intended

https://techjacksolutions.com/ai-brief/a-200-subscription-co...

I'm sure there are people here that will vehemently disagree with me, but we see the biggest tech companies pushing for age verification and we see founders and rich folk gleefully giving up their earlier pro-privacy stances in favor of supporting locking down identity.

I vehemently disagree, because this is not what is happening.

The Age verification domino toppled first in Australia, and then other governments found the example was good enough and followed suit.

The issue that HN conversations miss, is that the dominoes were set up over years. People have constantly been trying to deal with the many, many issues thrown up by social media. Issues ranging from the Myanmar genocide, content moderation, fraud, child safety, sextortion, to name just a random grab bag of issues.

Voters, governments, NGOs, victims and even tech firms, have been trying to figure out what to do for a decade+.

Voters, and non-tech-literate society used to complain about the status quo. The political will to change it reached critical mass, and is now in progress.

I believe the added point was that new tools identify a group of bugs, and then there aren’t more bugs to find. At which point it becomes another tool you use and don’t think about.

Also, from what I recall, Anthropic and OpenAI subsidize their prices by ~40x?

That said, given the prices*quality of recently released open models, I think the cost issue is moot.

I don’t think the author is arguing for compromise though?

I think the author would love to see more enlightened policy, but he doesn’t believe it’s going to happen. Hence the closer: “ Please make my predictions come out wrong.“

Also, these laws don’t simply originate from big tech. That particularly OSINT study was done by someone who wrote their own interpretation of the data points they uncovered, to create a narrative that was comforting to them.

Big tech has been aware that a reckoning was coming, because they and safety advocates have been fighting for years now. The fact that there was no actual teeth or ability to enforce e regulation, and that harms were evolving beyond reason and imagination was general knowledge.

Meta is lobbying to influence the shape of the manacles they know are coming.

It likely would! It is also probably the better solution all in all.

However society doesn’t have the stomach for that (yet).

Also, as the article correctly identified, society is gearing up to deal with other types of crime online.

Why do you use yourself as an example of parenting?

The issue isn’t the parents who can’t imagine bad parenting.

As a counter point, the larger point that the author was making resonated with what I have seen.

As in - women in tech who value privacy and rule of law, and are also advocates for child safety and (better versions of) age verification.

In addition, for obvious reasons, you see more (more being the key word) women showing up for issues like NCII.

While the root may have been religious, Lions and Rotary were things in other parts of the world, shorn of their religious connection. I know that my grandfather was a Lions club member, and that was in India.

Generally clubs of that nature have seen their membership drop, have they not?

All organized things started going down in participation no?

That isn’t the tool they are selling though. They say it is, and they promise it will be safe, but you cannot get it to say that without also crushing its capabilities in other domains.

There are also cohorts of users who will vehemently argue against “parental attitudes” and impairment of their experience with LLMs.

It’s taken 20 years to get pushback against social media. God knows how long its going to push back against whatever social malaise AI will create.

The opening example is of a person listening to their daughter’s voice on an unknown number, how would calling them back help? Or am I missing something obvious?

The Steerability point is one I would want to see more on.

This is an issue for tasks like content moderation and labelling. Judgements like this are subjective, highly dependent on context and generally messy.

Theoretically, you supply a policy and content, and the LLM labels according to the policy. In practice, the model has inertia which means you don’t get what you expect. Your large 5 page policy document only provides a minor improvement over a one line policy.

The other issue is that you may create carve outs for content in your policy, but the model will still flag it as violative. No matter how strong the carve out.

The most recent work I know of here is Zentropi’s policy steerability benchmark. They give a model the same content under two policies — one that says flag, one that says allow — and only score the pairs where it gets both right

If I am reading the numbers correctly, Opus-4.6 lands at 0.52 steerability — but that’s 0.97 positive accuracy against 0.54 negative. It flags almost everything it should, but 47% of the time when it shouldn’t. Sonnet, which is more deferent, is (somehow) less steerable.

I think this also implies that safety and Steerability are antagonistic to each other.

Yes! If anything both, the supply of veterans and the ratio of judgement capacity to generated content, will only go down with time.

So where are the new veterans going to be coming from? You can’t spin up experienced workers overnight.

Essentially the same supply chain issue which nation states are worrying about today, except for skill.

Privacy is losing ground, despite people understanding the stakes.

Privacy advocacy is losing the battle, because it is being framed as a choice between privacy and the status quo, and people vehemently dislike the status quo.

I’ve seen Roblox invest and conduct child safety research for several years now.

Maybe they are expanding into other industries, but the safety focus predates that.

You responded to a detailed/narrow example flagged as an example, with a reiteration of the rule.

Worse still, the lack of significant savings DOGE, indicate that the government was generally efficient.

This means that while specific cases may exist, the vigilance of the public and the set up of the US system had kept average performance in bounds.

Your model has a major gap in it.

Markets don’t come into existence ex nihilo. They come into existence by regulation setting the rules to ensure that the system works for humans.

You have nothing without regulation and government power, setting up the rules and enforcing them.

Further, when we look at the merits of the case which is driving this entire conversation, Meta acted in a manner that most people would consider evil.

And the market is rewarding them for it.

People prefer the more engaging option, based on testing when preferences are offered.

You aren’t really putting a fair choice here. A product that is tested and refined to be viscerally attractive to our brains, vs will power.

This means you have the illusion of individual choice, with practical addiction at scale.

Subreddits get banned for egregious shit. Reddit stays on the side of letting subs up, rather than banning.

Reddit has made it harder for mods to find out which subs you are active in, so that you can’t be banned for activity in subs that are in cold wars with each other.

Isn’t this all left leaning based on US standards?

That position, and this is true of many current discussion on free speech, is suited for the fights we fought and not the situation we face.

I do understand the challenge inherent in combatting inaccurate representations of facts.

This is a tactical issue, however the larger macro picture is very different from even the era of cable news.

So: > You can't ascertain what's true without free speech.

True, but the closer read is: > free speech is a critical component to a free, fair and competitive market place of ideas.

This is going back to the Abram’s dissent. The point of free speech was to ensure that a crucial tactic remained available for a competitive economy to exist.

The problem has since evolved, and the market place is no longer competitive.

This is for a variety of reasons, none of which go back to free speech.

For example:

1) The average person, is going up against content crafted by teams of people whose job it is to figure out how to convince or confuse them.

Individuals have free speech, but at scale the outcome will move in one direction.

You can argue that people can generate counter speech, however:

2) The velocity of content generation has increased: By the time content is debunked, a new crises can be brought up.

People have limited attention to use in a day, so its possible for the more resourced party to keep speaking and “flood the zone”.

3) Verification is hard, generation is easy: V/G - the more content we generate, the less the ratio of verified content is to generated content. This means that the average seek time for individuals goes up.

Free speech has been respected in all those cases, however the competitive exchange and evaluation of ideas has been hosed.

How? By that method you could compare a pebble to the moon and say you were extrapolating.

The only way you would do that is if you didn’t understand the shape/limits of the structures being compared.

I think your question needed to be discussed 10 years ago.

I don’t know the answer to your example, but I would very much like to know what the canonical answer is, and what the pros and cons are.

It’s not a topic that gets as much attention, even here on HN.

Yes! it should be:

Verification/generation

I’d scope it down further than cognition.

Cost of generation has been reduced, and is highly subsidized currently.

Cost of verification has effectively not changed. I’d say as a rule of thumb: verification is the tough part.

Our brains don’t fare well under constant review pressure. https://en.wikipedia.org/wiki/Ironies_of_Automation

Rate of generation/Rate of verification is a proxy for signal to noise ratios, just for work.

That ratio has changed, and verification is the hard part.

Verification is the point of all markets (and a decent part of human civ as well).

And review isn’t cost less - https://en.wikipedia.org/wiki/Ironies_of_Automation

We are all going to discover the Ironies of Automation.

https://en.wikipedia.org/wiki/Ironies_of_Automation

https://ckrybus.com/static/papers/Bainbridge_1983_Automatica...

They are difficult lawsuits for sure! But that is because the status quo is already messed up.

These cases are going to bring up questions on content moderation, engagement, ad revenue, age verification, and a whole host of insane things that people don’t want to think about.

This does not need to be the case though.

After so many years, the question on the societal utility of meta and similar services is finally being forced.

No matter what the verdict, the design and limits which make effective policy is still to be negotiated and figured out.

Yes and no.

At the scale of these firms, it is an issue of incentives, more than it is personal responsibility.

I know that many of the people who worked in safety flagged issues. I know NGOs and victims reached out to people in the firm over and over again.

Humans wanted to do the right thing. Its just that for other humans, they had to answer to shareholders and they had a far stronger set of incentives to ensure they made “number go up”.

Your string of people doing the right thing, makes little headway in the face of the tide of other humans who have incentives to increase time on site.