Tuesday assorted links

Comments

We've crashed the Scott Aronson website, it looks like.

Back up. :)

Respond

Add Comment

"Snowden: But I didn't come forward to be safe. If I wanted to be safe, I'd still be sitting in Hawaii, making a hell of a lot money, to spy on you."

God bless Snowden, defender of liberty and freedom and slayer of tyranny

Respond

Add Comment

Respond

Add Comment

5. I think the politics of this will come to the fore in time. Whether you are coming at it from the J. D. Vance angle, or the Bernie Sanders angle, this reinforces that we are increasingly a casino society.

Softbank is the one that got fleeced, and they're a Japanese company.

SoftBank is itself a winner in the casino, with Yahoo and Alibaba as past jackpots.

I am contrasting that to this:

Chris Arnade calls them “back-row kids.” They are the ones left behind by the global, post-industrial economy. They live in places like Flint, Michigan and Youngstown, Ohio. Some of them are the same “hillbillies” V.D. Vance described in his bestseller, Hillbilly Elegy. Far too often, these are the people struggling to find gainful employment and to form a sense of community in the post-industrial Midwest, the region sadly deemed the Rust Belt. Arnade, a former Wall Street trader, has spent the last three years taking pictures and talking to people in economically depressed regions.

I guess they just needed to create a no-profit unicorn, and hit up SoftBank, then they'd have enough for everyone.

Billions.

But they did not win. They lost billions and billions of their millionaire clients’ money.

This is the exact opposite, a few hundred millionaires lost big on a ill conceived investment. As soon as outside potential investors made their skepticism known the valuation was brought closer to reality.

What am I missing here.

Adam Neumann wins the big jackpot, as 4,000 workers lose their jobs.

Wait, your point is that Softbank made a huge mistake and ... what.

Your point is hedge funds should be legally prevented from making bad investments. Assumably because we need more laws to protect millionaires from losing money!! The ultimate pro inequality party stance. More laws to protect billionaires!

Your point is that Japanese or American billionaires should not be able to place a certain percentage of their assets in high beta high reward investments. It’s unfair because rich people should be protected from bad investments. We need to protect our landed gentry. What!!!

Your point is that no company should be able to receive outside funding. It should stand on its own from day one. Any firm that needs outside funding should be shut down. What!

Your point is that companies should not be allowed to receive funding outside of the government, which should receive ownership.

I’m lost. You speak against but not for. What are you for?

I am for skepticism about financial markets, and higher taxes on the lucky few casino winners.

Skepticism about what? That financial markets exist?

You realize in many countries actual casino winners are not taxed at all? Not sure why you think a Japanese bank investing in risky Chinese start-ups and making a return should be taxed more (by who?). You want to discourage new business formation and entrepreneurship in favor of wasteful consumption, or what?

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Never trust a rich man with an ugly wife. If he’s not fucking her, he’s fucking you.

Respond

Add Comment

Respond

Add Comment

SoftBank is itself a winner in the casino, with Yahoo and Alibaba as past jackpots.

My point is I'm not sure this is an American thing. Alibaba is a Chinese company, Softbank is Japanese. Ya see what I mean here? Also, the thing about a casino is that it's just a bunch of dumb games of chance. Alibaba is a real value-generating enterprise. You have your head firmly jammed up your ass once again.

How much do they pay their workers?

I should note that I'm not a raging, eat the rich, cancel billionaires, socialist.

I just don't think billionaires are saints, and I say tax them more. Use the money to improve Rust Belt schools.

I didn't think billionaires are more deserving than those schools.

Billionaire Betsy DeVos: Deep cuts to Special Olympics, student programs warranted

Billionaire Betsy DeVos requests massive education spending cuts for 2020

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Not sure I agree with this take. Softbank’s millionaire hedge fund clients made a bad investment. As with Uber and Lyft, a company whose private rosy valuations never matched with a realistic assessment by discounted cash flow analysis got burned when it tried to launch an IPO. Burning investor cash faster than it could hope to turn into income.

The financial markets worked about as well as could be expected given the efficient market hypothesis. I do not see what this has to do with Hillbilly Elegy or Sanders. Some Asian millionaires lost money. No retail investor lost as much as a dollar (or yen as the case may be.)

What are you reading into this that I might be missing?

Should the efficient market hypothesis extend to private exchanges? If so, and I think it should, this is more a disproof. This is more "shit happens" than "magic of the market." See also "scooters."

Now, on Vance and etc, we have to move on to the social promise in the Church of EMH - that we will have efficient allocation of capital. That's the reward which raises all boats.

And not just a casino increasingly paying off billionaire winners, as losers face an Angus Deaton style dystopia.

See also "the Sacklers."

There’s some emotional language and partisan framing in this response which I’ll address first, because I’m questioning whether this is a troll. If not I apologize in advance.

The Efficient Market Hypothesis is not a religion, and this is a terrible example if one were so inclined to challenge what we refer to as the strong version of the EMH. Which I do not agree with with!! Strong or weak, however, this is not the droid you are looking for :). You then speak to class issues and what I would call a gesture at unfairness and inequality, which involves ideology and politics that are entirely irrelevant to WeWork and I would prefer to leave that for you and others.

The EMH never claims that every investment by any hedge fund is a sound one. That would be by definition a preposterous claim. That’s akin to saying capitalism is proven false if any business ever goes under! It’s no claim at all and reveals a fundamental misunderstanding of the tenets of the EMH and basic economic theory.

The question to pose, sir, is did the financial market, as the wisdom of a collection of financially interested parties, correct the pseudo valuation of an overvalued firm over time. The answer is yes. In this case immediately.

The markets worked beautifully. No retail investor was ever hoodwinked into anything. Millionaires lost some cash and investors balked at the IPO because investors trust DCF over vague promises. Looks like the markets were so sound some millionaires lost money on a terrible call. EMH at work!!!

Cheers

I will stick to my claim that a foolish belief in stronger forms of the EMH, and therefore a disbelief in luck or chaos, is bound to a belief that winners are deserving. And not coincidentally, losers are undeserving .

In the worst cases it devolves to a kind of Church of EMH, with great men as saints, and great men are anybody with a billion (except Soros!).

But as Skeptic pointed out, the WeWork example has absolutely nothing to do with the 'strong' version of the EMH. And very little to do with the EMH at all.

Your non sequitur point I agree with somewhat.

Mine? I was answering an EMH claim.

Read carefully.

You made up a straw man version of the EMH to make a stupid partisan point. That’s not clever, it’s just dumb. Here are things I know for sure:

You don’t have an Econ degree
You went to a garbage school
You never worked in finance
You never took financial economics
You can’t explain the EMH
You can’t explain Beta
You can’t explain alpha
You don’t understand how hedge funds work

And no, SoftBank, a Japanese hedge fund losing money does not mean anything about Sanders or hillbillies. What next, Uber is overvalued so Lizardmen must be running things behind the scenes!

But I can hold it together.

The claim was "The financial markets worked about as well as could be expected given the efficient market hypothesis."

A billion in wins for one man, and 4000 others laid off .. should we really take that as the best we can expect from our economic system? From "efficient" markets?

That commenter is an ass.

Regardless, you are creating a straw man of epic proportions. In your view of the EMH, let’s call it AEMH (anons EMH) there can be no risk element, because that would prohibit instances whereby a fund or hedge fund invests in a firm that fails to produce a return. (Which is 90%+ of start up firms).

Sir your view of the EMH, the AEMH is not only contradictory to economic theory and econometric evidence, it’s self contradictory on the merits.

In an AEMH world there is no beta and no alpha. In an AEMH world, economic theory demands that the only return be based on risk free assets, so CAPM is gone, as well as the Fama 3 Factor. Any risk is a violation of the AEMH.

The only return would be based on the Lagrange solution of intertemporal utility maximization, since in your AEMH view all investments are risk free investments.

You are emotionalizing. You are so far down the mood affiliation hole you don’t realize millionaires making stupid investments decreases inequality.

You should applaud this, inequality is rapidly declining. $10 billion in rich people assets was thrown down the drain.

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

SoftBank is a public company.

Respond

Add Comment

Respond

Add Comment

#4..."Have you ever heard someone argue for minimum wage hikes and inelastic labor demand, yet claim that immigrants do not lower wages? Well, the latter claim about immigration implies elastic labor demand. "

This is essentially the point I've been arguing that, if you favor a guaranteed job, immigration is not helpful. I prefer no guaranteed job and immigration, but that's just me.

Respond

Add Comment

6. Quantum bullshit most likely from Google. At least that's their peer's view: https://www.nytimes.com/2019/10/23/technology/quantum-computing-google.html Everything coming out of so-called tech should be viewed with great skepticism. I accept nothing at face value.

Respond

Add Comment

4. Fortunately, real empiricism cuts through the propaganda: https://www.amazon.com/Unbound-Inequality-Constricts-Economy-about/dp/0674919319

Respond

Add Comment

3. Machine learning Beatles lyrics I am not impressed and still I do not fear The Singularity.
----

The singularity is being put in place. You are right., It is not fearful.

This is a big nothing. All lyrics sound lame out of the context of the music, the vocal intonation, the overall delivery. The actual words are only a small part of what makes a song effective.

However we are probably no more than a decade away from machine learning that could create actual Beatles-esque audio recordings, not mere words on paper. And create CGI music videos to go with them. And then you can fear the Singularity.

Someone ought to be working feverishly on this. Boomers are the final generation that still has money and is still willing to pay for music. It's imperative to cash in before they all shuffle off the mortal coil.

'The actual words are only a small part of what makes a song effective.'

As can be seen in this surprisingly well done reworking of a Beatle's album. The link is to something that is includes 'With Illicit Help From Your Friends' - https://www.youtube.com/watch?v=NYD3QtyEGGM

When machine learning is able to do something like this on a regular basis, well, it would be something to at least notice.

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

#5: That's a flawed comparison between Paulson and Neumann, I would say. Paulson made a substantial bet against mortgage backed securities, and had he been totally wrong, he'd have lost big and somebody else would have their name on the engineering school at Harvard. You could characterize that as zero-sum and say that maybe that wasn't the most socially useful form of economic activity. Okay, granted.

But as the article itself notes, Softbank's down about $10 billion on this deal, currently, while Neumann himself is up $1 billion. That's $9 billion just set on fire. That's not zero sum, that's negative sum. That's not simply not socially useful, that's socially harmful.

Correct me if I am wrong here, but didn't Paulson partner up with Goldman to commit a bit of massive fraud on its clients, in order to make sure his short bet panned out.

In sports betting at least, for the time being, that's illegal.

I honestly don't know all the details. My basic point, though, is that at worst, Paulson simply managed to transfer a lot of other people's money to himself, whereas Neumann managed to get his hands on a crapload of other people's money, lose 90% of it, and then pocket the remaining 10% for himself. From society's standpoint, that's worse.

oh no doubt it make one wonder if Neuman got his hands on the Epstein archives.

My point just below though is this is only a difference in degree. It's not like golden parachutes are anything new. And there's a long line of CEOs who oversaw the cannibalization of their companies via PE debt and asset sales and got rich in the process. Heck, the CEO of Sears managed to end up owning all the real estate while the company bled to death. Neat trick.

My understanding is this... Paulson wanted to bet against MBS. He partnered with Goldman. Goldman looked for clients to take the up side of Paulson's downside bet. In essence, failing to tell the upside client that they were buying investments their initiating investor expected to implode. Goldman then used Paulson's research to protect themselves by altering their investments anticipating a blowout in MBS. Someone can correct me, since it's been a while since I read The Greatest Trade Ever: The Behind-the-Scenes Story of How John Paulson Defied Wall Street and Made Financial History by Gregory Zuckerman.

Respond

Add Comment

I'm with you 100% on golden parachutes for failed CEOs.

Respond

Add Comment

Respond

Add Comment

Well, the 90% of the money that "got lost" went to other people too. It's not real resources that got destroyed, I don't think.

Indeed. Did it ever really exist at all? Ones and zeroes.

Until you buy a yacht with it.

Respond

Add Comment

Au contraire. WeWork paid people to lease, subdivide and build out office suites and "flexible work spaces" that were worth less in the rental market than it cost to develop them. And they did this on a pretty large scale. That's the definition of wasted resources.

Sure, but it doesn't come to 90%.

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

5. There's nothing new or different about the WeWork sh*tshow, it's just more better bigger. end-stage simulacra

Respond

Add Comment

Immigrants are both a supply of labor and a source of demand for labor.

Somewhat higher minimum wages both increase the cost of labor and increase demand for labor as more money ends up in the hands of people with a higher marginal propensity to consume.

These aren't one sided changes.

Respond

Add Comment

Any follow up on the Libra dud? Damn bureaucrat's fault right?

Any thoughts on Hillary entering (strong-arming) the race and probably helping Trump win again with her educated charm?

Hillary is not entering the race, hun. That's just some bernie bros fever dream.

Respond

Add Comment

It's not SOP for either.

Respond

Add Comment

Respond

Add Comment

#1...Good interview.

Respond

Add Comment

2. The last time I asked a woman if the visitors matched the art works, she slapped me.

Did you grab her p**sy when you said it?

Respond

Add Comment

Respond

Add Comment

I too do not fear the singularity. Life and technology follow a sigmoid (S) shape; they do not climb exponential curves forever. Eventually a new constraint is found, and the positive exponential curve turns to a negative exponential curve at the inflection point. Nobody knew we were at the inflection point for population in the late 1980s until at least 5 years later -- you can't know; a sigmoid is indistinguishable from a positive exponential curve until you are at or past the inflection point.

What we label AI is only machine learning, neural networks, a 1980s idea made practical with 2019 computer power and huge data sets via the internet. Machine learning allows high dimensional non-linear interpolation of enormous data sets, which is very useful for solving some problems that involve interpretation of a lot of data, like computer vision. It is not a solution to General Artificial Intelligence, the ability to learn and deal with unstructured, poorly defined problems with contexts that are difficult to limit. There has been no progress is advancing General Artificial Intelligence in 40 years. We don't really know how to even ask the question. Note how machine learning was able to help a great deal with the observability problem of driving (how to interpret a lot of sensor data), but is choking on the controllability problem (from a partially observed context, construct a model of how the universe works, predict several seconds ahead, and take action accordingly to accomplish various conflicting goals).

Quantum computing will also allow us to solve a difficult set of problems that have proven impossible for current computers. That's nice, but there's nothing about it that suggests that it will help solve the general AI problem.

Well done, +5 internet points

Respond

Add Comment

You make valid points. And yet...

And yet just in the past few years, machine learning has blown past a number of yes-buts. As in "OK, so it can do X now, but what about [much harder problem] Y".

Go is way harder to solve than chess, and just a few years ago we had no idea how to tackle it. We were told that it might be decades before computers could beat the best humans. And suddenly they could.

And then the next frontier was games with incomplete information like Starcraft II and poker. But as of a few months ago the beat-the-professionals milestones have been hit and the rest is just details.

Starcraft II is also real-time rather than turn-taking, and has a large "action space". It shares those characteristics with the autonomous driving problem.

In all these cases, researchers sprang a new solution on a world that had no idea it was even possible, let alone imminent. If general AI is possible, it will very likely unfold in the same way. It's foolish to bet against it happening in your lifetime, however old you may be.

Has there been a new update on Starcraft? The original AI was not able to compete on a fair basis- it had to have a supernatural click-rate enabled by the direct machine interface, full view of the entire map rather than the human's limited view.

Yes, AlphaStar's wins earlier this year against professional players were under certain favorable conditions: playing only one of the game's races against other players playing the same race, and interacting directly with the raw interface of the game engine rather than relying on the "camera" view.

But apparently as of July 2019 it is training by playing incognito in regular mode against human players in the public multiplayer ladder. DeepMind claims that the version using the camera interface is almost as strong as the version using the raw interface.

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Is there anything more dull than trying to mathematically model Beatles lyrics? You'll never get an "understanding" of them that way -- a grasp of the tacit knowledge that goes into writing good lyrics. If you'd like to write lyrics like the Beatles you should listen to a lot of the Beatles, and a lot of their influences (Chuck Berry, Little Richard, etc.), and then just start writing. I guess its a neat trick to use machine learning to construct Beatles-like songs, and i understand the authors aren't trying to get rich off the songs their model creates, but i just think its kinda dumb belies the point that AI is still far, far, far off from humanity. I think people who understand tacit knowledge understand this point.

Anyway, I didn't really read the article. Just kinda skimmed it. Thank you for reading my MRCommentTalk

Respond

Add Comment

#3: Yeah, the article took a rather strange approach: instead of explaining neural networks, deep learning, etc. it treats natural language processing as a computing problem and even offers examples of code as if this is something for hobbyists to code in their homes. There's plenty of machine learning software available to the public, but to actually create one's own machine learning software? Telling the reader about bigrams and trigrams doesn't exactly equip them to understand what's going on with GPT-2.

And yeah, I'm not impressed by the article-Beatles lyrics either.

A slightly better but still not very good article about GPT-2 is in this week's New Yorker:
https://www.newyorker.com/magazine/2019/10/14/can-a-machine-learn-to-write-for-the-new-yorker

It does point out the interesting aspects of Open AI, a non-profit firm producing open-source AI software. But the reporter's emotional reactions are over-blown. Both articles should've done a better job of describing what GPT-2 can and cannot do. There's rapid progress being made here, but we need better reporting than these articles provide.

Respond

Add Comment

#2: very cool and amusing. He probably could've taken a photo of someone wearing a van Gogh t-shirt standing in front of that same van Gogh, but I like that he took a more challenging approach by finding non-literal or non-exact matches.

Respond

Add Comment

#4. "If you don't like minimum wage hikes, you can't believe that immigrants are pushing down wages. If you like higher minimum wages, you must believe that immigrants are terrible for labor markets. This proposition should be equally uncomfortable on the left and the right."

Say what? No, you can disagree with things for different reasons. Minimum wage pushes a bad wage subsidy tax onto businesses and results in fewer workers making more money. Low skilled immigration increases the labor supply and puts downward pressure on low skilled wages. Furthermore low skilled immigrants pay less than the median tax rate and thus shift the tax burden to everyone else.

Respond

Add Comment

This Beatles example is one of the least interesting and impressive of all GPT-2 applications I've ever seen - it uses the least powerful GPT-2 and a tiny corpus incorrectly formatted. Why is this being singled out on MR as 'why the Singularity is not near' as opposed to any of the full-strength GPT-2 samples OA released half a year ago, or any of the good hobbyist examples since, or the long and good New Yorker article, or the CTRL or Megatron samples? What's this strawman doing here?

Respond

Add Comment

I finally just clicked on 5, but the headline alone inclines me to view it very favorably.

Respond

Add Comment

Respond

Add Comment