Former YouTube content moderator describes horrors of the job in lawsuit | Hacker News Reader

Former YouTube content moderator describes horrors of the job in lawsuit

cnbc.com

52 pointslaurex24 comments

There is a positive psychological correlation between the Orderliness (Conscientiousness) personality trait and Disgust Sensitivity. Agreeableness (Compassion/Politeness) and Neuroticism (generally anxiety) are also positively correlated to Disgust Sensitivity. [0]

Lets assume that certain personality traits, such as Orderliness, Compassion, and Politeness, are most conducive with the role of moderator and they are the most susceptible to feeling (perhaps very intensely) disgust when confronted with such content (especially repeatedly over a long period of time).

It may be a catch-22. We theoretically want orderly, compassionate, and polite people moderating the public discourse. If we filter out those personalities, leaving only the most Open people who are least likely to have disgust sensitivity, then we have the people who are the least capable of making important distinctions about what is acceptable to the rest of the us.

That said, perhaps that's exactly what we need, instead of having moderators making sophisticated distinctions, we just need moderators who are capable of and willing to identify what is (universally) criminal/evil (exposing them to the worst of humanity). Leave the remaining distinctions to the community.

[0] https://www.sciencedirect.com/science/article/abs/pii/S01918...

It's locked up behind a paywall, but people confuse "correlation" with "equivalence." For this to be a catch-22, you'd need an amazingly strong correlation.

Also, most psychology results fail to replicate. A single paper has about a 60-70% chance of being nonsense, probably more.

Especially because they used PCA of all things to figure out what these personality traits actually are.

I'm aware of the replication crisis and that correlation does not equal causation. Are you aware of any research that contradicts the original statements?

I think your statement confuses who has the burden of proof when proposing a new theory (or just how sorry the state of science is).

I'm a commenter on HN who made a comment for the sake of discussion, not a researcher proposing a paper. You avoided the comment entirely, cited the replication crisis and the old hat "correlation != causation" argument when my comment didn't even have that problem.

> We theoretically want orderly, compassionate, and polite people moderating the public discourse.

I understand why you would think that WRT moderating discussions or comments, especially where that entails interaction with the participants (eg. what dang does here on HN), but those considerations don't really apply to making a determination of whether a video should be taken down.

Yes, I said that in the next paragraph. The debate is about whether or not a certain personality types are..

a) more appropriate as moderators.

b) able to (emotionally & psychologically) withstand the type of content they are tasked with moderating.

It's therefore my contention, based on the research I could find, that an Open personality is sufficient to fulfill both criteria assuming that we don't want moderators making distinctions beyond a certain point (e.g. is this content criminal/evil?)

I guess I didn't state my point clearly.

"Moderation" isn't just one thing. Even in theory the personality types you would want for different forms of moderation would be different.

Here are some more citations: "orderliness is associated with disgust sensitivity"

https://www.youtube.com/watch?v=dMl8HwDpae0

"We found that trait core disgust was associated with higher orderliness"

https://research.vu.nl/en/publications/cognitive-and-persona...

"We propose that trait disgust is associated with the specific motivation to create and maintain order"

https://www.cell.com/current-biology/fulltext/S0960-9822(14)...

"The results found that Orderliness mediated the association between disgust sensitivity"

https://tspace.library.utoronto.ca/handle/1807/77383

"trait disgust encourages a generalised search for order"

https://www.narcis.nl/publication/RecordID/oai%3Aresearch.vu...

I remember reading an article some time ago about thr experiences of a developer who I believe was working on the most recent Mortal Kombat game. They talked about needing to watch lots of gore and similar content for the games cinematics, and about how it started to affect their mood. I don’t think I’d be able to work a job like that for very long.

Was this the article?

https://news.ycombinator.com/item?id=19888220

Yeah man. You gotta be cutout for this shyt. Just reading the article alone changed my mood. I don’t see how one can process that amount of information and not have psychological issues.

That would be one place where deep learning could sort things and preserve humans, a few errors could be tolerable while sparing ppl lives.

The Verge also had an article about Facebook moderators[1]. Absolutely atrocious. Not just the content they have to deal with, but the working conditions. Barely any spare time or support despite the fact that people have to deal with crazy shit, employees who turn into conspiracy theorists after being exposed to it, people who do drugs at work to deal with it, simply crazy.

That looks like the future of tech. A bifurcated economy where a few developers collect stock options and enjoy fancy food delivery every day and an army of serfs who have to crawl through the garbage.

[1]https://www.theverge.com/2019/2/25/18229714/cognizant-facebo...

Thanks for the link, I read that a while back and have been trying to find the article again ever since then, but hadn't been able to find the right terms to put into google.

Surely you could extract stills for the reviewer in some way and no sound? Feels like a management issue.

It seems apparent to me that many things could be explored to reduce the impact of disturbing content. How about converting the images to black and white? How about slowly fading in the content, so that the moderator can cancel display of the image as soon as they judge it, reducing the intensity of otherwise detailed memories? How about using machine learning to judge whether content is in slow motion and, if so, increase playback speed so details aren't so easy to see? How about--and this one I'm just throwing out there with no confidence--rotating the content 90 or 180 degrees to maybe change how the brain interprets the image?

I was wondering the same thing. Still images, maybe added blur. As the other commenter suggests, black and white.

Maybe combined with some machine learning to interpret what images show, giving a description before showing it to the moderators? There must be a large amount of horrendous pictures online with some form of labelling. I cant imagine being the person who has to sift through all that stuff though

Anyone know of any remote content moderation positions going? Looks like YouTube is in person only.

I've watched most of the content listed in the article and it did nothing to me. In need of some casual work and this seems like easy money.

It's low paid work done in India, Phillipines and Malaysia so I would find remote unlikely.

That's what I had previously assumed it was. However article is about a legal case occurring in California and Youtube appears to be hiring content moderators in Michigan. They apparently need english speaking Westerners for some aspect of it.

https://careers.google.com/jobs/results/118369454816928454-c...

A bit of a tangent, I’m at extreme range on my WiFi and I got 5 FOUC’s loading this page.

Reply on news.ycombinator.com