Oriel — AI & technology worth noticing, every day.

Humanizer has about 60,000 stars. On 12 public tasks my linter plugin scored 2.67 higher

Post · https://dev.to/efe_genc/humanizer-has-about-60000-stars-on-12-public-tasks-my-linter-plugin-scored-267-higher-2cdg

ai-slop-linter against six humanizer tools and no tool: 12 public tasks, graded blind, 58.67 of 60 against 56. On a held-out set the top tools tied.

First seen by Oriel on .

AI summary

A benchmark found the ai-slop-linter outscored six AI-writing cleanup tools and a no-plugin baseline on 12 public tasks, but not on six held-out tasks.

  • Across 11 arms and three Sonnet 5.5 runs per task, the linter scored 58.67/60 versus 56.00 for Humanizer called by name.
  • The linter uses deterministic rules, no model, and --compare to detect lost facts, qualifiers, and newly added first-person or claim words.
  • On held-out tasks, unslop scored 30.00/30 while the linter scored 29.67; four arms scored at least 29.67, limiting separation.
  • The linter cost 2.43 times the no-plugin baseline, or $0.125 versus $0.051 per public cell at list price.
  • The benchmark, harness, evidence, and recomputation steps are public, with reproduction requiring Claude Code 2.1.308, Node 24, and about nine hours.

#ai writing #claude code #linters #benchmarking

Generated by gpt-5.6-sol. Check the original for details.

Metric trends

Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained

Reactions

Total public reactions reported by the source; distinct from points and likes.

  • DEV Community: 13, latest observation ; +2 over observed period since

Comments

Comment count reported by the source.

  • DEV Community: 22, latest observation ; +7 over observed period since

Aliases

SourceExternal IDURL
DEV Community4909647https://dev.to/efe_genc/humanizer-has-about-60000-stars-on-12-public-tasks-my-linter-plugin-scored-267-higher-2cdg Read discussion