Censorship by Algorithm Does Far More Damage Than Conventional Censorship | Hacker News Reader

Censorship by Algorithm Does Far More Damage Than Conventional Censorship

caitlinjohnstone.substack.com

55 pointsk1m28 comments

There's a reason why humans are and will remain judges&jury.The point of a punishment is redemption, not malice or revenge.If it's not redemption then the judge will be affected by his own hand, sooner or later.He will thus become more just in his actions.No this cannot happen behind a screen

Then again this discussion is pointless, since big corporations enjoy both the privileges of being "private"(ironically while also treated as public forums by the legal systems) while not being hold responsible for their own content.Something censored by 'the alg'?Well, Section 230 vaguely says anything can be censored as long as the motivation behind it is also vaguely explained,and of course they cannot be held under scrutiny because technically "it's a private company[...]".

What people should remember is that every action has a price.You not only vote for your favorite candidate when elections come but you also vote with your wallet, the products you buy, services you use, every decision you make in your life.Complain about google when registered with a gmail account? Not an argument.We cannot allow ourselves to continue in an uneven playing field due to morality.The bigger player doesn't have morals.

>There's a reason why humans are and will remain judges&jury.

https://www.theatlantic.com/ideas/archive/2019/06/should-we-...

Unfortunately, I don't think we're far from AI deciding what the terms of plea deals are, and what sentences to hand down in a trial.

Skimming through the article something jumped at me and that's where indeed the actual danger is: ai-based "recommendations" for someone who is or not high-risk.In this case from what i see the decision was still made by a human, but the fact is that most of the judging came from an algorithm.I don't think humans will be replaced as ultimate deciders(let's hope for this, though i'm pretty optimistic), but we need to be careful about who, or better said what we introduce in the pipeline.Because a decider will use the algorithm as a scapegoat for making a decision, just like we already use any other person/object/variable as a scapegoat for our decisions.

Algorithms and such systems are a double-edged sword.They're good for analyzing potential threats for a defensive approach(thus why they maybe incriminate too much)but they're not, and >will not< be able to do the proper judgement for a person.Because they inherently cannot comprehend, simulate, or understand the human potential, spirit, that ultimately someone can fundamentally change(whether they choose to do so is another question).This is ultimately a very philosophical concept of what and what cannot science explain, and for the most part it cannot explain how a person who is at it's deepest point can become a very good person given the promise (mostly self-made) of redemption.

"Unfortunately, I don't think we're far from AI deciding what the terms of plea deals are, and what sentences to hand down in a trial."

AI is just a puppet of those who program them. And a convenient scapegoat to hide behind. Never forget that. There is nothing magically more objective about a AI.

The problem I see here is that there is no alternative to an algorithm. What do you expect Twitter to do? Twitter's users want to see the content that is most interesting to them, and to do that Twitter has to design algorithms to achieve that. Maybe that algorithm is literally just "Here's everything anyone you follow has posted chronologically" - but that's got problems, firstly obviously you're more interested in what some people have to say than others so you're going to have a low average value reading all the tweets, secondly it'll incentivize people spamming the timeline because they know their tweets will be buried by other tweets if people aren't reading live and that will damage how much influence they have. So, I don't think Twitter using such a simple algorithm would be good and I think a competitor would likely beat it with a better algorithm. So a state where Twitter doesn't use algorithms is likely to tend to a state where twitter is bankrupt and some other company is using algorithms.

I think it's pretty clear you need some algorithm. Ok, so maybe the line we draw is that we don't make value judgements on the content and only prioritize by meta data. Ie, your algorithm looks at engagement, reach etc. This is basically what Facebook did, and what they found was that it was much easier for people to produce highly engaging content by playing on peoples fears, misrepresenting information, and all the classic "click bait" tricks. High value content is expensive and time consuming to make, and since it doesn't get rewarded by the algorithm it gets priced out of the market. So sure, it's possible to draw that line, but again, I don't think that that's actually a neutral position, we know the type of content that benefits from content neutral algorithms. And remember - engagement at the end of the day is just a way of measuring what impact the content of the article is having, you're saying you want to promote content that is "engaging" as opposed to other priorities that you might have such as "truthful" or "High quality".

It's difficult not to think that what we're actually seeing is that the beneficiaries of the previous algorithms - the voices that were raised up by the facebook style algorithm are simply angry now that the algorithm has changed and not to their benefit. The author cites how popular their facebook pages used to be, but that wasn't a neutral algorithm either - they were just happy when the algorithm was rigged to benefit them rather than harm them.

I don't think you're entirely wrong but I think there is a big question of how prone an algorithm is to gaming and what measures the company provides to adjust it. For example, a simple chronological feed has drawbacks — your buddy in Europe posted something while you were asleep — but it also has advantages like not directing people to click-bait.

The fake news glut came from several areas where the major social media companies really weren't even trying:

1. Provide true negative feedback mechanisms: Facebook, Twitter, Google, etc. rewarded content which got clicks without having corrective mechanisms which actually worked. That provided a big incentive to crank out cheap agitprop since there was basically no way where you'd lose money as long as your costs were low enough. That also lead to a glut of ripping off other people's content, which appeared to be pretty safe as long as you weren't ripping off a major media company. They have a bit of a feedback loop now but enforcement is still quite lax and the algorithms will continue promoting things to strangers even after a lot of people mark them as junk or block the account. Not promoting things on behalf of your friends (e.g. Twitter making Likes public) or otherwise tamping down the “recommend this to other people” signal would be a good way to reduce spread without preventing anyone from seeing content they're actively looking for.

2. People could open accounts pretending to be media sources or authoritative in some way and on the off chance that the companies eventually did something, they'd just re-open a new one. A lot of the impact of fake news sites could have been reduced by requiring non-personal accounts to list verified ownership, country, etc. and other pages owned by the same entity — that'd both make it easier for people to see that Red White and Blue Breaking News is actually some random person in Macedonia and it'd increase the consequences of breaking the rules if you had to, say, burn something like a Dun & Bradstreet number rather than just creating a new page or getting another burner email address. It certainly wouldn't solve every problem but increasing the costs of bulk content-mill operations would make that business less profitable without preventing more legitimate businesses.

From a user-benefiting pov there is quite a simple solution. Just let the user pick their algorithm - including third party algorithms. Of course, that will never be allowed because it will likely reduce revenues.

N.B. Caitlin Johnstone journalistic style is click-baity. She might be being banned by humans. At first glance what she writes _looks like_ politically motivated junk. Potentially a human is not reading her articles in depth or perhaps they are, and it's still getting banned.

>The problem I see here is that there is no alternative to an algorithm.

That's not what she is arguing at all. She's pointing out the algorithms have intentionally been tainted. Biased. Censorship by omission. Algorithms aren't independent things, they are programmed and share the biases of their programmers.

Indeed I think she sums it very eloquently here: "You've probably noticed this if you've tried to search YouTube for videos which don't align with the official narratives of western governments and media lately. That search function used to work like magic; like it was reading your mind. Now it's almost impossible to find the information you're looking for unless you're trying to find out what the US State Department wants you to think."

If you happen to not have experienced that, I might point out you just might be in the right bubble to where it's not yet noticeable for you, or maybe not happened enough to be more than an annoyance easy to write off as something else. Perfect example: "The author cites how popular their facebook pages used to be, but that wasn't a neutral algorithm either - they were just happy when the algorithm was rigged to benefit them rather than harm them." When algorithms that used to distribute attention fairly uniformly now are biased to certain political ideologies to the extent that even the size of these services can't hide it - when does that become more than the mere coincidence you are trying to pass it off as? In your mind is there ever a threshold that would no longer pass a sniff test?

I have long noticed the usefulness of search features on the major services has dramatically declined, especially if the topics are even remotely political. So I no longer rely on search, especially for anything related to current events. I'm back to where I was pre-Google - a large, diverse group of sources I routinely visit. I've rediscovered the joys of RSS and self-hosted aggregation - it greatly helps one do this in a pretty trivial manner. No wonder Google first took over RSS with Reader then killed it. With such functionality I don't need some all knowing centralized "cloud" of knowledge to consult - I can easily gather information on my own and make my own decisions.

Heaven forbid that be allowed to continue. It's why it's past time for people to take control back themselves. It's a little more work, but well worth it if you really care about getting an unfiltered view of the world. I'm constantly amazed by how many times a topic will come up in conversation as being new and I'll usually have heard about it days if not weeks before - and often with significantly more detail and context.

And it's not really that new - before the internet you could spot people who relied on TV/Radio, those who read newspapers too, and those who routinely dug deeper into long form magazine or other articles.

What's really insidious about the Internet is it gives a great illusion to having unfettered access to all information without really living up to it in a way most people will never notice.

Until something they are passionate and have detailed knowledge of gets distorted. Then it sticks out like a sore thumb. The sunglasses in *They Live*, if you will. Just wait - censorship never ends. Sooner or later you will have your own jolt. The real question is will it be too late to do something about it?

"The avalanche has already started - it's too late for the pebbles to vote"

Tolerance or excuses for this behavior just enables it to be more firmly entrenched.

This article is only half-right. Some of the other things you don't see because the algorithm doesn't promote them are:

- scientists and educators that make 1 hour long videos to explain how something is incorrect. Example: https://www.youtube.com/watch?v=H95VCYLBh-A (less than 0.1% reach of the original)

- long threads of detailed analysis with cross-references

- long educational material that explaines e.g. confounders in order to explain why some statistical results observed are not possible to interpret in a certain way

The reason you don't see this kind of content is simple: the level of its engagement is low. Its not clickbaity, its not exciting, it doesn't have a rebel perspective, it doesn't claim to uncover something nobody has ever seen before. Its often kind of a downer too, pointing out other people's mistakes.

Social media sites make their money through engagement. The longer you stay on their site, the more ads they can show you. As such, their default algorithms promote engagement at any cost. It turns out that this trains creators to spread clickbaity, potentially misleading, false but exciting and novel "information" (see paper [1] as well as Veritasium's youtube specific explanation [2])

You know what other thing spreads a lot? Outrage. Extreme reactions to spread of misinformation are also engaging. People will fight over them due to strong feelings: one side frustrated by seeing the effort put into debunking having absolutely no effect (because it doens't get promoted by the algorithms, nobody finds out about it) while the other side feeling like the core values we have (free speech) are under attack.

Unfortunately, there is no way for social media companies to fix this problem without taking huge hits in revenue. The only way we can fix it is by becoming aware that what we're seeing isn't free speech, its some distorted picture of the world designed to engage us longer at any cost.

[1]: https://www.science.org/doi/10.1126/science.aap9559

[2]: https://www.youtube.com/watch?v=fHsa9DqmId8

> The only way we can fix it is by becoming aware that what we're seeing isn't free speech, its some distorted picture of the world designed to engage us longer at any cost.

I like your assessment, but the conclusion is depressing me. Because that awareness is never going to reach the wide population and simply understanding something, sadly, does not make you immune to it. You can understand the reason things are the way they are but still feel the frustration and revolt inside you. I know I do.

I have some hope based on the development of awareness of "yellow press" (https://en.wikipedia.org/wiki/Yellow_journalism). I don't see another solution that can reach "both sides".

I just wanted to add - for anyone that thinks the spread of viral false information isn't just as serious of a problem as lack of free speach is, I invite you to read this story https://threadreaderapp.com/thread/1460368038859919361.html

> You've probably noticed this if you've tried to search YouTube for videos which don't align with the official narratives of western governments and media lately.

You want your unpopular ideas to be promoted, and you think that it's your right. Everything cannot be promoted, what is there suggestion of the article? To promote the less popular ideas, the most extreme.

I see social problems there, but morality apart, it becomes a "Nazi Bar problem", you don't want to appeal mainly to niche audiences because money.

In the author's opinion. We should also remove HN voting system, as popular ideas to up and unpopular are hidden.

> You want your unpopular ideas to be promoted

She didn't say "unpopular". She said those that go against the official narrative, regardless of popularity.

> and you think that it's your right.

She didn't say that either. She's merely drawing attention to how social media secretly manipulates what is allowed to become or remain popular.

> In the author's opinion. We should also remove HN voting system

I don't think she said anything of the sort? Her article is very short on solutions. But we should be able to admit a problem exists and discuss it even if we don't yet have a solution.

> But we should be able to admit a problem exists and discuss it even if we don't yet have a solution.

I doubt there even is a solution. At no point in human history have things ever been different to the current situation - and for good reason.

Today it's SV tech bros, in the past it was priests, kings, bible translators, newspaper publishers, radio- and TV stations. This is nothing new and complaining about it is like complaining about the desert being dry.

It's especially amusing coming from a country were puritanical views are codified in media regulations such that nakedness and profanity is still taboo, whereas hate mongering, spreading misinformation and outright defamation and lies are promoted under the banner of "free speech". A tad hypocritical, but maybe that's just me.

Complaining is fine. No change ever started without complaining beforehand. If nobody complains, lots of people won't even notice the issue.

Wonder what will be next overlord though.

> She didn't say "unpopular". She said those that go against the official narrative, regardless of popularity.

That's always been a thing, though. The incentive for big platforms has never been to promote the most popular ideas, much less the most correct ones. It's to promote "engagement" to the greatest possible extent, while ensuring that they're showing advertiser-friendly content at least most of the tme. And big advertisers are easily swayed towards not letting their ads appear beside "misinformation" that contradicts official and quasi-official narratives. You can't get rid of this dynamic except by re-decentralizing search and curation of content.

> She didn't say "unpopular". She said those that go against the official narrative, regardless of popularity.

Yes, she said that because it sounds more sympathetic than “I am owed an audience” or admitting that many of the people who saw their reach reduced were in fact trafficking in mistruths. The message that this is the “establishment” suppressing the truth is a very powerful one which inspires a strong emotional reaction from many of us, but it's also far too simplistic — for example, she scare-quotes “fake news” to dismiss it without acknowledging that there was in fact a ton of provably untrue material heavily promoted by the major social networks during the 2016 election cycle to the point where many poll respondents would express belief in falsehoods. Valuing free speech doesn't mean we should pretend that there aren't problems with real impacts.

She's saying the opposite that algorithm discards what may be popular/trending. So she'd probably be for HN voting system.

Surely there's a difference between "promoting" something and showing me what I'm explicitly searching for.

Is a library promoting Critical Race Theory by keeping books about it on the shelves?

I think you're both right. I read the article and the author does seem to be suggesting what critics of cancel-culture critics always claim we're suggesting - that popular platforms must give "equal time" to all points of view. This bit of ongoing misdirection is doing a good job of muddying the definition of "censorship" to make it look like the actual memory-hole-ing that Twitter and Facebook and Google have been engaging in is somehow purely organic. So the person you're replying to is correct that the article is completely wrong - so wrong that I kind of wonder if the author isn't trying to pull some weird "false flag" pro-censorship move.

But that's not what is actually happening on these platforms.

You can search for something like "is coronavirus dangerous to me" on YouTube and you'll get a list of specific banal videos, which don't actually answer your question from about 12 months ago. It almost looks hardcoded. You can tweak the search parameters and you'll get the same thing - videos which talk about coronavirus, but not the thing you searched for.

No other topic has this applied to it. If you search for e.g. "is the ryzen 7 worth it" the videos vary, they're not all adverts from AMD and they will answer your question.

We're not talking here about what's on the front page or marketed, we're talking about explicitly searching for material and having that be hidden from view i.e. censored.

If I go to the library and pick up a copy of the communist manifesto, or Mein Kampf, or a book on Stalin, or whatever, the library isn't _promoting_ those at all, they're not expressing an opinion. There's been a weird shift in discourse recently (see no-platforming) which implies that allowing something to exist is promoting it.

No one wants their unpopular ideas to be "promoted" she gives anecdotal accounts of popular creators being gradually shadowbanned.

It is the very point that propagation of speach is not dependent on popularity but increasingly on alignment with powerful interests and it does a lot of damage. How could you miss that?

> she gives anecdotal accounts of popular creators being gradually shadowbanned

How "popular" are we talking, though? Taking a big picture view, "official" narratives are pretty popular whilst "alternative" points of view are niche. And modern AI-powered algorithms don't do a great job of promoting niche, long-tail content in general. For all we know, this might be the outcome of a random AI tweak that results in burying some content because it gets predicted as "unpopular".

Anti-wokeness is very popular. Not trusting the presidential elections is very popular. Not trusting big pharma is very popular.

All of these are algorithmically suppressed.

If the definition of popular is what you see on twitter, etc, then nothing popular is being censored by definition.

Source? I've seen this claim many times but is it just a feeling or is there at least a whistleblower?

There are some whistleblowers indeed, but for the most part many dominant tech companies semiopenly proclaim these groups to be "dangers to society" and many admit to use "throttling", shadow-banning, and "machine learning fairness" to reduce the exposure to these thought crimes.

The only reason there are limited numbers of whistleblowers is that Tech companies actively brainwash their employees to think that the marginalized opinions are just some "small dangerous fringe nazi thing", similarly to how the commenters I replied to seem to think.

Reply on news.ycombinator.com