That fixes many of the abuse problems with downvotes.
HN situations are also monitored by the site operators themselves, who intervene. YT is too vastly big for that.
That fixes many of the abuse problems with downvotes.
HN situations are also monitored by the site operators themselves, who intervene. YT is too vastly big for that.
No it's not, it's our society that's fucked. It's our CEOs' legal requirement of fiduciary responsibility that causes these problems.
YouTube was 10.9% of Google's revenue for the year 2020, with something around $19,770,000,000 of income. How many employees could Google hire to manually investigate and take action against users? Even 10% of that income to address this problem would hire literally tens of thousands of people to tackle the issue.
The "problem" with Big Tech is that they're convinced of their own intelligence and they all believe they can automate their way to Utopia. Well you can't. You're not that smart. You're not that wise. You need bodies on the ground, you need hands doing work. We may get there some day, but it won't be today, tomorrow, ten years from now, and it may not even be a hundred years from now.
A little painful hubris in the form of some of these enormous social media platforms collapsing not under their own weight, but under their own arrogance in the form of thinking they can automate their problems away, would be a good dose of medicine sorely needed.
The entire revenue of $20B could be blown through on reviewing 6B comments.
If you want to keep this activity to 10% of revenue, that caps you to 600M comments.
> No it's not
And I specifically said the site operators; people hired by Google to moderate are just hired help. They are not equivalent of a dang.
Also, I think your overestimating the number of YT comments. Let’s assume 1 billion hours of watch time per day, and say 20 billion videos, but only 1 in 10,000 views get a comment. That’s just about your 600M threshold per year picked from super inflated costs.
GIGO, your numbers are incredibly rediculous.
What do you think is the average number of quality, thoughtful reviews you can obtain per hour, from an individual, week over week, month over month?
Don't forget that people eat lunch, pee, call in sick and have productivity slumps.
And how little do you think they should be paid?
Lets just assume perfect efficiency and then handicap it to account for lunch, peeing, pooping, killer bee attacks, etc. How about peak efficiency * 3/8, which basically says you only expect your workers to 'work' 40 hours per week, but in an 8 hour day only actually get 3 hours of work done?
Lets say a long youtube comment takes about a minute to read at most, and since we aren't going to quiz people for comprehension later we just need them to see if there is anything objectionable in the comment, we can take 1 minute to be a hard upper bound for the time it takes to handle one comment. The shortest comments take around 5 seconds. Let's assume the average is about 15 seconds (most comments are less than 1/4 the length of the longest comments).
So given the above they can do (3 hours/day * 60 minutes/hour) * 4 comments/minute == 720 comments/day. If we are forced to hire in the US we can easily find people to do this work for $25/hour working from home.
So every human review costs ((8*25)/720) == 27 cents. This is like 1/10th of your estimate and I feel like I'm going out of my way to make the numbers unrealistic in your favor.
Realistic numbers would be that it takes 15 seconds per comment on average and you hire people part time so they don't get burned out on reading comments (or full time but only part of their day is spent reviewing comments). With the same numbers as above, but not paying them to poop and pee for 5 hours out of an 8 hour day gets the cost per review to like 8 cents.
Could they do it better than an algorithm which already does it for free?
You do know most moderation is contracted for much less than $60/hour. Potentially to offshore corporations paying less that US minimum wage.
It's doable, but an uncomfortable reality.
This is nonsense pretty much every time it’s trotted out. There is no objective “requirement of fiduciary responsibility” that governs every action a CEO might take in a given situation, especially not for dumb minutiae like “allowed unfettered downvoting.” If it was this simplistic, we wouldn’t need CEOs at all.
Tens of thousands of people moderating comments sounds like the worst of both worlds. Not nearly enough to keep up with volume, and way too many to ensure quality moderation. You'll get a horde of randos deleting stuff in idiosyncratic ways while bad stuff keeps getting through anyway.
Use your preferred search service to search for fiduciary duty myth.
Agreed. Another way of framing this is that HN is sponsored by an organization that makes a lot more money per HN user than Google makes per YouTube user. In other words, HN users don’t have to pay for their own moderation by generating revenue in the same way YouTubes users do.