Not sure what a thinly resourced piece like this is doing so high up on HN. There’s really very little evidence that this has anything to do with racial slurs.

From a quick read: last June, the channel was banned for “harmful and dangerous” content. The ban was removed in under 24 hours. YouTube has not confirmed the reason. And yet:

> Experts suspect that it was the usage of words like "black" and "white" that confused the Yutube's AI filters.

Well, if experts said it, it just be true! The experiment they ran is speculative as best and the article just ran with it as absolute fact. If the filter was caused by use of “black” and “white” wouldn’t every chess channel fall victim to it? The filter isn’t new, why was it triggered once last June?

A lot of questions raised and an article that has no interest in answering them. It’s credited to “Buzz Staff”… I think it’s always suspect when no one wants to attach their name to something they wrote. A smell of clickbait running all the way through this one.

I recently took some actions to reduce the damage if Google ever ban my account.

I'm now using a custom email domain (although it's just redirecting to my Gmail for now), and doing regular backups of my emails (through Thurderbird) and Google Drive.

With these news being more and more common, you never know when it'll happen to you.

What is amazing is how there is no way to recover from all these algorithmic banning. I'm getting banned on Twitter every other day from following anyone because I follow a lot of academics. It's all legitimate but algos detected some pattern and I'm now toast. Twitter provides exactly zero ways to report this or fix this. They simply don't care if some monority users suffer. We see same things when Google disables account and you get locked out from your electronic life.

I used to think working with governments is very hard. Getting support in places like DMVs and passport renewals is super hard. But never thought so futuristic tech world will be so much more arcane and worse than even the governments.

inb4 someone will suggest that FIDE should change the color names to be more inclusive + why does white have the first move advantage, why can't black be the first. Perfect breeding soil for just another day of pointless race drama on Twitter

It gets more ridiculous with every day.

I fully expect people to start developing vocabularies to harden their group communication against these automatons.

Also, please, don't use term "AI". Calling it intelligence is too charitable for something that can't distinguish game of chess from hate speech.

The mistake a lot of companies are making is they view ML as a replacement for humans. It's not. At least, it's nowhere near that now.

You should view ML as an enhancement for humans.

So the problem here is that no ML system without essentially being an AGI is capable of fully comprehending context, in this case use of "black" and "white" in Chess that has absolutely no racial context.

So the mistake here is (likely) automated action against the channel. That should never happen. At least, not for this particular signal.

What should happen is the channel should be flagged for review by a human. Any human would appreciate the context and take no action.

This is the "enhancement" part. With no automated assistance, a human might be capable of moderating (made up numbers) 100 channels. That doesn't scale at Youtube scale. ML systems might enhance that human's moderating ability to 10,000 channels.

Even better, by flagging for review, you're actually producing training data for your ML. The P-R for your automated disciplinary system needs to be incredibly good before you take humans out of the equation. False positive bans or blocks on channels is bad for the individual channel. It can be devastating in fact. The corporate view is that "one channel doesn't matter". And they're right... to a point. At some point it undermines confidence in the platform and then it does matter.

I guess the decision to take down the channel wasn't as black and white as they originally thought.

This was only speculation. It's a pretty weak story:

> Even though the channel was restored within 24 hours, the YouTube did not explain why it had blocked Croatian chess player Antonio Radic, also known as 'Agadmator,' from its platform briefly, the Dailymail reported.

> Experts suspect that it was the usage of words like "black" and "white" that confused the Yutube's AI filters. They found that 80% of chess videos that were flagged for hate speech actually ahd terms like 'black,' 'white,' 'attack' and 'threat.'

YouTube didn't say why, and they simply "suspect."

It's just as likely that is was flagged for other reasons. For example, mass "reporting" of the channel as offensive by people with an axe to grid against this particular chess YouTuber. (It's been my personal experience there's a lot of politics in the chess community.)

Incidentally, if you like chess the channel is outstanding. (And back online now.)

https://www.youtube.com/user/AGADMATOR

I enjoy watching on my lunch breaks. Great breakdowns of games between top players past and present.

I'm not sure how to word this... I'm confused that we are still calling people "black" and "white", when their actual color is nowhere near those two. It's no longer cool to call people "yellow" or "redskin", so why are those still accepted?

People I know actually use "dark-brown", "light-brown", and "light-skinned" and other such more accurately descriptive terms when actually describing someone's appearance, and I like that much better.

> They found that 80% of chess videos that were flagged for hate speech actually ahd terms like 'black,' 'white,' 'attack' and 'threat.'

Surely effectively all chess videos, banned or not, have those words in them? This all feels pretty speculative.

“ The researchers now suggest that social-media platforms should incorporate chess language in their algorithms to avoid future incidents like this.”

Or maybe don’t automate banning YouTube channels via nebulous black boxes? Or can advertisers stop being afraid of anything and everything?

Their algorithms seem to be looking for the words "black" and "white". I posted a comment to a YT video just today describing my daughter's experiences in high school, the school's recent heavy emphasis on black lives matter (the topic has crept into History, Cooking, Math, and even Chemistry) and the kids, including the few black kids at the school, think it's excessive. The comment disappeared when I refreshed. I reposted (have learned to always cmd-C), and it was taken down again. So I substituted "bite" for white, "back" for black, and this time it stayed up. Ridiculous and counter-productive, in my opinion.

I didn’t know Agadmator was censored so I looked for information and found a video of him talking about it here: https://m.youtube.com/watch?v=KSjrYWPxsG8

He doesn’t know why his podcast was flagged. He mentions a few possibilities-one of which is the black and white thing, which would be a mistake. It’s notable how guarded his speech is, as though he might say something else verboten.

I’ve watched hundreds of chess videos on YouTube and never heard of one getting flagged until now.

Amadgator is a one of the top chess youtubers, after carlson and hikaru. They should have a human-check placed by channel size - if for nothing else just to not get bad PR like this.

Isn't the biggest structural problem with ML that it can't explain how it came to a specific conclusion?

Anyone knowledgeable on this subject?

Delegating prosecution of possible hate speech to commercial entities was always going to be bad.

Youtube is essentially a public forum with audience of entire countries, where the people who run it choose what gets amplified, what gets shunned and what get outright blocked.

Who would be happy with a private entity deciding the future mindsets at this scale?

What’s funny is how this shows how bad poor algorithms can be at scale. Even with human intervention.

These cases have lots of exposure and people who notice. So the real QA is in these edge cases where customers find bugs. I worry about how not so visible algorithms.

I had a company market to me about sentiment analysis. I asked their accuracy and they answered that they can’t actually test but their training model has x%. So to make decisions accordingly. But then I had coworkers doing stuff like what percent had what sentiment vs the other with no assumptions.

I think the problem is people are kind of bad with probabilities so something either is or isn’t. And I seem to witness a gulf where there are people who kind of understand and then people who don’t but trust the output.

Again with the lies. Youtube did _not_ block agad's videos for racial slurs... "Experts suspect" that may have been the reason youtube blocked the video(s). Youtube didn't say why.

Lots of talk about false positives. As a human, when dealing with fuzzy logic, I like to focus on recognizing when something is wrong to reduce false positives.

Maybe Google should put more effort in training AIs to recognize when they don't know instead of shoehorning an answer.

An analogy is I would expect a non-stupid human who has never seen a screw before, but trained to work with hammering nails to quickly recognize something is wrong the first time they're presented with a screw and not attempt to hammer it flush.

YouTube should not be blocking any channels or taking down videos in all but the most egregious circumstances. These companies are absolutely out of control and have been for a while.

Reminds me of https://en.wikipedia.org/wiki/Scunthorpe_problem

Their either needs to be competition for Youtube or they need to face regulation. You can't have it both ways. The system does not work without one or the other.

I wonder if this could be manipulated the way redditors used to manipulate Google image by posting 'this is a house' with an image of a potato and after enough upvotes, the potato would appear first if you searched house on Google image.

What if people start publishing hate speech but replace the hateful words by, say, 'Alphabet' ?

Will the AI learn that Alphabet is a hateful word ?

Surely someone with enough resource should be able to do that.

This must be nipped in the bud before it just turns into a billionaire’s tool for shutting down conversations on class consciousness.

It should be a criminal act to ban someone from your platform without pointing exactly to why they were banned. "Oh but we can't expose details of our ban process for people to game" Bullshit. They just don't want to have to answer to anyone for cutting them off.

the chess is a very aggressive game of premediated threats and attacks with the goal to achieve dominance and control over another player. It fosters individualism and combativeness instead of teamwork and collaboration. I think it should be age restricted like say alcohol and driving.

Agadmator’s channel is truly a gem. He provides excellent analysis of interesting games and includes some great dives into chess history as well. I hope google gives an adequate explanation for this and makes sure it doesn’t happen again (not holding my breath)

Except no racial slurs were used. Just “black, white, attack, threat”.

People love making clickbait titles.

> They found 82% of the comments flagged in a sample set didn't include any obvious racist language or hate speech,

So 18% did include obvious racist language or hate speech?

Is it just me, or did this article have an incredible amount of typos and grammar mistakes given its length?

Various missingspaces, Carnegie Melon[sic], etc.

"In this position, black is better."

I know I've got an audio plugin (AU/VST) called Bit Shift Gain, because it applies the exact gain changes required to just bit-shift the exponent of the floating point value one bit without touching the mantissa. It's a handy plugin for clean gain changes (completely reversible without loss so long as you're gain boosting, not clipping, and then cutting again).

I can't say its name on youtube on my own channel, 'cos YouTube thinks I'm saying something else. (not sure what would happen if I kept on doing it. Maybe they would start sending me apprentice Nazi traffic :D )

are there any mirrors to youtube? i tried to search this but nothing promising showed up. i think this might be due to the sheer size of the video site. but is there a site that saves the newest videos for the last 3 days?

How come that in 21st century we are still using the racially degrading terms, like black and white, in so many places?

Does it take a lot of effort to replace them with neutral non-offensive colors?

Speaking about chess, don't you think that chess being mostly-male mostly-white is a direct consequence of the language they use?

Why is Google so averse to "human in the loop" solutions?

Response from the YouTube support team:

> Sorry about that

I wait for the day that chess will be banned in USA, or some fork of chess will emerge, because obviously chess is racist: white is always first.

It is fascinating to witness tech giants and the academy move towards a singular focus on melanin and gonads in the name of wokeness.

Click-bait title.

fitting for a site named news18.com

Clbuttic.

What triggered it white pawn takes black queen

So, assuming YouTube makes some manual adjustments to their algorithm to not flag/ban chess content, I wonder if actual hate speech will start pretending to be chess content in order to avoid detection.

IIRC the technology industry has moved away from master and slave terminology, is this still true, and could that get someone blocked?

We're never going to achieve justice Justice until we ban those ultimate bastions of hate, the English language and Roman alphabet.

Anyone who has ever used these instruments of injustice is tainted.

Oh, wait.

Isn't this old? We were just talking about this very problem seven months ago. It's hard to believe Youtube didn't put in some hack or retrained for chess to avoid this high profile embarrassment in seven months.

https://news.ycombinator.com/item?id=23729156