The Algorithm and the Race to the Bottom

[Bootstrap]

Separating this into a topic on its own ...

It seems you advocate for putting the burden of choice on the government which could choose to put the burden on companies that provide content or platforms for content. Others advocate for the burden of choice being on the individual, along with whatever consequences that may bring.

I don't think that is a fair description of what I am advocating. I am in favor of letting people and platforms choose, but I would like to enable them to make informed choices.

I think a lot of what is on the internet, including everything from news outlets to social media to general websites and everything in between, is not generally expected to be scientifically accurate. A huge amount of it is just simply entertainment. If reasonable people are looking for what is true, they probably already have sources they know and trust. They take the rest with a grain of salt.

I think a lot of the Internet is less innocent than you describe above. And I think people have no way of knowing a lot of things without specialized expertise. Consider the current discussion of GPT - people believe claims they have heard on the internet, and they have no way of knowing if they are true. They assume that they really know what GPT is and what the algorithms are, or that their sources are telling them the truth.

When users sign up for a platform, they click on interest fields that say they want to discuss gardening and recipes and crafts, but the engine keeps pushing political content at them because it is the stuff that gets the most mouse clicks. And it gradually ramps up the emotional grab of the politics. It targets you and your friends at the same time.

Teens who sign up for social media because they want friends get introduced to sexual content, which also ramps up over time. Online, the sexual content that gets the most mouse clicks is porn, and it has an outsized influence on all discussion of sex and sexuality, even if teens are not allowed to view it on a given platform. So they go out and look for porn, which is where most teens get their sexual education first. Parents should know this. It's not just teens, of course, the sexual content that goes viral is also the stuff that leads to the worst behavior. Just as it does in politics.

Teenage girls who sign up for social media also get drawn into body anxiety and are more likely to be drawn to anorexia and bulimia. Social media usage also increases suicide rates.

Social media is widely used to recruit for sex trafficking. Child porn is distributed almost exclusively online - often with mothers offering to film their kids doing sexual acts for pay.

Here's what all of these things have in common: High-engagement content is being promoted. The stuff that keeps you doom scrolling or looking for the forbidden fruit. The stuff that has you clicking for more. That is what generates the most profit for the companies providing this content. They will push the boundaries. And they make it really easy to share to draw people back into the platform. "Going viral" is what they want most. They are willing to put up the guardrails that people ask for, but most people are not remotely informed enough to even know what guardrails to want.

All of this is having an effect on our society. It's one of the major reasons society has changed so much since introduction of the Internet. We should at least be able to measure what is going on and make it public. Most people don't know these things, they are not prioritized in the algorithm. Researchers in AI and GPT pretty much all know these things. Most are concerned.

In some areas, like child pornography and child sex trafficking, this is clearly in the government's role. In anything that borders on 1st Amendment rights, the government should tread cautiously, ensuring that independent parties can measure and report.

[Bootstrap]

I don't think that is a fair description of what I am advocating. I am in favor of letting people and platforms choose, but I would like to enable them to make informed choices.

Requiring platforms to make certain tools available for users doesn't quite sound like the same thing as "letting people and platforms choose." It sounds like something coercive.

It is definitely coercive - like requiring people to provide ways to know if they are dumping toxic chemicals in our water supplies.

Because of 1st Amendment concerns, it stops short of requiring them to report if they are polluting our water. Or telling them they can't do it. In some areas, like child pornography, it is already illegal, and there may be other areas where that level of intervention is justified.

But for now, I'm really just asking for easier ways for users to know if they are poisoning our water.

[Bootstrap]

On this sub-topic, I am now responding in this thread.

As it is, we have no way of knowing if Bootstrap is intentionally promoting false content over true content, or how he is manipulating us.
Nor do we have any way of knowing that about any of the major news websites, or for that matter even about whitehouse.gov.

I think the benchmarks threads I am creating give you some information on that. I have told people the prompts and methods I am using, and explained how they can do it themselves. I think the Covington thread does the same. In general, I think I try to provide people with information they can use to verify things for myself.

I think it is quite naive to think that companies can #1 cheaply, #2 easily, build tools that will #3 "make it really easy for people" to know how content is created/curated without #4 in the process creating an opportunity for people to learn how to game the algorithms and make their misinformation game less easily detected.

You'll excuse me for saying this, but as person who works in this field, I think I may be the least naive person in this conversation. I think I understand a fair amount about AI, GPT, APIs, this kind of research, and engineering software platforms at scale. I probably know more about the data that these engines use, how they analyze it, how it is customized for specific groups of users.

These companies already use tools like this internally to measure things they care about. If researchers and news sources can provide objective ways to measure what they do, and these are measures people care about, then social media companies will tweak things to make them look good. Of course, some social media companies may not care because their users already know what they are giving them.

At some point, the question becomes "Do we really need, or even want, a world where every last detail of everything including your bathroom schedule has to be subject to government regulation and publicly available. I get the impression that you think the world is made better when you have the ability to look into every closet and every drawer - and that the more regulation the merrier. I get the impression that you think government can generally be trusted to do the right thing. My experience and observation is that government cannot be trusted to do the right thing; that government will lie, cheat, twist the facts, hide the evidence, whatever they think they can get away with to get the results they want. I think that there is no practical difference between Meta or X post their algorithms, and making the NYT publicly disclose their internal deliberations on how and whether to report on the news.

My proposals basically amount to asking companies to make it easier for us to know if they are putting toxic chemicals in our water. They don't even tell them they can't do it, just that they have to make it easier for us to know.

See the OP for reasons that I believe at least THAT level of regulation is warranted. Transparency.

I believe that 1st Amendment concerns are real and important. I also think the balance of powers is real and important. And I think transparency is important. Particularly when the government starts working closely with social media, dark money is exchanging hands, and they are asking for practically no regulation at all. I do not trust the government that much. I do not trust social media that much.

[mike]

I don't think that is a fair description of what I am advocating. I am in favor of letting people and platforms choose, but I would like to enable them to make informed choices.

Requiring platforms to make certain tools available for users doesn't quite sound like the same thing as "letting people and platforms choose." It sounds like something coercive.

It is definitely coercive - like requiring people to provide ways to know if they are dumping toxic chemicals in our water supplies.

Because of 1st Amendment concerns, it stops short of requiring them to report if they are polluting our water. Or telling them they can't do it. In some areas, like child pornography, it is already illegal, and there may be other areas where that level of intervention is justified.

But for now, I'm really just asking for easier ways for users to know if they are poisoning our water.

The comparison to dumping toxic chemicals is interesting, but it's not the same thing at all. One is a scientific process. The other is a subjective process subservient to arbitrary standards set up by the government.

[Bootstrap]

The comparison to dumping toxic chemicals is interesting, but it's not the same thing at all. One is a scientific process. The other is a subjective process subservient to arbitrary standards set up by the government.

Maybe I'm not clearly explaining what I am asking for? It's certainly not what you describe. I think my analogy is fairly precisely true.

Do you know what an API is? What do you think I am asking for?

[mike]

The comparison to dumping toxic chemicals is interesting, but it's not the same thing at all. One is a scientific process. The other is a subjective process subservient to arbitrary standards set up by the government.

Maybe I'm not clearly explaining what I am asking for? It's certainly not what you describe. I think my analogy is fairly precisely true.

Do you know what an API is?

Generally speaking, yes. What I am attempting to critique here is the of dumping toxic chemicals into our water as being analogous to the posting of false or misleading information online. I am saying they are two very different things. The former seems to be a scientifically quantifiable issue and the second a much more subjective process.

[ken_sylvania]

On this sub-topic, I am now responding in this thread.

As it is, we have no way of knowing if Bootstrap is intentionally promoting false content over true content, or how he is manipulating us.
Nor do we have any way of knowing that about any of the major news websites, or for that matter even about whitehouse.gov.

I think the benchmarks threads I am creating give you some information on that. I have told people the prompts and methods I am using, and explained how they can do it themselves. I think the Covington thread does the same. In general, I think I try to provide people with information they can use to verify things for myself.

I think it is quite naive to think that companies can #1 cheaply, #2 easily, build tools that will #3 "make it really easy for people" to know how content is created/curated without #4 in the process creating an opportunity for people to learn how to game the algorithms and make their misinformation game less easily detected.

You'll excuse me for saying this, but as person who works in this field, I think I may be the least naive person in this conversation. I think I understand a fair amount about AI, GPT, APIs, this kind of research, and engineering software platforms at scale. I probably know more about the data that these engines use, how they analyze it, how it is customized for specific groups of users.

These companies already use tools like this internally to measure things they care about. If researchers and news sources can provide objective ways to measure what they do, and these are measures people care about, then social media companies will tweak things to make them look good. Of course, some social media companies may not care because their users already know what they are giving them.

Thank you for deigning to explain this to us mere mortals. Would Your Majesty be pleased to comment on the critical issue of how a "cheap" reporting API such as would be used for internal purposes would avoid enabling sophisticated purveyors of misinformation to use that information to refine their models to game the algorithm and avoid detection? I notice your response ignores the central core of my comment, which was that a cheap, easy solution will be vulnerable to abuse.

[Bootstrap]

I will respond to the last two comments either tomorrow or the day after. I have limited amounts of time for this, and I have to pace myself.

Ken: I think I did answer your question, but I will find a better way to explain what I said. I'm used to discussing these things with very different people in very different settings, and I think we are talking about technical things that take time to understand. I'll try to do better tomorrow.

I don't think most "mere mortals" have thought deeply about GPT technologies or understand exactly what they do. I also don't think most "mere mortals" know what I am talking about with the APIs. I need to think of a better way to explain that. Are there ways I can share expertise and facts without triggering some kind of "us versus them" here? I am not trying to say I am better than you or smarter than you, but I do think I have expertise in this particular area.

[Bootstrap]

I haven't forgotten - I still intend to do what I said, some time in the next few days.

[Bootstrap]

I finally got around to this - with the help of a little GPT.

First, the motivation:

The Invisible Force Shaping Our Lives

Imagine waking up in the morning and scrolling through your favorite social media platform. What you see—what grabs your attention and influences your thoughts for the day—isn’t random. It’s chosen for you by an algorithm. This algorithm doesn’t just decide what you see; it shapes the conversations we all have, whose voices are heard, and what ideas become prominent. But here’s the thing: we don’t know exactly how these algorithms make their decisions, and that’s a big problem.

When Innocence Takes a Dark Turn

At first, it might seem harmless. Someone who loves gardening or enjoys looking up recipes might notice their feed changing. What started as a peaceful hobby now includes heated political arguments or sensationalized stories. A young girl looking for connection and friendship might find herself inundated with content about body image, diets, and ideals that fuel anxiety, eating disorders, or even self-harm. Teenagers trying to navigate questions about relationships might be pushed toward explicit and harmful content they didn’t seek. People interested in understanding politics might get pushed further and further into extremist views.

This isn’t accidental. These algorithms are designed to keep us engaged—pushing us toward content that provokes stronger reactions, even if it’s not what we wanted to see. Why? Because our attention drives their profits.

Real-World Consequences

The effects of these algorithms go far beyond what we see online. They affect how we think, what we believe, and how we interact in the real world. They deepen divisions, influence public discourse, and can steer us toward decisions and behaviors we might not have chosen otherwise.

Shining a Light on the Algorithms

We need to understand how these algorithms work and what they prioritize. This isn’t about regulating speech or controlling content—it’s about gaining the information we need to make informed decisions about what we consume online. By creating tools to study these algorithms—tools that let researchers, journalists, and experts examine their effects—we can better understand their impact and, if necessary, advocate for changes that align with our values.

Protecting Free Speech, Promoting Transparency

Importantly, this initiative does not regulate content. It doesn’t tell anyone what they can or cannot say online. Instead, it focuses on making the invisible systems behind social media visible. It’s about accountability, not censorship. These tools will empower researchers and journalists to study how content is prioritized or suppressed—without interfering with free expression.

This approach defends First Amendment values by ensuring the rules of the game are transparent while keeping the government out of regulating ideas or conversations. It’s about giving us the tools to understand the digital spaces we rely on every day and enabling us to shape them in ways that reflect our collective priorities and needs.

And these days, the "Public Square" that matters most is online, governed by these algorithms. If we really want to protect free expression, we need ways to see how the algorithm works.