# 93% of Paint Splatters Are Valid Perl Programs

An empirical study explores the intersection of abstract art and programming language syntax.

- **Source:** [Hacker News](https://news.ycombinator.com)
- **Published:** 2026-11-11
- **Topic:** [technology](https://brifing.si/c/technology)
- **Original article:** <https://www.mcmillen.dev/sigbovik/>
- **On Brifing:** https://brifing.si/s/074d77ae-7c16-442e-b33e-45469b62d6a0

## The Core Research Question

Researchers investigated whether random paint splatters could function as valid Perl code. By applying optical character recognition to various artistic patterns, they sought to determine if visual chaos inherently contains executable logic. The study specifically aimed to answer if it is possible to create paint marks that fail to parse as Perl.

## Empirical Methodology and Results

The team utilized OCR software to analyze a dataset of 100 distinct paint splatter images. The empirical findings revealed that 93% of these splatters successfully parsed as valid Perl programs. This high success rate highlights the permissive nature of Perl syntax when interpreting arbitrary character strings generated from visual noise patterns.

## SIGBOVIK Conference Context

This paper was formally accepted for publication at the SIGBOVIK 2019 conference held in Pittsburgh on April 1st. The work received an Unwitting Participation Ribbon, a humorous distinction for papers containing genuine artifacts. The author documented their findings on page 174 of the official proceedings, contributing to the spirit of open science.

## Parsing Visual Data

The researchers provided a gallery showcasing the paint splatters alongside their corresponding Perl source code. For images that did not parse correctly, the text was highlighted in red. In instances where OCR settings yielded multiple interpretations, the author manually selected the most aesthetically interesting result based on their own personal judgment.

## Specific Code Evaluations

Several splatters yielded surprising results when processed through the OCR engine. For instance, one splatter was interpreted as the string lerzfijglpFiji-j, which evaluates to the number 0 in Perl. Another image recognized as -*? also evaluates to 0, demonstrating how the language handles unexpected or malformed input strings during the execution process.

## String Interpretation Examples

Further analysis revealed complex string evaluations from the splatter dataset. One specific image was recognized as the string ;i;c;;#\?z{;?;;fn':.;, which evaluates to the character c. Another example, ;E,'__', successfully evaluated to the string E__. These examples illustrate the diverse ways Perl interprets random character sequences derived from non-textual visual data sources.

## Data Availability Constraints

The author provided a tarball containing the 100 paint-splatter images used as the primary dataset for this research. However, the associated source code remains unavailable to the public at this time. The author cited personal difficulties with managing GitHub repositories as the primary reason for the current lack of accessible code documentation.

## Implications for Language Parsing

This study serves as a playful exploration of how programming languages interpret noise. While 93% of splatters were valid, the remaining 7% confirmed that not all visual patterns can be forced into executable code. It underscores the unique, often unpredictable nature of Perl's parser when confronted with non-traditional, image-based input data structures.

## Sources covering this story

- https://www.mcmillen.dev/sigbovik/ — 

---

_Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._