Sample

Buse/91

Human rating uses the original dataset scale.

DatasetBuse
Human2.09
Lines7

Scores

MethodScoreScaleDataset rank
(0–100)
Posnett00–1; higher is more readable
2.0Δ 0.0
Scalabrino0.2140–1; higher is more readable
0.0Δ 2.0
Dorn0.230–1; higher is more readable
13.1Δ 11.1
Mi0.4010–1; higher is more readable
28.3Δ 26.3
LLOC6Logical lines; fewer is more readable
11.6Δ 9.6
DeepSeek V4 Pro70–20; mean of three runs
3.0Δ 1.0
GPT-6.1 Sol5.670–20; mean of three runs
0.0Δ 2.0
BTTF0.2970–1; higher is more readable
9.1Δ 7.1
Human2.091–5; original rating
2.0

Views

Download official ZIP ↗

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

Loading feature diagnosis…

Source regions

Feature contributions

Feature contributions

FeatureValueContribution
(log-odds)
Halstead volumeHow much symbolic information the code contains. It combines the total number of operators and values with the number of distinct ones.413.594−13.649
Byte entropyHow varied the bytes in the source text are. Repeated bytes give a lower value; a more even mixture gives a higher value.4.856−7.284
Source linesThe number of non-empty source lines.7+2.8

Intercept 8.87 + feature contributions = -9.262 (log-odds).
Score = sigmoid(-9.262) = 0.

Score0.214

Feature contributions

FeatureValueContribution
(log-odds)
Comment layout (horizontal)How rapidly the visual pattern of comment characters changes from left to right.0−0.804
Comment layout frequencyHow frequently the layout switches between lines containing only comments or whitespace and lines containing code.8+0.753
Indentation frequencyHow rapidly indentation changes from line to line.7−0.495
Operator visual areaThe share of the code’s visual area occupied by operators such as +, =, and *.0.137−0.387
Identifier layout (vertical)How rapidly the visual pattern of identifier characters changes from top to bottom.6.747−0.316
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.93.875−0.311
Number layout frequencyHow rapidly the number of numeric values per line changes through the code.6+0.283

Intercept 0.07 + feature contributions = -1.207 (log-odds).
Score = sigmoid(-1.207) = 0.23.

Score0.401
Score6
Run 1

The code is difficult to read due to unclear variable names (old1, active1, old2, active2, jjMoveStringLiteralDfa18_0), unexplained magic numbers (16, 17, 0L), and complex bitwise operations combined on one line. It has high action density, making individual logic steps hard to follow. The structure is fragmented and relies heavily on generated parser conventions, reducing readability for general developers. While the nesting is shallow and exception handling is simple, lack of comments or descriptive naming obscures intent. The snippet ends abruptly, suggesting incomplete context, which further hurts clarity.

Run 2

This code is difficult to read due to poor structure and unclear intent. The method name and parameters (old1, active1, old2, active2) are cryptic, making the purpose ambiguous. The bitwise operations with compound assignment in the conditional are hard to parse at a glance, and the method abruptly cuts off mid-catch block, suggesting incomplete or tangled logic. Nesting is minimal but the flow is unnatural—jumping from a conditional return to a try/catch with no connection makes the reading disjointed. Action granularity is mixed: the if statement packs multiple operations into one line, while the rest is sparse. Overall, the code lacks meaningful naming, logical organization, and clear separation of concerns, making it highly unreadable. It would require significant refactoring with descriptive names and restructured logic to be maintainable.

Run 3

The code is difficult to read due to dense bitwise operations and cryptic variable names like 'jj', 'old1', 'active1', etc. The intent is unclear without external context, and the logic on the first line combines multiple operations, reducing clarity. The structure is somewhat logical but tightly packed, and nesting is minimal. However, the reading flow is poor because the code jumps from a condition to a return to a try-catch with little explanation. The action granularity is adequate but suffers from ambiguous naming and lack of comments. Overall, it's understandable only to someone familiar with generated parser code, making it unreadable for general purposes.

Run 1

Code Structure: The fragment follows a recognizable lexer-state pattern, but incomplete blocks prevent understanding the full organization. Nesting: Relatively shallow, with an early return and a try/catch. Clarity of intent: Opaque generated-style names, numbered states, and unexplained bit masks make the purpose difficult to infer. Code length: Short and compact, though brevity comes at the expense of explanation. Action granularity: The condition combines two assignments, a bitwise operation, and a comparison, requiring careful mental parsing; the inline try block also compresses operations. Reading flow: Execution proceeds naturally from filtering active states to reading a character and handling failure, but inconsistent formatting and the abrupt ending disrupt readability.

Run 2

Code Structure: The snippet has a recognizable guard followed by input reading and exception handling, but it is incomplete and lacks closing braces. Nesting: Shallow nesting makes the visible control flow relatively easy to follow. Clarity of intent: Opaque generated-style method names, numbered variables, bit masks, and unexplained constants obscure the purpose. Code length: The snippet is short, although brevity does not compensate for its dense expressions. Action granularity: The condition combines two assignments, bitwise operations, and a comparison; the try block also compresses several structural elements onto one line. Reading flow: The early return provides a natural progression, but inconsistent formatting and the abrupt ending disrupt readability.

Run 3

Code Structure: The visible code follows a lexer-style sequence, but missing closing braces and a truncated catch block leave its structure incomplete. Nesting: Shallow nesting keeps the snippet relatively easy to scan. Clarity of intent: Generated-looking method names, numbered variables, and unexplained constants obscure the purpose; bitmask updates embedded in the condition require careful interpretation. Code length: The snippet is concise, though dense expressions reduce accessibility. Action granularity: The conditional combines two assignments, a bitwise operation, and a comparison, while the try block places reading and assignment on one line. Reading flow: The progression from state filtering to character reading and exception handling is logical, but inconsistent formatting and the abrupt ending disrupt comprehension.