On a bit of a whim I've spent some time on and off during the past couple of weeks looking through the New Answers to Old Questions feed, and I have noticed a recurring pattern of plagiarism. A few times each day, an answer is posted that consists entirely of multiple lines of code copied from some other answer to the same question without attribution. Can some sort of automated filter be implemented that blocks such trivially plagiarized answers?
Adding some details, the pattern I have noticed is:
A new answer to an old question.
By a 1-rep account.
Consisting entirely of multiple lines of code with no explanatory text.
That is copied exactly (including matching case) from some code block in another answer to the same question with no attribution.
Frequently the formatting is copied also, making the copying quite obvious: Examples:
- Toggle Bootstrap button text on button click?,
- Stop pasting html style in a contenteditable div only paste the plain text,
- Display a loading bar before the entire page is loaded,
- fullcalendar timegridweek don't show events time properly,
- Issues implementing SHA256 hash, then pbkdf2Sync and finally cipher with AES-256-cbc - c# to Nodejs,
- Overlay image on div
(subsequently edited to change<!-- begin snippet: js hide: false console: true babel: null -->to<!-- begin snippet: js hide: false console: true babe2: null -->) - PDF file to be displayed on the dialog modal via bootstrap,
But sometimes the formatting is not copied so the posts look quite different. Examples:
Occasionally code is copied from multiple answers: Examples:
Or the code is truncated. Examples:
One or two plagiarized answers per day doesn't sound like a lot, but that's just what I noticed by checking the NATO feed for a few hours on several days in the past couple weeks. Assuming my sample was random we're talking hundreds per year. It's enough that I can manually raise "Plagiarism" flags faster than mods seem to be handling them. And while reviewers in the "Late Answers" queue are, according to this answer to Is a plagiarism audit justified?, supposed to check for internal plagiarism:
the Late Answers queue is a bit different from others insofar as checking for internal plagiarism (i.e. copies of other answers to the same question) should (IMHO) be an inherent part of the review process.
it doesn't seem to be working.
Could we implement some automated way to block such answers or screen them out? And if so, what should it be? I'm not sure that "better onboarding" would be the solution here, such trivially obviously copied answers border on being abusive. If we warn the poster about their plagiarism while they are creating their post, then (assuming they are not bots) they will probably just tweak the post a little, so maybe a plagiarism flag should be automatically raised instead?
Update January 2025.
Note that this is still going on, see e.g.:
- https://stackoverflow.com/a/79359035
- https://stackoverflow.com/a/79359199
- https://stackoverflow.com/a/79356743
- https://stackoverflow.com/a/79348878
- https://stackoverflow.com/a/79341289
- https://stackoverflow.com/a/79331190
My observation is that I see at least one plagiarized answer per day in the NATO queue. The discussion under the question How to handle code-only answers that are entirely copied, verbatim, from the question itself? suggests that most of these plagiarized answers are created by someone who picks "Copy snippet to answer" from some other answer, then somehow submits their answer, so maybe we could try to prevent that from happening somehow?
Update May 2025.
Still happening, e.g.:
- https://stackoverflow.com/a/79627134
- https://stackoverflow.com/a/79607264
- https://stackoverflow.com/a/79605558
- https://stackoverflow.com/a/79601463 (with added copy/paste nonsense line)
- https://stackoverflow.com/a/79581429
- https://stackoverflow.com/a/79564672
Seems to be less frequent though. Instead I'm seeing more nonsense answers.
Update June 2025.
Two separate plagiarized answers for issue when trying to split url in js
Two plagiarized answers for How to copy URL on button click? as well:
Update November 2025.
Still happening, e.g.
Answer to How to search and play a video on YouTube using Selenium in Python?. (Screen shot).
Answer to How do I make my text in HTML look like the code in here.
Answer to Android, Kotlin, Google Maps Markers.
Answer to Aligning elements left, center and right in flexbox.
One new twist is that sometimes the new "Copy" button is used, which auto-inserts an attribution.
