Hacker News Reader

Comment by bandyaboot | Hacker News Reader

I feel compelled to point out that the outrage regarding “human experimentation” was nearly universal here when it was researchers making intentionally bad commits to the Linux code base.

View on HN

I assume (and I don't want to assume too much) you are saying the reaction here is hypocritical?

If so, I don't agree.

The Linux kernel is not too different from an emergency service. Interfering with its operation/development is dangerous to a large number of people on the planet, maybe almost all of them.

A shitty journal article that got accepted to peer review in an academic journal is not really affecting most people's day-to-day life, just academics and philosophers, and we (the collective we, maybe HN) don't really care as much about that as we do the Linux kernel, or at least care proportionately to the importance of the target of that human experimentation.

That would be fine to not care as much and this human experimentation would likely pass the bar for being ethical. The point is that it’s a line in the sand to not cross ever.

It’s like mandating code review. You don’t let the programmer who made the code decide it doesn’t need code review this one time, and you don’t let researchers decide they can experiment on unconsenting humans this one time.

That's a fair point. I may not fully understand who he experimented on, but it seems like the target of experimentation for the professor here was the reviewers and the journals, but he didn't experiment on people to write the paper, just that submitting the paper was the experiment. I feel that is maybe one layer removed from sending bad commits to real people then publishing the results, but I guess after thinking it all out they are close enough to pass a smell test and be critiqued along side each other! Thanks for your example.

I think it is always wise to get ethics (/institutional) review board clearance if you have people participating in your experiments and particularly they do not consent in advance. I understand from our own experience that one avoids those boards sometimes, but IHMO one needs to accept criticism if one hasn't done. I hope he would have got clearance anyways. He could have complained I guess if not.

Nevertheless the penis paper proved a point and we use it in our teaching as an example how peer review sometimes "works".

That's a fair point. I guess any time you try to unofficially exploit a system's weaknesses to prove a point, and you fail, you should expect to be treated like any other person who was busted trying to exploit the system. Otherwise, the system still failed.

If that researcher actually got a change into millions of user's linux systems, he would look like a genius, new checks would be put in place to catch his techniques, and everyone would be auditing every past commit on various open projects looking for other similar attacks. But instead, the system worked as intended, and now you can't go back and say "lol just kidding, it's a social experiment bro!" It's a risk you'd better be willing to take.

Not sure if I understand you completely, but I can't help but say bad commits don't necessarily get merged into the trunk / master / main.

Research being published in any old journal is not the same as merging a patch into main.

As I pointed out in another thread, the journals he published his work to were low tier, open access journals. Anyone can start a journal, and these journals will accept almost any research that's on topic.

So if I were to stretch the metaphor, it's more like he found some fork of the official Linux repository, he submitted a patch that the maintainer just merged in without thinking, and then he used this to show how easy it is to get malicious code into the official Linux repo.

Reply on news.ycombinator.com