Arias: Human proof for FOSS contributions [LWN.net]

LWN<br>.net<br>News from the source

Content Weekly Edition<br>Archives<br>Search<br>Kernel<br>Security<br>Events calendar<br>Unread comments

LWN FAQ<br>Write for us

User:<br>Password: |

Log in /<br>Subscribe /<br>Register

Arias: Human proof for FOSS contributions

[Posted May 26, 2026 by jzb]

Rodrigo Arias Mallo, maintainer of the Dillo web browser, has written a<br>blog post<br>with a proposal on one way to ensure that a contribution is written by<br>a human and not AI; he suggests asking new contributors to record<br>their programming session using asciinema.

In the same way that LLMs generate patches, they can also generate<br>the asciinema recordings themselves. Then, the contributors can lie to<br>the reviewers pretending to have made the edits. Perhaps surprisingly,<br>this is not a easy task for LLMs, at least from my observations. The<br>corpus of recordings of developers making mistakes and thinking the<br>whole process of editing a file is not as large as the corpus of FOSS<br>programs and patches in which to train an LLM. During my very simple<br>tests I haven't been able to generate an asciinema session that<br>remotely resembles what I would expect from a human, and even less so<br>from a human with a nice editor theme and editing an existing Dillo<br>source file.

The Dillo project is not yet requiring asciinema recordings, but he<br>said that he would like to test the theory further. LWN covered asciinema in<br>January 2026.

to post comments

This seems unwieldy

Posted May 26, 2026 18:03 UTC (Tue)<br>by dskoll (subscriber, #1630)<br>[Link] (1 responses)

Reviewing an editing session sounds like a very painful expenditure of time. And also, I suspect many would-be contributors would simply refuse to do it. Imagine doing this for every single patch.

I think continuing to have contributors attest that they have not used AI is the best we can do. After all, we already just trust that contributors haven't ripped off proprietary code they happened to have seen, without asking for a recording of their programming session.

This seems unwieldy

Posted May 26, 2026 18:20 UTC (Tue)<br>by tlamp (subscriber, #108540)<br>[Link]

Also, it'd bet it be trivial to fake such session too.

IOW, this just makes it again harder for humans but not really for bots, given that the latter has no problem with wasting time and nerves on such things.

This isn't just anti-ai, it's also anti-gui

Posted May 26, 2026 18:43 UTC (Tue)<br>by jbills (subscriber, #161176)<br>[Link] (2 responses)

I suppose this guy also only wants contributions written in emacs/vim/other terminal text editor. This idea is very silly. If you wanted to be realistic about it, require a video recording, not asciinema. Maybe one day all programmers are forced to become vtubers to contribute to open source. That seems like a good idea.

Speedruns do require this

Posted May 26, 2026 19:19 UTC (Tue)<br>by tialaramex (subscriber, #21167)<br>[Link]

Yeah, that's potentially interesting. Speed running is done this way in some categories. Nobody cares that you have some inputs which, if they were made in real time, would constitute the fastest SMB Any% because just that is called TAS (Tool-assisted speed run, typically assembled over an extended period perhaps by a large group co-operating) and already exists separately from "manual" speed running, if you want to claim an actual speed run they need video showing you and your hands making inputs, as well as the video game itself.

Since you're on video constantly while making attempts you might as well be on Twitch where you might even recoup a little bit of money from revenue share. Most of these communities are tiny, maybe you have fifty people peak watching your attempt at Blue Prince no-major-glitches Bequest speed running but you need to record video anyway or your attempts are just hearsay so eh, might as well. And hey, maybe you attract a following, the "hot girl bonus" factor applies on Twitch like most of life, but you might be a more entertaining watch than other players for all kinds of other reasons.

This isn't just anti-ai, it's also anti-gui

Posted May 26, 2026 22:34 UTC (Tue)<br>by nix (subscriber, #2304)<br>[Link]

Not just that -- it would have to be *in a terminal*. I have used X for my Emacs for as long as it's been possible to do so, because I like my fruit salad and my different fonts and my right-click context menus and my keybindings that terminals can't pass on and all that stuff Emacs users are supposed to revile and abhor. Am I supposed to switch to writing code in the straitjacket of a terminal emulator just in order to prove that I'm not slopping? You trust me that little, I don't contribute, sorry.

Nonsense

Posted May 26, 2026 19:05 UTC (Tue)<br>by gmprice (subscriber, #167884)<br>[Link] (4 responses)

I understand the general anti-AI sentiment for things like art, but code is code.

There's little difference between well designed code generated by an LLM and well designed code written by hand - except that one took tokens and...