Hacker News Reader

Comment by concordDance | Hacker News Reader

The idea would be that the two of them collaboratively agree on some observable prediction they differ on. E.g. level of officially reported government spending in 4 years time or gdp growth rate next year or number of plane crashes next year or what have you.

Just some observable metric.

If they literally can't come up with a single observable predictive difference then the predictive aspects of their models are actually equivalent and they are only narratively different and don't "really disagree". Like Copenhagen interpretation vs many worlds.

View on HN

Many things don't have quantifiable metrics like that. For example, is USA still a democracy in 4 years? Are people more or less free?. You know, important questions that aren't just economic numbers. Even semi-quantifiable stuff like "are Americans better educated" is debatable on many topics if you can't agree on truth. Oh, and that GDP growth rate number? That relies on a lot of trust as to who's doing the reporting. For example, many people don't believe China's reported GDP numbers. What makes you think the USA doesn't devolve to such a distrust as well.

If they affect your life they can be observed.

If "democracy" is just metaphysics then it's irrelevant. But if it has actual tangible effects such as "can you vote?", "can you protest the government?", "is the leader of the opposition arrested?", "do most people think they live in a democracy?", "how popular is new legislation compared to previous years?", etc...

Then you can make predictions about it and test them!

You can even do local predictions if both can agree, such as "will the combined incomes of my family be higher or lower in 4 years time?" as low coupling proxies for gdp. (Ideally one would use probabilities for loosely linked proxies like that and use the probability differences the two theories assign to give bits of evidence to one over the other, so you'd want many many such proxies, ideally uncorrelated ones)

> can you vote? can you protest the government? do most people think they live in a democracy?

Was Jan 6 a protest of the government or an insurrection? Can Russians vote or are elections a sham? Do the majority of Russians believe they live in a democracy if they’re afraid of whose conducting the polling (or the MAGA non response to polling)? Those are values question that require you to have an agreement on reality.

> You can even do local predictions if both can agree, such as "will the combined incomes of my family be higher or lower in 4 years time?" as low coupling proxies for gdp

Your personal income has absolutely no predictive value on gdp. It’s more predictive of whether you personally made successful bets or even if you’re better at sucking up to the current power structure. It tells you nothing about population level metrics if you have no way of conducting reliable population level surveys. For example Donald Trump’s personal net worth sky rocketed under Biden because he won the election while as the leader of the opposition to the democrats was looking at jail time and whether that was legitimate or not depends on which political lens you look through it.

> If they affect your life they can be observed.

Ah, but if either side distrusts the other about whether the observation made is truthfully reported, how do you solve that? It requires some amount of trust and right now there’s a very clear divide there.

There are definitely tangible predictive differences in the case of, say, Russia vs USA. Things like "If you go to the capital with a bunch of friends carrying placards saying '$LEADER is corrupt and evil and should be replaced by $OPPOSITION' how many of you end up in a jail cell in the next day?".

If there is literally no tangible difference then it's just label games and metaphysics and doesn't matter.

> Your personal income has absolutely no predictive value on gdp.

It actually is correlated (admittedly in most day-to-day cases it's just a lagging indicator, but things like natural disasters hit both). It's not the strongest correlation but it would still be evidential. Definitely under 1.0 bits though... One would need a LOT of such observations and having them not screen each other off to start getting a convincing number of bits.

Probably not realistic to have humans manage these sorts of numerous tiny updates though...

/nitpicks

> Ah, but if either side distrusts the other about whether the observation made is truthfully reported, how do you solve that? It requires some amount of trust and right now there’s a very clear divide there.

Yeah, it gets much trickier like that. But I do think two reasonable people from the opposite political sides could agree on some sort of observable to the extent their disagreement is anything other than narrative.

> Things like "If you go to the capital with a bunch of friends carrying placards saying '$LEADER is corrupt and evil and should be replaced by $OPPOSITION' how many of you end up in a jail cell in the next day?".

If the other side calls it a violent riot does it still count as people getting put in jail? Cause the Jan 6 insurrection and BLM protests occurred at about the same time and are viewed very differently depending on which political lens you put on.

> If there is literally no tangible difference then it's just label games and metaphysics and doesn't matter.

You’re discounting feelings as if it doesn’t matter. But if people believe or feel like they live in a dictatorship, what quantitative data are you going to use to disprove that. Moreover, why aren’t feelings valid when talking about politics which is fundamentally an emotionally driven human activity and not a data driven one? By the way the left believes they live in an authoritarian dictatorship under Trump while the right believes they lived in an authoritarian dictatorship under Biden. And political power literally is the power to emotionally manipulate others because you individually can’t actually accomplish anything by yourself.

Reply on news.ycombinator.com