Sample

Buse/6

Human rating uses the original dataset scale.

DatasetBuse
Human2.19
Lines4

Scores

MethodScoreScaleDataset rank
(0–100)
Posnett0.7250–1; higher is more readable
75.8Δ 70.7
Scalabrino0.7930–1; higher is more readable
57.6Δ 52.5
Dorn0.7520–1; higher is more readable
72.7Δ 67.7
Mi0.6210–1; higher is more readable
60.6Δ 55.6
LLOC5Logical lines; fewer is more readable
29.8Δ 24.7
DeepSeek V4 Pro7.670–20; mean of three runs
5.1Δ 0.0
GPT-6.1 Sol6.670–20; mean of three runs
1.0Δ 4.0
BTTF0.4960–1; higher is more readable
47.5Δ 42.4
Human2.191–5; original rating
5.1

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)
Byte entropyHow varied the bytes in the source text are. Repeated bytes give a lower value; a more even mixture gives a higher value.4.217−6.325
Halstead volumeHow much symbolic information the code contains. It combines the total number of operators and values with the number of distinct ones.96.211−3.175
Source linesThe number of non-empty source lines.4+1.6

Intercept 8.87 + feature contributions = 0.97 (log-odds).
Score = sigmoid(0.97) = 0.725.

Score0.793

Feature contributions

FeatureValueContribution
(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.39.8+1.98
Comment layout frequencyHow frequently the layout switches between lines containing only comments or whitespace and lines containing code.5+0.859
Comment layout (horizontal)How rapidly the visual pattern of comment characters changes from left to right.0−0.804
Indentation frequencyHow rapidly indentation changes from line to line.1−0.703
Identifier layout (vertical)How rapidly the visual pattern of identifier characters changes from top to bottom.3.512−0.361
Number layout frequencyHow rapidly the number of numeric values per line changes through the code.3+0.352
Operator visual areaThe share of the code’s visual area occupied by operators such as +, =, and *.0.126−0.284

Intercept 0.07 + feature contributions = 1.11 (log-odds).
Score = sigmoid(1.11) = 0.752.

Score0.621
Score5
Run 1

The snippet is a short fragment with some structure but suffers from poor formatting and unclear intent. The code is mostly flat with minimal nesting, which is good, but lacks braces around the outer if statement, making the logic ambiguous and error-prone. Variable names like 'xsp' and 'jj_scanpos' are cryptic and require context to understand. The action granularity is acceptable, with one main operation per line, but the return statement is on the same line as the nested if condition, reducing readability. The reading flow is disrupted by the lack of indentation for the nested block and the missing closing brace. Overall, the logic is not immediately clear, and the formatting does not support easy comprehension.

Run 2

The code is short, which helps reduce cognitive load, and each line represents a single logical action. However, readability suffers significantly due to a lack of context: variable names are cryptic (`xsp`, `jj_scanpos`), and the numeric token literals (100, 101) obscure intent. The false branch of the first `if` is empty (indicated only by the closing brace placement), and there is no surrounding structure or comments to explain why the scan position is saved and restored. The flat, unindented cascade of if-statements is not deeply nested, but the absence of blank lines or formatting makes the control flow hard to follow. Overall, the logic appears to be a fallback parser check, but without meaningful names or comments, the purpose remains ambiguous.

Run 3

The code is poorly readable due to several issues. The structure is fragmented and lacks context, making it hard to understand the overall logic. Nesting is relatively flat, but the lack of indentation consistency and missing braces for single-line if statements reduces clarity. Intent is ambiguous because variables like 'xsp' and 'jj_scanpos' are cryptic, and the token numbers (100, 101) lack meaningful names. Action granularity is mixed—some lines contain multiple actions, such as 'jj_scanpos = xsp; if (jj_scan_token(101)) return true;', which hinders readability. The reading flow is disjointed, and the code feels like an excerpt without surrounding logic. Overall, this snippet would benefit from meaningful variable names, consistent indentation, clear braces, and separation of actions for better comprehension.

Run 1

Code Structure: The scan-and-reset sequence has some organization, but the incomplete block weakens it. Nesting: Only one visible nesting level, though inconsistent indentation obscures scope. Clarity of intent: Cryptic identifiers and numeric token values make the purpose difficult to infer. Code length: Very concise, with no obvious redundancy. Action granularity: Assignments are separate, but the final line combines a condition and return. Reading flow: The sequential operations are easy to follow locally, but the missing closing brace and unexplained state reset interrupt comprehension.

Run 2

Code Structure: The snippet appears incomplete, with an unclosed conditional block, and its organization is weakened by inconsistent indentation. Nesting: Nesting is shallow, but the inner conditional is not visually distinguished. Clarity of intent: Saving and restoring the scan position suggests token matching with a fallback, but abbreviated names and unexplained token values obscure the purpose. Code length: It is concise without obvious redundancy. Action granularity: Most lines contain one action, though the inner conditional and return share a line. Reading flow: The sequence is locally understandable, but poor formatting and the missing closing brace interrupt the flow.

Run 3

Code Structure: The apparent fallback token check is logically related, but the snippet is incomplete and lacks a closing brace. Nesting: Only one level of nesting is present, but missing indentation obscures it. Clarity of intent: Saving and restoring the scan position suggests parser lookahead, though opaque identifiers and numeric token codes make the purpose difficult to understand. Code length: Very concise, without obvious redundancy. Action granularity: Most lines contain one action; the final line combines a condition and return, which is compact but less explicit. Reading flow: The sequence is understandable with parser knowledge, but inconsistent indentation and the unfinished block interrupt reading.