony 🏝” by Robert Grimm.\n\n  * A [__custom build script__](build.sh) in the repository root takes care of\n    repetitive tasks. The one optional argument is the name of the task to\n    execute.\n  * By default, i.e., when invoked without argument, the build script runs\n    `pdflatex` and `bibtex` to __create the PDF document__ from the LaTeX files\n    in the [source](source) directory.\n  * Since LaTeX and BibTex are incredibly noisy in their output, the build\n    script contains custom logic to __detect actionable warnings__ and then\n    error out.\n  * To work with the ACM's new (but arguably not improved) publishing flow, the\n    paper uses only approved LaTeX packages and __compiles with `pdflatex`__. To\n    produce my own copies, it also compiles with `lualatex` when the build\n    script is given the `lua` argument.\n  * Unfortunately, that leaves only one subpar option for __color emoji__,\n    namely simulating them by including graphics files. I wrote my own LaTeX\n    package, emo, to take care of that and then some. Emo is included with the\n    paper sources, but may be outdated. Check out [its\n    repository](https://github.com/apparebit/emo) or\n    [CTAN](https://ctan.org/pkg/emo).\n  * Transparency data and Jupyter notebooks with the code for __analyzing the\n    data__ are inside the [supplements](supplements) directory.\n  * The build script assumes that the __virtual environment__ with Python\n    packages necessary for running the notebooks is contained in the `.venv`\n    directory. When invoked with the `venv` argument, it checks whether that\n    directory exists, creating the virtual environment and installing packages\n    otherwise, and then activates the virtual environment.\n\n\n## (C) Copyright 2023 by Robert Grimm\n\nThe shell script and Jupyter notebooks included in this repository have been\nreleased as open source under the Apache 2.0 license. Otherwise, all rights are\nreserved, including for the paper itself.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fapparebit%2Fpenal-colony","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fapparebit%2Fpenal-colony","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fapparebit%2Fpenal-colony/lists"}