Wednesday assorted links

Comments

2. I believe this sort of data, but possibly because I do believe it, I under-consume and miss some happiness. Conundrum.

Comments for this post are closed

Actually– no chance she's First Lady anymore… Guzmán lost hist appeal to the National Board of Elections and his candidacy was overruled this morning.

Comments for this post are closed

2. The most important (cowboy/male) things in life according to a country song: Younger women; older whiskey; faster cars/horses; and more money.

My experience, more money makes you want more money. Whether it makes one more happy depends on the individual and facts/circumastances.

Don't forget a good dog to go fishing with you.

Comments for this post are closed

But it could buy me a boat.

Comments for this post are closed

Money doesn't buy happiness. But you can rent it.

Or, if you can be just as happy with money as without, I prefer to have some.

Comments for this post are closed

Comments for this post are closed

#3 is great

It is the quintessential puff piece. Not a hint as to her husband's policy platform, other than heavy handed illusions to him being an outsider, plus a warm fuzzy foray into the culture war. It should be taught in J-school for propaganda 101. Instant classic!

Like anybody cares about Peruvian politics lol

Comments for this post are closed

What is interesting is that marrying foreigners seems to be pretty common for Peruvian presidents and presidential candidates:
https://en.wikipedia.org/wiki/Eliane_Karp
https://en.wikipedia.org/wiki/Pilar_Nores_de_Garc%C3%ADa
https://www.google.com/search?q=Mark+Vito+Villanella
https://en.wikipedia.org/wiki/Pedro_Pablo_Kuczynski#Family_and_personal_life

Comments for this post are closed

Comments for this post are closed

That's all he could pick off the DC scrap heap? Even for a third world president of an insignificant country, he could do better than that.

Comments for this post are closed

Comments for this post are closed

#2: Left unconsidered is the indisputable proposition that if money does not buy you happiness, it nonetheless allows you to cry in comfort.

#2: The article, pretty much starts with the research that supports the idea that more money increases happiness. It's just a geometric increase and it takes a lot of additional money to make a little bit more happiness.

Can I interpret the Wolfers-Stevenson "Life satisfaction and income" graph to mean that once I hit $2,048k in annual income, I will exist in a state of maximal bliss?

I think you could interpret it to mean that you'd have 0.5 less bliss than someone making $4,096k in annual income, but 0.5 more bliss than someone making $1,024k in annual income.

But in reality, the chart doesn't show any data greater than $64k in annual income.

http://www.brookings.edu/~/media/research/files/papers/2013/04/subjective%20well%20being%20income/subjective%20well%20being%20income.pdf Figure:2

I'm extrapolating, but the y-axis label says "10=highest possible".

In reality, if the effect is so constant across such an immense range of incomes, I don't see why it would stop at any particular nearby point. It might eventually stop, but I can't easily guess where.

Comments for this post are closed

Comments for this post are closed

I would settle for minimal bliss.

Comments for this post are closed

Comments for this post are closed

I really liked this logarithmic model of happiness. I don't think that there is much of a cut off and it fits with my observations.

happiness ~= log2(income)

is a pretty good model.

Comments for this post are closed

Comments for this post are closed

Comments for this post are closed

5. They have goods that they can barter.

Yeah, I thought it was obvious too. Not sure I learned a thing from that article.

That was my reaction and surely Professor Cowen is aware of this too.

Comments for this post are closed

Comments for this post are closed

"Sometimes Better to Use Another Username"

LOL, yes.

Comments for this post are closed

It did say brides and the Chinese sex industry.

What is interesting about that article is that they also say young men are forced into the Army for a very long during their peak-escape years. That is probably something few people think about.

Comments for this post are closed

Apparently male privilege gets guys an all-expenses paid return trip to one of those special all-inclusive North Korean Superfun Holiday Camps.

I'm waiting for Hillary Clinton's take on this to follow up on her statement that, "Women have always been the primary victims of war. Women lose their husbands, their fathers, their sons in combat. Women often have to flee from the only homes they have ever known. Women are often the refugees from conflict and sometimes, more frequently in today’s warfare, victims. Women are often left with the responsibility, alone, of raising the children."

Maybe there's a corollary about why the vast majority, up to eight times as many, of the asylum seekers in Europe are male..

LOL, "women are always the primary victims of war because they survive while men die"

I forget if it was on this blog that I read rape is a bigger problem for men in Africa than for women, that many men are crippled by being repeatedly raped by soldiers and that their wives universally leave them if they find out they have been raped. But no one even wants to acknowledge the problem because they are afraid it will take away funds for women.

Comments for this post are closed

Comments for this post are closed

Comments for this post are closed

#1: You have misspelled his name.

Comments for this post are closed

#6 I have to admit that every time I see computers beat a human at something it bums me out. I feel like humans are a little bit diminished. I know that humans coded the software that did this, but still, it bugs me.

The big computer advantage in Go is that it won't make one-off mistakes. Even the best human will, 1 time out of 100, screw up a basic process or fail to recognize an opponent's strategy that they normally would. Computers always play the best game they are capable of.

Looking at the ending board position, it seems that Lee failed to make a crucial cut against one of the computer's formations, and the computer maneuvered him into giving up a large uncontested moyo (area of the board that you will surround and turn into territory unless specifically prevented). Had Lee managed to invade the moyo or cut off the computer's formation, he would have won handily. I'm sure in retrospect there are 1 or 2 key moves that, if he could do it over again, he would do differently and win.

Had he won, there would be no such instances on the computer's side--it would be unable to make that kind of mistakes, or realize them after the fact.

The question is what approach the programmers are taking. With chess, the computer cannot consider all possible moves. There are too many of them. So the computer looks at the player and examines his most likely moves. It looks at the most promising possible moves the opponent might make. And Big Blue successfully freaked his opponent out because playing a computer was so new and hence so weird.

Much of that is likely to be even more true with Go. So humans can cope. If the computer studies its opponents, its opponents can study him. One approach is simply to make a dumb move. If it is dumb, the computer will not have examined that possibility. It won't have time to go back and re-examine every possible option that results from that move. It will have to discard all its previous calculations.

Commentary says that Sedol tried that with his opening, and now regrets it

Comments for this post are closed

While Go has a very large branching factor, the magic of AlphaGo is that it can prune the tree effectively using it's value network. The idea that playing 'dumb' will some how prevent it from going quite deep into the search space is ludicrous.

I think that Lee attempted to play very agressively in an attempt to confuse AlphaGo, but in the end he is the one who got 'confused', this is understandable, determining the values of the moves near the end of the game was very confounding.

He will probably play defensively today and I would guess that we will see a narrow win by AlphaGo.

AlphaGo won match 2. Se-Dol, at least, doesn't think it was close.

http://arstechnica.com/information-technology/2016/03/google-ai-begins-battle-with-humanitys-best-go-player-tonight/

Comments for this post are closed

I expected 5-0 but by smaller margins. It was known that it was superhuman.

I'm going to guess that running out of time hurt Lee.

Comments for this post are closed

Comments for this post are closed

Comments for this post are closed

Comments for this post are closed

Do you expect machines to be better than you at drafting a legal brief? Driving a car or truck? Writing music or poetry? Designing a skyscraper or train station? Inventing new and smarter machines and then programming them in ways you don't understand? The world is about to change for all of us.

I expect my autonomous car to drive significantly better than I do. And hugely better than the lady who passed me on the left yesterday texting with both hands.

And she wasn't using her arms to steer. There was nothing touching her steering wheel but good intentions.

Comments for this post are closed

I drove past a guy practicing a piccolo trumpet a few months ago.

Comments for this post are closed

Comments for this post are closed

Then what will you be doing in 70 years? Or your children and grandchildren?

Comments for this post are closed

Well that's the tough part--naming a valuable endeavor that will not be dominated by machines in 70 years. Or 35 years or 100 years. The debate of how long it will be is now not even an order of magnitude.

Comments for this post are closed

Comments for this post are closed

Cheer up!

After all, the ability of a computer to beat the world's top Go player is just grabbing low hanging fruit...

Comments for this post are closed

Demis Hassabis, the founder and chief executive of Google’s artificial intelligence team DeepMind, the creator of AlphaGo, called the program’s victory a “historic moment.”

Demis Hassabis was born and grew up in North London. He is of Greek Cypriot and Singaporean descent...

I knew it, I detected a Greek surname and now I see he's mixed race, the best kind. Explains everything.

Comments for this post are closed

Comments for this post are closed

Disappointing to see another otherwise good paper cite Woolfers ahead of Stevenson. How to make female economists happier. Looks like a free lunch to me (would the male lose that much utility if the correct emphasis was used in citations)

The article does have some places where they are cited as S and W and some where it is W and S.

Comments for this post are closed

It switches but isn't the last citation more prestigious? Suggests you ran the lab. Or is that not a thing in economics?

Comments for this post are closed

Comments for this post are closed

BDK: "Well that’s the tough part–naming a valuable endeavor that will not be dominated by machines in 70 years. Or 35 years or 100 years. The debate of how long it will be is now not even an order of magnitude."

Yes, it's amazing to me how little thought is given to the implications of generalized AI that is superior to humans in every objective way. Compare the time spent thinking about that versus thinking about global warming, for example. I'll bet the total number of person-hours spent globally thinking about global warming outnumbers the hours spent thinking about the implications of general AI that is superior to humans by at least a factor of 100. But put a person from the year 2000 in the year 2100, and the effects of generalized AI will be much more apparent than the effects of global warming.

But do these type of games give you any information about general AI development? They have rigid and easily determined rules and winning conditions. Life is different pretty fundamentally, isn't it?

No.

To the first question, or the second?

Comments for this post are closed

Comments for this post are closed

"But do these type of games give you any information about general AI development?"

What has really impressed me was that when the Go program beat the European champ (in January!) it seems like the general consensus was that it wasn't ready to be #1 in the world. Now, it seems like the general consensus is that it might very well be the best in the world.

But I was really impressed with Watson beating Ken Jennings and Brad Rutter, in terms of seeing how far computers are coming in similarity to human capabilities.

But even more than the software, what impresses me is the hardware progression. I plotted the results of a paper by Martin Hilbert and Priscila Lopez, and projected the rates of improvement a few decades into the future.</a

Based on a human brain being about 20 quadrillion instructions per second, all the computers in the world were only equivalent to 1 human brain in 1993. But in 2016, all the computers are equivalent to about 2 million human brains. And by 2040, the number of human brain equivalents will be about 10 *quadrillion*.

With that sort of hardware power, it's hard for me to believe computers won't be able to do almost all of the jobs humans can do now.

D-oh! Sorry about that link. A computer would never do that...watch the Tech/Florida State game on TV, and miss the close on the hyperlink.

Comments for this post are closed

How good are computers at, say, writing an essay? That doesn't require more than one human brain but will trillions of human brains of power help the computer to do that? That doesn't seem like something that would be impossible for AI to achieve by any means but is that something that it can currently do? I think writing a coherent essay is a lot easier than playing Go at a top professional level, but...

I bet you're in the top 5% in terms of generating quality essays. Machines might still be closer to 50th percentile but it won't be long. http://mobile.nytimes.com/2015/03/08/opinion/sunday/if-an-algorithm-wrote-this-how-would-you-even-know.html?referer=&_r=0

Comments for this post are closed

Comments for this post are closed

Comments for this post are closed

Comments for this post are closed

The comparison of Global Warming and AI is astute.

AI is almost a certainty, and within a somewhat short order. Global warming has an ill defined cost function and any large effects are very far off.

Worrying about or thinking through the implications of generalized AI is pretty pointless. We really have no way to predict what will happen. Live your life now to the fullest! That's about all we can do.

I disagree very strongly. Managing AI is a programming problem.

Comments for this post are closed

Comments for this post are closed

What AI cannot do is set goals that are important to humans. xkcd has an interesting comic portrayal on the reasons for Asimov's ordering of the "laws of robotics" which may be relevant to concerns about AI: http://xkcd.com/1613/

Comments for this post are closed

Computers will probably not help us think about global warming but they could help us model it. The current computer models used to study global warming are cruddy but that doesn't imply that it is not a problem amenable to computer modeling. The mechanisms involved are well understood but the system is incredibly complex with a lot of positive and negative feedbacks and current models just don't have the adequate complexity to deal with them. But it is not infinitely complex. It is something you can throw money+CPU time+modeling expertise at.

Nonsense. 20 years ago there was an imperfect model which failed to precisely predict values in a highly volatile series, therefore all future modelling efforts are necessarily worthless. Throwing good money after bad .. /sarc

Comments for this post are closed

"Computers will probably not help us think about global warming but they could help us model it."

My point was that it's silly to do any thinking or modeling of global warming, since global warming's impacts will be trivial compared to the impacts of generalized artificial intelligence.

Comments for this post are closed

Comments for this post are closed

Comments for this post are closed

Women are more likely to escape North Korea because they can always marry some Chinese guy.

The Chinese guy does not control access to the border. The North Korean border guards do. He can bribe them if he has the cash and can speak Korean.

Or she can have sex with them.

The question is really whether North Korea has become so corrupt people pay good money to become border guards. I would guess they do.

More likely it would involve some combination of nepotism and signalling purity with regards to state sanctioned dogmas than monetary corruption, I think. While I don't doubt that it might happen on occasion, presumably it's rather hard to extract sex as a bribe at a border crossing, since it would involve collusion with other guards.

Comments for this post are closed

Comments for this post are closed

Comments for this post are closed

As a 53 year-old man, I think it would be an absolute hoot to be the first lady of Peru all of a sudden! Where do I sign up?

Comments for this post are closed

Giving money to someone living on $1,000 per year in the developing world will do far more to improve their lives than giving $1,000 to someone earning $25,000. The correlations above suggest that it’s about 25 times more valuable. If you want to help people, this is a major reason to focus on international poverty rather than helping the relatively poor in richer countries.

Someone need to tell that to Bernie Sanders.

Comments for this post are closed

Comments for this post are closed