Kenneth Arrow Has Died | Hacker News Reader

Kenneth Arrow Has Died

nytimes.com

158 pointsmsabalau26 comments

This is possibly one of the slickest arrows theorem videos, made by the exploratorium last year.

Wow I never knew the exploratorium made youtube videos. That was one of my favorite places when I was a child.

Thank you for that video - it was wonderful. Are there any other channels that focus on things like this (logical stuff)? I like CGP Grey, so I'd love suggestions that are similar.

You might like 'undefined behavior'. The following video covers 'Range Voting' in relation to Arrow's Theorem:

https://www.youtube.com/watch?v=e3GFG0sXIig

Related: For hackers, a well know example of community using a method other than the simple majority rule is Debian. They use the Condorcet method [1] for the election of their project leader [2]. It's a ranking method «that elects the candidate that would win by majority rule, in all pairings, against the other candidates».

[1] https://en.wikipedia.org/wiki/Condorcet_method [2] https://www.debian.org/vote/

It's a bit more precise to say that they use a Condorcet method. There are lots of different vote-counting systems that are Condorcet methods; the one they use is particularly refined:

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

Didn't knew that, thanks!

Really appreciate your sharing this video. I will use this when discussing voting paradoxes and how poor most of our electoral systems are, given what's possible.

yeah! it actually came to my attention because my friend's girlfriend's brother is the voice in the video (he is a tech/master builder at the exploratorium - with, a very alluring voice). I got to see the early version and corrected a misspelling/mispronunciation of "monotonicity".

Two of his interesting contribution were his Arrow-Pratt measure of risk aversion, which is invariant to affine transformations (so consistent with revealed preference) and his Arrow-Debreu general equilibrium proof. Both are quite intuitive:

https://www.google.co.uk/url?sa=t&source=web&rct=j&url=http:... https://www.google.co.uk/url?sa=t&source=web&rct=j&url=http:...

Here's a picture of me hanging out with him at his home in Palo Alto a couple years ago.

https://www.flickr.com/photos/brokenladder/9002915694/in/alb...

He was a really nice guy. He signed some of his books for me.

He did so many things. I personally think the most interesting/stimulating contribution was his paper "The Economic Implications of Learning by Doing" from 1962.

You have made me curios, so I'm going to read the paper. It lives under this DOI [10.2307/2295952](https://doi.org/10.2307/2295952) (paywalled).

I am having a hard time understanding the paper. Would one of you be able to summarize the implications of the paper?

I found this part interesting: "I advance the hypothesis here that technical change in general can be ascribed to experience, that it is the very activity of production which gives rise to problems for which favorable responses are selected over time."

There is actually a little blurb in the linked NYT article on this topic:

Take “learning by doing,” a notion that Professor Arrow examined in the early 1960s. The basic idea is straightforward: The more that a company produces, the smarter it gets. Decades later, economists incorporated this idea into sophisticated theories of “endogenous growth,” which have a country’s rate of economic growth depending on internal policies that promote innovation and education — the very forces that Professor Arrow’s writings anticipated."

edit: I always forget which markdown pieces HN uses/doesn't use. Italicized rather than failed attempt at blockquote

Add this to the search string: "http://eclass.uoa.gr" should lead you to a non-paywalled version of unknown legal status.

The paper is freely available at: http://faculty.msmc.edu/hossain/grad_bank_and_money_policy/t... (you should always try to search for a copy on Google Research)

Experience would presumably need to be retained and transmitted to have the desired effect in reducing the costs of new models in production.

I'm wondering how our modern approach to workforces facilitates knowledge retention?

At the scale of an economy though, a flexible and transferable work force could facilitate the dissemination and expansion of access to institutional knowledge, benefiting both individual organizations (which receive as well as lose knowledge) and society as a whole. So I imagine there's a balance to be struck between retention and transfer.

The 1990s management fad of "core competencies" addresses this, so long as your correctly identify core competencies.

Studying Arrow's work encouraged me to think about the complexities of aggregating preferences. I look forward to reading more of his other work. With economics there is more than meets the eye.

Reading Arrow is one of the things that warmed me up to the European School (Roy et al). Try to avoid policies that create big losers, instead of trying to optimize a national utility function. It's also really interesting that in the 1970s the Soviets were trying to do the latter -- it's one of the things they wanted to do with their crazy mirror-universe arpanet.

Read "Gaming the Vote".

Also...

http://ScoreVoting.net/ArrowThm.html

The story at the end is really good.

more at http://www-siepr.stanford.edu/ArrowShovenMay09.pdf

Is there a source for that Spencer research?

OT: speaking of whales, that reminds me of something I've wondered about.

I recall reading an article a few years ago about some species of whale that lives in the Atlantic and spends the winter in warmer southern waters, and then migrates north to their breeding grounds.

The article talked about how researchers had put trackers on some of the whales, and found that the whales followed much straighter paths than the researches expected. The expectation had been that there would be a lot of east/west variation along the paths of individual whales, because all of the geographic inputs to navigation the whales are thought to use (magnetic sense, position of sun, maybe polarization of light from the sky, visible landmarks, and so on) have a fair amount of uncertainty.

I did a bit of Googling, and found that there were tens of thousands of these whales. The article didn't say how many whales had trackers attached, but I would assume that it is only a very tiny fraction of the total population. I also found that the migration is spread out, not all at once. Some whales will leave, then a little later more will go, and so on, so that at the height of the migration whales are all along the general route.

So what I've wondering is if the overly straight routes of the tracked whales could be explained by most of the whales listening for the calls of the whales further along in the migration and swimming in the direction of the calls, and giving that priority over following geographic navigation clues. Since they only track a small fraction of the whales, I could see them easily only having data only from whales that are not near the leading edge of the migration.

The reason I think following would lead to straighter routes is that if B is following A, with a gap of several kilometers (which should be easily doable, because whale sounds can be heard quite far away under water), then if A deviates side to side by a certain amount, B will also deviate, but B's deviation will be smaller. The farther back B is, the smaller B's deviation will be.

Now if there is a C following B a few kilometers back, C's deviations will be even smaller than B's.

With a spread out migration, these chains should be long, and those whales back a few steps in the chain should be going quite straight. If you only have data on the whales that are a few steps back or more, it would look like the whales have some super accurate navigation system.

Reply on news.ycombinator.com