How do profiling algorithms rewrite history?
- Filed under
- Techno-Oligarchy
- Type
- Background entry
- Sources
- 9
Image: pixabay.com
Are profiling algorithms quietly creating new historical records that erase inconvenient lives and legitimize novel forms of social power?
Summary
The mainstream evidence emphasizes profiling as summarization, inference, and selective persistence. These systems can make large histories usable for personalization or auditing, but should distinguish temporary context from stable preference and expose the sources behind sensitive inferences. Here, “rewriting” mainly means changing representation and retention, not literally changing the past.
- Step-back Profiling improves personalized scientific writing by distilling user histories into concise profiles of traits and preferences.
- Source-linked profiling can make attribute inferences more auditable by tracing predictions to the posts and relationships that supported them.
- Writing every observed interaction into persistent profiles can harm alignment with users’ later behavior; selective write controls reduce that harm in the reported experiments.
Deeper Insight
Your profile is a history someone else wrote. You can't read it, can't argue with it, and it decides what you're offered next.
Every click, purchase and pause gets compressed into a profile, a short version of you a machine can act on. That version shapes hiring screens, credit and what your feed shows next, and it can carry old prejudice in its math. When it gets you wrong, you may not even have the words to say how. The past is still yours. The record isn't.
- What it could changeA right to see and contest your profile would turn a hidden verdict into an argument you could win.
- How it touches todayEngagement-driven feeds already decide which moments of the public record stay visible and which fade.
- Our futureHistory may increasingly be what the machines kept, not what happened.
- Algorithmic profiling can produce hermeneutical injustice when people lack the resources or standing to make profiling-related harms intelligible and contestable.
- Critical analyses argue that large-scale models can encode prejudice, operate opaquely, and impose unequal consequences in hiring, credit, and criminal justice.
- Engagement-optimized platforms can encourage prolonged attention and emotionally charged consumption, changing which traces and interpretations become prominent.
Sources · 9
Featured · Primary source · ar5iv.labs.arxiv.org
[2509.09294] Altered Histories in Version Control System Repositories: Evidence from the TrenchesThis work was supported by France Agence Nationale de la Recherche (ANR), program France 2030, reference ANR-22-PTCC-0001. This work was made possible by Software Heritage, the universal source code arch
[2509.09294] Altered Histories in Version Control System Repositories: Evidence from the Trenches # Altered Histories in Version Control System Repositories: Evidence from the TrenchesThanks: This work was supported by France Agence Nationale de la Recherche (ANR), program France 2030, reference ANR
Read sourceAnalysis · researchgate.net
Algorithmic profiling as a source of hermeneutical injustice
# Algorithmic profiling as a source of hermeneutical injustice Philosophical Studies. Published: 2024-02-05. 26 citations. ## Authors - Silvia Milano (University of Exeter): h-index 8; 1,069 citations; corresponding author - Carina Prunkl (Utrecht University): h-index 9; 672 citations ## Topics - Fe
Read sourceAnalysis · worldhistory.medium.com
Historians Complicate; AI Flattens | by George Dillard | Medium
Historians Complicate; AI Flattens | by George Dillard | Medium Sign up Get app Sign up Member-only story # Historians Complicate; AI Flattens ## AI is shaping ourunderstanding of the past, and not in a good way 8 min read Feb 21, 2026 -- 19 Share Press enter or click to view image in full size Scre
Read source