In Search of the Science in Economics | Hacker News Reader

In Search of the Science in Economics

bloombergview.com

11 pointsbpolania6 comments

"You don’t need matrix analysis to know that if you tell everyone to wash their hands, infections will go down."

No, but you will need a well designed study or two, maybe a few accidental experiments (situations where a single variable is changed among two randomly selected groups of people), to tell if it's actually true.

And it may be that your hypothesis is incorrect. It may be that telling everyone to wash their hands actually makes infections go up in some specific contexts. I can imagine a couple scenarios where that might be the case. One is the false sense of security of the washing hands message leads to a slightly sloppier application of other protocols such as sterilization of tools or cleaning furniture. Or the means of spreading the message its self, say on plastic posters, could be a breeding ground of airborn bacteria. Sure, these are highly improbable, and there's even more I can't imagine, but I wanted to point out that the original hypothesis isn't as airtight as the author intended... if taken literally.

Damn straight! THere are a lot of things in medical science where people say "Oh, this is obvious. X causes Y, so if we eliminate X, Y will go away." Running careful experiments with proper controls and randomization is the only way to come to the appropriate knowledge required to make good policy. This is just as true in economics and every other science as well.

Nobody argues that social sciences are useless because it's impossible to find patterns. People argue that social sciences are useless because they find random pattern in a specific data and don't attempt to reproduce the result. And when attempted, turns out that most of them are just random patterns that only occur in that study data and don't correspond to everyone.

The main problem of social science like history or economy is that they are political shaped sciences. Both pick those pattern that fit into their world view, to form a religion like truth of how the world worked. And state payed historian or economic scientist will normally preach the view of their system.

I did not have time for this year Riksbanks Rris, but e.g. in 2013 we saw a typical price " should discard its mathematical pretensions " and better just tell its political agenda straight. Instead they used a positivistic computer simulation, which basically means: I have a row of theories (CAPM, OPT, EHM, ...) that have several unknowns and I tweak their parameters till simulation shows the results that I want to show.

So they find random pattern in specific data. It wouldn't be bad if their results fit some political agenda, as long as the results are truth. Problem is it's not just a matter of showing results that fit political agenda and remaining silent about ones that doesn't fit. Problem is that almost everything, fits agenda or not, is false, ireproducible random noise. You almost never see the results building on earlier results in social science, like it is in all other sciences. You only see "studies" spewed out like from some kind of social study generator, and none of these matter even to themselves as they are forgotten as soon as they are published.

the idea that economics is only a social science

That's exactly how I see it. And the only way to put science into “social science” is the proper application of statistics. Kurt Tucholsky's observation that sociology was invented so people could write without experience is still true today.

Reply on news.ycombinator.com