Hacker News Reader

Comment by shoes_for_thee | Hacker News Reader

Somewhat tangential...

I've been thinking lately that there is a need for a space in which political discussions can be had where people are as committed to the values of civility and good-faith argument as we are at news.yc.com

I have not a clue how such a space could be moderated. It's obvious to me that a person violating guidelines would quickly claim suppression if their posts were moderated, and the space itself would quickly devolve into the things it tries to avoid. Eg, echo-chamberiness, sophists, trolls.

It seems like a paradigm shift is required, or that the medium just isn't appropriate. It also seems widely-accepted by nearly everyone that our current medium is destructive and no longer serves constructive use. Reputation, ranking, and moderation based on the 'wisdom of crowds' is no longer viable in its current form, but I'm utterly unwilling to cede control to an authority...

How does one establish trustless good-faith?

View on HN

People are committed to civility at HN and good faith arguments mostly because HN is a small community, much smaller than twitter e.g. It is also inhabited by mostly people with intellectual curiosity, and accounts violating the guidelines get banned quickly in their infancy.

I followed quora since when it was a very small website. It was full of very useful and well written content. As the userbase grew, the quality of the content went on continuous downward trajectory. If HN had 100 million users, you would see the same content quality at HN, and the volume would make moderation hard, if not impossible. Not to mention, you might start seeing calls to cancel user XYZ on HN too if it reaches a userbase as large as twitter.

A lot of approaches to moderation seem destined to fail, afaic.

I think first principles should be:

1. There is no such thing as objective moderation. The universe gives us no way to conclusively prove something is good, fair or true.

2. You cannot moderate a forum at internet-scale using exclusively paid employees.

3. A tech company is not the government. There is no shame in a company looking out to protect its own good name.

Some takeaways: (1) and (3) suggest a social network should admit its point of view, and moderate according to its tastes from the top-down. (2) suggests the bulk of moderation should either be AI-based, or delegated down via a hierarchy of its users. Since AI won't be capable of competent moderation for at least a decade, the primary method should be the community. (3) means that the the positioning on moderation should be heavy-handed. Eg: "If we dislike your contributions or feel you make us look bad, we will ban you. We offer no 'court of appeals', because it's an esthetic judgment."

This ideology of my model is rather depressing, but the current state of the internet (compare HN or Tildes with 8chan or Gab) hints that it's better than the status quo.

> 2. You cannot moderate a forum at internet-scale using exclusively paid employees.

Sure you can. You just may not be able to print money Facebook style while doing so.

I agree that these companies could afford to hire tens of thousands of people, and have them follow a flowchart. I don't see that environment ever delivering competent moderation. Not to mention, it's unethical to subject employees to several hours a day of troll vomit.

It needs to follow the rules in place for advertising and completion law. Deliberate deception should be handled with a heavy hand, most other things should be community managed.

I like the idea of moderating through providing feedback on the arguments people make. So it isn’t enough to moderate for civility, it’s moderation of the discussion. Like flagging things for logical fallacies. And also promoting those who make sound arguments. Like literally a debate moderator. I think if enough things are flagged people’s pattern matching begins to pick up on those fallacies everywhere and they would become better debators

I took a stab at this a couple of years ago and ended up in some interesting discussions. Mostly what I found is that a lot of times people are arguing for one policy when they really mean for that policy to be bundled with a second policy. For example, in the USA, shelter-in-place is a good response to COVID if it is paired with a social safety net to handle lost wages. But if you don't trust the government to hold up the safety net, then maybe a masks-only policy would have been better.

I called my discussion method, Humble Americans for Math and Reason (HAMR) and it was sort of tongue in cheek because I knew how annoying it would be if that was an actual group trying to insert itself into conversations all over the place.

The discussion system boiled down to three guidelines. Do the math. Be transparent about values. Be humble about the math.

Here's the longer explanation:

-------

#1. Do The Math

The main goal of HAMR is to promote mathematical reasoning when we evaluate problems and solutions. For example, the media often covers new dangers that are probably overblown. The novelty of the danger far exceeds the actual risk.

The HAMR style response would be to develop a standard of safety that everyone can relate to.

Let’s call it the car standard, or Automobile Fatality Comparison Scale to be precise.

One person dies per 2 million hours driving in a car.

So, when we hear a story that someone died during a Tough Mudder competition, HAMR’s job is to apply the car standard.

The math looks approximately like: 750,000 people have entered a Tough Mudder, the average Tough Mudder takes three hours to finish, and there has been one fatality.

That’s (1 death / (750,000 participants * 3 hours)) = 1 death per 2.25 million hours.

In other words it’s a 0.9 on the Automobile Fatality Comparison Scale, i.e. 10% safer than driving.

Now people can make a mathematically reasoned decision about the dangers of a Tough Mudder.

-------

#2. Value System Transparency and Neutrality

Reason is actually a value system — but let’s ignore that.

What HAMR needs is to be able to be both transparent and neutral about other value systems.

For example, if we were to analyze the Iraq War we could look at several value systems: * Imperial: will the United States come out ahead in terms of opening and protecting new markets. * Respect for Human Life: How many Iraqi’s will die in the conflict? What change can we expect in their quality of life? * Respect for American Life: How many Americans will die in the conflict.

One of those three angles is much more compelling to my value system. But the point of HAMR isn’t to convince you my value system is better than yours. It’s simply to help people be better at mathematical reasoning.

And I picked Iraq as an example, because I’m nearly positive that the Iraq War was a bad decision by almost any value system: * The Iraq war cost the US: $2 Trillion dollars (i.e. $6,289 per US citizen) * Led to the death of 4,486 US soldiers. * Led to the death of 110,000+ Iraqi civilians.

For every calculation, there needs to be transparency about the value system prompting that calculation.

Then, having transparency about the values, there needs to be a huge amount of respect for other people’s values.

It’s an oxymoron, but the HAMR party is non-partisan. Democrats, Communists and Republicans should be able to contribute together.

-------

#3. Embrace Humility

Anyone who spends any time with data knows that data is messy. You need to be open to better data and better math at all times.

For example, I just gave numbers from the Iraq war above. I got those from a Google search. Do you have better numbers or a deeper analysis of how those numbers were calculated?

Undoubtedly, someone does.

I really rather like this approach. I have 2 comments which are that some people are so fixed on one view that they really cannot change. Give them the maths, watch them ignore it.

Example: in a drugs discussion I posted some info showing nicotine to be minimally addictive, complete with reference in the form of a link, extract from the link so people didn't even have to click on it, and added my own experience that I did not find nicotine alone addictive you need a 2nd type of chemical called an MAOI which makes nicotine very addictive wgen taken together). That got downvoted because nicotine = bad. All the facts you like but some react by instinct.(Note, I use nicotine/drugs that just as an example).

The 2nd point is unconscious bias, which you have yourself. You've even included 'Anmericans' in the name, more or less implicitly discoubnting others - plaese don't! We exist in other parts of the globe.

Anyway, I really do like the principle.

The acronym didn't work as well if I extended it to Earthlings. HAMR > HEMR.

Agree not everyone values changing their minds. But since you do, I think you'll enjoy this article about falsification. The idea is to think in terms of "I believe X, but if A, B or C were true then my belief would be false."

https://medium.com/better-humans/the-falsification-mindset-h...

I'm already aware of popper and his falsification, but I'll spend a bit more time on the article, thanks.

My problem is I can easily and happily change my mind if the evidence is concrete but where it is more a matter of common sense I tend to fail; "This is interesting and fun and may be useful - is it worth investing the time in?" And because it is all these, I seduce myself into spending the time on it whereas I suspect it may not be the best use of my time. Deffo not my strong point, that...

Going back to the original issue of evidence swaying (or not) people, I really feel lost looking at how to some people, reality is just a hindrance to what they know to be true.

I love these guidelines.

The 2nd one feels like the starting assumptions for a mathematical proof. Everything in the argument has to start with your values.

In most debates, people talk over each other because they haven't agreed to disagree on their value systems. They don't focus on the flow of reasoning that follows.

> "people talk over each other because they haven't agreed to disagree on their value systems"

Love this summary.

Reply on news.ycombinator.com