Toxic Code Snippets on Stack Overflow
GitHub's Copilot tool has brought the twin issues of software licensing and the exploitation of open source contributors to the fore once again. (See for example this discussion and this one.) For those who missed it, GitHub used millions of lines of code from the repositories it hosts to train a nuclear-strength autocompletion engine. It's a cool idea, but it would have been ever cooler if GitHub had asked whether the licenses on those repositories allowed it to do what it did, or offered some kind of compensation to the programmers whose labor they took advantage of.
Intellectual property isn't the only potential problem with Copilot; as we've learned the hard way, models are only as good as the data they're trained on. Ragkhitwetsagul2021 casts light on both of these concerns by looking at the quality of code snippets posted to Stack Overflow. Among its findings:
- 69% of answerers never check for licensing conflicts between their copied code snippets and SO's CC-BY-SA license. (In fact, 85% of answerers are't even aware that SO enforces that license.)
- 66% of people don't check for licensing conflicts when using code snippets.
- Of the 2,289 non-trivial clone candidates they studied, 214 (9.3%) could potentially violate the license of the original software. These clones appear 7,112 times in 2,427 GitHub Projects.
- 153 of the clones had been copied from the Qualitas corpus (a collection of Java code used for research). Of those, 66% were outdated and 6.5% were buggy.
Stack Overflow isn't GitHub, but I'd struggle to believe that code taken from the latter would be of higher quality than the answers on the former. I would also be surprised if people who ignore licensing issues on one site are scrupulous about them on another. AI-assisted software tools are clearly coming our way; no matter what else they do, they will provide lots of material for discussion in software ethics classes.
Ragkhitwetsagul2021 Chaiyong Ragkhitwetsagul, Jens Krinke, Matheus Paixao, Giuseppe Bianco, and Rocco Oliveto: "Toxic Code Snippets on Stack Overflow". IEEE Transactions on Software Engineering, 47(3), 2021, 10.1109/tse.2019.2900307.
Online code clones are code fragments that are copied from software projects or online sources to Stack Overflow as examples. Due to an absence of a checking mechanism after the code has been copied to Stack Overflow, they can become toxic code snippets, e.g., they suffer from being outdated or violating the original software license. We present a study of online code clones on Stack Overflow and their toxicity by incorporating two developer surveys and a large-scale code clone detection. A survey of 201 high-reputation Stack Overflow answerers (33 percent response rate) showed that 131 participants (65 percent) have ever been notified of outdated code and 26 of them (20 percent) rarely or never fix the code. 138 answerers (69 percent) never check for licensing conflicts between their copied code snippets and Stack Overflow's CC BY-SA 3.0. A survey of 87 Stack Overflow visitors shows that they experienced several issues from Stack Overflow answers: mismatched solutions, outdated solutions, incorrect solutions, and buggy code. 85 percent of them are not aware of CC BY-SA 3.0 license enforced by Stack Overflow, and 66 percent never check for license conflicts when reusing code snippets. Our clone detection found online clone pairs between 72,365 Java code snippets on Stack Overflow and 111 open source projects in the curated Qualitas corpus. We analysed 2,289 non-trivial online clone candidates. Our investigation revealed strong evidence that 153 clones have been copied from a Qualitas project to Stack Overflow. We found 100 of them (66 percent) to be outdated, of which 10 were buggy and harmful for reuse. Furthermore, we found 214 code snippets that could potentially violate the license of their original software and appear 7,112 times in 2,427 GitHub projects.