Sample

Dorn/cuda/109

Human rating uses the original dataset scale.

DatasetDorn

Human4.16

Lines10

Scores

Method	Score	Scale	Dataset rank (0–100)
Posnett	0.9	0–1; higher is more readable	89.1Δ 7.0
Scalabrino	0.737	0–1; higher is more readable	80.2Δ 15.9
Dorn	0.64	0–1; higher is more readable	55.7Δ 40.4
Mi	0.701	0–1; higher is more readable	84.1Δ 12.0
LLOC	1	Logical lines; fewer is more readable	97.9Δ 1.8
DeepSeek V4 Pro	15.33	0–20; mean of three runs	79.9Δ 16.2
GPT-6.1 Sol	17	0–20; mean of three runs	80.4Δ 15.7
BTTF	0.603	0–1; higher is more readable	74.4Δ 21.7
Human	4.16	1–5; original rating	96.1

Views

Download official ZIP ↗

Select the downloaded ZIP or its extracted folder. Files stay in your browser.

Source code

Official dataset source · Dorn/cuda/109

Loading feature diagnosis…

Source regions

Feature contributions

Feature	Value	Contribution (log-odds)
Byte entropyHow varied the bytes in the source text are. Repeated bytes give a lower value; a more even mixture gives a higher value.	5.107	−7.66
Source linesThe number of non-empty source lines.	10	+4
Halstead volumeHow much symbolic information the code contains. It combines the total number of operators and values with the number of distinct ones.	91.377	−3.015

Intercept 8.87 + feature contributions = 2.194 (log-odds).
Score = sigmoid(2.194) = 0.9.

Score	0.737

Feature contributions

Feature	Value	Contribution (log-odds)
Identifier layout (horizontal)How rapidly the visual pattern of identifier characters changes from left to right. It measures spatial frequency, not the meaning of the names.	33.909	+2.229
Operator visual areaThe share of the code’s visual area occupied by operators such as +, =, and *.	0.195	−0.954
Comment layout (horizontal)How rapidly the visual pattern of comment characters changes from left to right.	6.182	−0.705
Indentation frequencyHow rapidly indentation changes from line to line.	1	−0.703
Comment layout frequencyHow frequently the layout switches between lines containing only comments or whitespace and lines containing code.	11	+0.647
Identifier layout (vertical)How rapidly the visual pattern of identifier characters changes from top to bottom.	9.943	−0.271
Number layout frequencyHow rapidly the number of numeric values per line changes through the code.	7	+0.26

Intercept 0.07 + feature contributions = 0.574 (log-odds).
Score = sigmoid(0.574) = 0.64.

Score	0.701

Score	1

Run 1

The code is highly readable. It has a clear, logical structure with a comment separating sections, defines constants with descriptive names, and includes a useful macro. It is flat, concise, and each line expresses a single intent. Minor deductions for potential macro pitfalls and commented-out include.

Run 2

The code is short and readable but suffers from incomplete context. It includes commented-out includes, two well-named constants, and a simple MAX macro. Structure is flat and clear, with one action per line. However, intent is ambiguous: the constants' purposes are unexplained, the commented include is unclear, and the macro lacks parentheses around parameters, which is a correctness risk. Reading flow is natural, but the file appears incomplete.

Run 3

The code is minimal and mostly readable but has minor issues. It includes a commented-out include statement, which can cause confusion about whether it is needed. The constants MAX_EPSILON_ERROR and THRESHOLD use float literals with an 'f' suffix, but MAX_EPSILON_ERROR is defined as 5.00f while THRESHOLD is 0.30f — both are fine, but consistency in formatting (e.g., 0.30f vs 0.3f) is slightly inconsistent. The MAX macro is unsafe because it evaluates arguments multiple times and lacks parentheses around parameters in the expression; it could cause side effects or precedence bugs. The overall structure is logical: system includes (commented), constants, then macro. Nesting is flat, intent is clear, length is concise, and each line performs one action. Reading flow is natural. The score is lowered mainly due to the unsafe macro implementation and the leftover commented-out include.

Run 1

Code Structure: Simple, logically grouped definitions, although the broad header comment and commented-out include add clutter. Nesting: No nested control flow; the ternary expression is straightforward. Clarity of intent: The constant names are descriptive, and MAX clearly indicates its purpose, but the constants lack contextual explanations. Code length: Very concise, with minor unnecessary blank lines and inactive code. Action granularity: Each definition occupies its own line. Reading flow: Natural top-to-bottom progression, though inconsistent spacing slightly reduces polish. The MAX macro would be clearer and safer with individually parenthesized parameters.

Run 2

Code Structure: Simple, logically grouped macro definitions, though the includes heading and commented-out include add clutter. Nesting: No nested control flow; the conditional macro is easy to scan. Clarity of intent: MAX_EPSILON_ERROR is descriptive, but THRESHOLD lacks context, and the constants' purpose is unexplained. Code length: Very concise, with a few unnecessary blank lines and an obsolete-looking include comment. Action granularity: Each active line defines a single macro. Reading flow: Straightforward top-to-bottom flow, slightly weakened by inconsistent alignment and spacing.

Run 3

Code Structure: Simple organization, though the includes heading and commented-out include add clutter. Nesting: No nested control flow; the macro expression is easy to follow. Clarity of intent: Constant names convey general roles, but their context is unspecified, and MAX_EPSILON_ERROR is less precise than the other names. Code length: Very concise, with slightly excessive blank spacing. Action granularity: Each definition occupies its own line. Reading flow: Natural top-to-bottom flow, with minor formatting inconsistencies between definitions. The MAX macro would be clearer and safer with individually parenthesized arguments.