Friday, June 5, 2020

In which a film critic reviews his own early work and realizes he'd been a huge asshole.

As previously mentioned, I'm an admirer of film critic Bob Chipman, in part because of his willingness to re-examine his own work and, when appropriate, apologize.



I was going to troll Andrew Gelman with this human Spirograph clip, but we've had to reschedule this post so often I doubt anyone remembers the original comment.



And finally a bit of Wes Montgomery to kick off the weekend.


Posted by Mark at 6:30 AM 0 comments Links to this post

Thursday, June 4, 2020

And the view from a year ago.


Tuesday, February 5, 2019

Rational actors, stag hunts and the GOP

We have hit this idea in passing a few times in the past (particularly when discussing the Ponzi threshold), but I don't believe we've ever done a post on it. While there's nothing especially radical about the idea (it shows up in discussions of risk fairly frequently), it is different enough to require a conscious shift in thinking and, under certain circumstances, it can have radically different implications.

Most of the time, we tend to think of rational behavior in terms of optimizing expected values, but it is sometimes useful to think in terms of maximizing the probability of being above or below a certain threshold. Consider the somewhat overly dramatic example of a man told that he will be killed by a loan shark if he doesn't have $5000 by the end of the day. In this case, putting all of his money on a long shot at the track might well be his most rational option.

You can almost certainly think of less extreme cases where you have used the same approach, trying to figure out the best way to ensure you had at least a certain amount of money in your checking account or had set aside enough for a mortgage payment.

Often, these two ways of thinking about rational behavior are interchangeable, but not always. Our degenerate gambler is one example, and I've previously argued that overvalued companies like Uber or Netflix are another, the one I've been thinking about a lot recently is the Republican Party and its relationship with Trump.

Without going into too much detail (these are subjects for future posts), one of the three or four major components of the conservative movement's strategy was a social engineering experiment designed to create a loyal and highly motivated base. The initiative worked fairly well for a while, but with the rise of the tea party and then the Trump wing, the leaders of the movement lost control of the faction they had created. (Have we done a post positing the innate instability of the Straussian model and other systems based on disinformation? I've lost track.)

In 2016, the Republican Party had put itself in the strange position of having what should have been their most reliable core voters fanatically loyal to someone completely indifferent to the interests of the party, someone who was capable of and temperamentally inclined to bringing the whole damn building down it forced out. Since then, I would argue that the best way of understanding the choices of those Republicans not deep in the cult of personality is to think of them optimizing against a shifting threshold.

Trump's 2016 victory was only possible because a number of things lined up exactly right, many of which were dependent on the complacency of Democratic voters, the press, and the political establishment. Repeating this victory in 2020 without the advantage of surprise would require Trump to have exceeded expectations and started to win over non-supporters. Even early in 2017, this seemed unlikely, so most establishment Republicans started optimizing for a soft landing, hoping to hold the house in 2018 while minimizing the damage from 2020. They did everything they could to delay investigations into Trump scandals, attempted to surround him with "grown-ups," and presented a unified front while taking advantage of what was likely to be there last time at the trough for a while.

Even shortly before the midterms, it became apparent that a soft landing was unlikely and the threshold shifted to hard landing. The idea of expanding on the Trump base was largely abandoned as were any attempts to restrain the president. The objective now was to maintain enough of a foundation to rebuild up on after things collapsed.

With recent events, particularly the shutdown, the threshold shifted again to party viability. Arguably the primary stated objective of the conservative movement has always been finding a way to maintain control in a democracy while promoting unpopular positions. This inevitably results in running on thinner and thinner margins. The current configuration of the movement has to make every vote count. This gives any significant faction of the base the power to cost the party any or all elections for the foreseeable future.

It is not at all clear how the GOP would fill the hole left by a defection of the anti-immigrant wing or of those voters who are personally committed to Trump regardless of policy. Having these two groups suddenly and unexpectedly at odds with each other (they had long appeared inseparable) is tremendously worrisome for Republicans, but even a unified base can't compensate for sufficiently unpopular policies. Another shutdown or the declaration of a state of emergency both appear to have the potential to damage the party's prospects not just in 2020 but in the following midterms and perhaps even 2024.

So far, the changes in optimal strategy associated with the shifting thresholds have been fairly subtle, but if the threshold drops below party viability, things get very different very quickly. We could and probably should frame this in terms of stag hunts and Nash equilibria but you don't need to know anything about game theory to understand that when a substantial number of people in and around the Republican Party establishment stop acting under the assumption that there will continue to be a Republican Party, then almost every other assumption we make about the way the party functions goes out the window.

Just to be clear, I'm not making predictions about what the chaos will look like; I'm saying you can't make predictions about it. A year from now we are likely to be in completely uncharted water and any pundit or analyst who makes confident data-based pronouncements about what will or won't happen is likely to lose a great deal of credibility.
Posted by Mark at 3:20 AM 0 comments Links to this post

Reviewing our take on the state of the GOP seven years ago.


Not especially prescient but not much I wouldn't still stand by (though I wish I would have considered the possibility -- and implications -- of a wealthy, well-known candidate appealing directly to the base).

Tuesday, August 6, 2013

Metablogging -- stag hunts, misalignment, principal agents and all that jazz

Shortly after Josh Marshall posted this analysis of recent events in the GOP, Joseph called me up to compare reactions. We've been having this conversation for so many years that much of it has devolved into a self-referential shorthand. As an illustration, at one point, after a fairly long-winded comment by me, Joseph simply said "Stag hunt." and tossed the ball back to me and we moved on to the next topic.

We agreed (with, I assume, fingers crossed on both ends of the line) that we'd write some posts on the subject, but I'm starting to think that it might be more useful to step back for a minute and talk about how we've been framing the question of what's going on in the Republican Party (politically, not socially or in terms of policy. Those are entirely different metaposts).

For years now, the two of us have been talking about the post-Tea Party GOP in terms of a multi-player stag hunt. Over the past few years the stakes (particularly the costs of failure) have increased. At the same time, participation rates required to take down the stag have also increased. As a result, progressively smaller groups have gained the power to kill the enterprise. (In a different conversation Joseph pointed out that, in a military context, shooting deserters is also a predictable result of this situation.) We could dig deeper into examples and implications (particularly with respect to the trade off between the power of an alliance vs. its stability) but for now I want to limit the discussion to framing.

(there might also be a place here to talk about symmetry breaking, but I'd need to give that some thought first.)

Another way of looking at the story we've found useful is to look at misalignment of interests, especially what looks to us two non-economists as a particularly nasty two-level principal agent problem where a small group of big donors determine the pool of viable candidates and a relatively small but coherent subgroup of the primary voters make the purchasing decisions for the entire party. You'll notice that, like the stag hunt frame, under this scenario small groups can acquire disproportionate power.

And of course there's the mandatory Influence reference, framing the story in terms of social psychology. If you check out the chapters on commitment and consistency, social proof, and scarcity you'll find all sorts of applicable discussions of the ways groups united by a common belief system deal with ideological challenges and the loss of dominance.

Nothing particularly fresh or profound here, but these idea have proven a pretty good framework recently. I'm not saying they should be the basis of the standard narrative -- I'm not sure there should be a standard narrative -- but they do come in handy. More importantly, I think you can make the case that too little of the public discourse is spent examining underlying assumptions and asking about the different ways to frame our questions.
Posted by Mark at 3:17 AM 0 comments Links to this post

Wednesday, June 3, 2020

Deferred Tuesday Tweets








Lifestyles of the Rich(?) and Famous.

 I'm not going to weigh in on the scientific debate, but the politics of hydroxychloroquine is worthy of study.


 
Flack-to-Hack ratio at work.

 
Mandatory hyperloop reference.

 
Common sense from Krugman.

 
At least the Onion's on a roll.

 
Zuckerberg is a horrible person...


 
... and yet still better than....


 
One of the stories we're following closely.

 

Amazing how a company this well run could have lost over eight billion dollars in 2019.

 

Brought to you by the good people of MOD.

 
Time to revisit our dysfunction thread. Excellent by Fallows.


 The revolution will be tweeted



 



 


 



 





 




 




 


Posted by Mark at 6:30 AM 0 comments Links to this post

Tuesday, June 2, 2020

LA is still not burning


I have great respect for the resilience of NYC journalists. If the New York Times were consumed in a great conflagration, I have  no doubt that the survivors would crawl from the wreckage, mix their blood with ashes for ink, and scrawl out on op-ed on how bad things are in Los Angeles.

Here are some tweets I posted Sunday.

[Should have been more clear, but the next part refers mainly to the national situation.]
Later that evening things got more intense, but not in a way that disproved my point.

Here's where the old city/county question reappears. When people talk about NYC , they are referring to a city comprised of various counties. When people talk about LA, they are usually referring to  a county comprised of different cities. The city/county distinction is important because, while the city is very big, the county is huge, with over 10,000,000 residents and land area of over 4,000 square miles.

In this case, almost all of the footage of riots and looting came from the cities of Long Beach and especially Santa Monica. This is clearly a county level story. Therefore it is essential to remember that the scenes that dominated the news were largely limited to two neighborhoods in a place with considerably more residents than NYC.

This is not to downplay the anger here or its causes or the damage it can do to innocent bystanders when it gets out of control, but when the media focuses all its attention on a tiny sliver of a town, the rest of the country gets a false impression, particularly when some people use the misinformation as disinformation.



Posted by Mark at 6:30 AM 0 comments Links to this post

Monday, June 1, 2020

A few points to consider when reading about Tesla (2020 edition)

1. [As previously mentioned] Tesla's valuation has always depended more on the story than the fundamentals.

Finally, it is essential to remember that maintaining this “real-life Tony Stark” persona is tremendously valuable to Musk. In addition to the ego gratification (and we have every reason to believe that Musk has a huge ego), this persona is worth hundreds of millions of dollars to Musk. More than any other factor, Musk’s mystique and his ability to generate hype have pumped the valuation of Tesla to its current stratospheric levels. Bloomberg put his total compensation from Tesla at just under $100 million a year. When Musk gets tons of coverage for claiming he's about to develop telepathy chips for your brain or build a giant subterranean slot car race track under Los Angeles, he keeps that mystique going. Eventually groundless proposals and questionable-to-false boasts will wear away at his reputation, but unless the vast majority of journalists become less credulous and more professional in the very near future, that damage won’t come soon enough to prevent Musk from earning another billion dollars or so from the hype.

2. Since I wrote that, Musk made a big bet on the stock price and it has worked out very well for him.
Tesla confirmed that CEO Elon Musk earned the first tranche of his massive incentive payout, in a document filed with the Securities and Exchange Commission on Thursday.
The tranche is comprised of about 1.7 million shares of Tesla, and would be valued around $775 million based on Thursday’s closing market value. Shares in Tesla closed at $805.81 on Thursday, and the options have a strike price of $350.02.

Thursday’s filing, which also set a date of July 7 for the company’s annual shareholders meeting, said: “As of the date of this proxy statement, one of the 12 tranches under this award has vested and become exercisable, subject to Mr. Musk’s payment of the exercise price of $350.02 per share and the minimum five-year holding period generally applicable to any shares he acquires upon exercise.” It is not clear if Musk has yet exercised the options.

3. With this in mind, the recent surge looks a bit convenient.

As you can see from the 5-year chart below, Tesla's stock didn't do much for quite a while then, about a year ago, it more than quadrupled in about three quarters.




It's useful to compare this to the stock performance of the far larger and more profitable GM.


Particularly these numbers:
vs. Tesla's

4. These incentive plans can incentivize some bad behavior.

Jack Welch demonstrated how easy it was to nudge the books to create the profitability picture Wall Street likes. Running up a stock's value isn't that dissimilar, and there are lots of questions about Tesla's books. Add to that the power of the company's narrative, the cult of personality around Musk, and investors' willingness to reward even the distant promise of viability from any business that can label itself a tech company (we live in the WeWork age).

When Musk compares covid-19 to a cold, pushes miracle cures or lobbies hard for a quick end to the lockdown, it's important to remember...

5. What's good for Tesla isn't necessarily good for America.

Posted by Mark at 5:31 AM 0 comments Links to this post

Friday, May 29, 2020

University as a business

This is Joseph

François Furstenberg has comments on cuts at Johns Hopkins, including the high pay of leadership. The whole thing is worth reading but I wanted to highlight a couple of things:
Then there is the issue of deferred compensation for top executives. According to the university’s latest audit, total liabilities related to deferred compensation amounted to over $130 million — or $30 million more than the institution will save by suspending contributions to its thousands of employee retirement accounts this year.
While a handful of top administrators will take a modest pay cut this year, the university has not said whether any of its executives will forfeit the sums accumulated in their deferred-compensation plans. I assume they won’t. There is a searing irony in the fact that these well-paid officers may keep their lucrative deferred-compensation packages even as staff and faculty sacrifice the value in their retirement funds — which are deferred compensation on a far more modest level. Altogether, these practices do not paint a portrait of an institution with robust mechanisms of oversight and accountability.
And the speed of the university being in crisis:
How does a university with a $6-billion endowment and $10 billion in assets suddenly find itself in a solvency crisis? How is one of the country’s top research universities reduced, just a month after moving classes online, to freezing its employees’ retirement accounts? 
This brings up, in my view, a deeply philosophical question about the role of Universities. If they are just tax advantaged businesses then the deep subsidization of these institutions needs to be carefully considered. If they are enduring institutions intended to last for the long term then isn't the point of the endowment to weather storms?

This fits into another post I read lately from Matt Reed:
Adjunct classes cost less than they take in; in business terms, they’re profit centers. The highest-paid faculty cost more than they generate; in business terms, they’re loss centers. By state administrative code, we have to eliminate the profit centers to protect the loss centers. In business terms, that’s backward. But those are the rules. 
This is an insane framing, but I think that was his point.  Obviously the tenured faculty also contribute to reputation and mission. But the point is that a business engaging in cuts would have very different rules and the consequences would be quite different. Imagine a University closing in year 3 or 4 (!!) of a professional program. The cost to students would be . . . severe. Similarly, reputation and integrity are why you can't be a "premium" student with an automatic high grade.

This is a challenging time for everyone and it is a good time to reflect on the social role that we ascribe to universities. This does not mean that I want to disrupt it -- disruption is a great plan for things that can fail or fall apart without huge consequences. Is anybody seriously considering disrupting the police or military?

But there is a real question about the priorities of our academic environment and how we move forward in a word where we start being asked to be more like a business. To me, the real question is why? Just to enrich people with an MBA?
Posted by Joseph at 8:30 AM 0 comments Links to this post

Thursday, May 28, 2020

Q-ing theory

For a long time now, we’ve been talking about conservative movement disinformation, and for almost as long (since 2015 at least) we’ve been suggesting that the movement has lost control of the process and while the disinformation is still being cranked out, it no longer serves its intended function.

2020 has provided an abundance of examples.
 


 

 


These last two represent a new development, Republicans objecting to right-wing conspiracy theories, but before anyone chimes in with a chorus of "good people on both sides," a little history would be useful.

I was living in Arkansas in the 90s and witnessed up close the birth of the modern style of weaponized batshit crazy conservative conspiracy theories. It was the beginning of what Charles Pierce calls happy hour at the Mena Airport cocktail lounge. Bill was a murderous drug dealer who raped his lesbian wife who, in turn, was also having an affair with Vince Foster before she had him killed.

(If you think I am exaggerating in anyway, read up on the subject. If anything, I’m understating the case)

In order to be truly effective, this approach required the disinformation to be carefully channeled and for the mainstream media to largely turn a blind eye. The lies you could tell on Rush Limbaugh‘s show got a dog whistle on Fox. The lies you could tell on Fox got a dog whistle on Meet the Press. The lies you could tell on Meet the Press is a topic for another post.

Not all journalists were AWOL. There was a great deal of outstanding reporting on the subject by people like Josh Marshall, Jonathan cheat, Jane Mayer, Ornstein  & Mann and, of course, Charles Pierce but these stories  though well reported,never really gained traction and the sources of the disinformation paid no real penalty.

The pattern would continue for a couple of decades, from the invented quotes of Al gore to the swiftboating of John Kerry to the birtherism and other absurd “scandals” of the Obama era all the way through the coverage of Hillary in 2016. Throughout all of this, you would have to look long and hard to find a Republican in good standing who objected with more than lip-service and crocodile tears.

But sometime in the Obama administration, the system started to show cracks. Theories and memes that had been relegated to dark corners and right wing radio call-in shows were suddenly showing up in quotes from Republican congressman and Senators. To those who had not been following the story closely, it appeared that there had been a huge uptick in craziness among conservatives. For those who had been keeping up, it was not at all clear that the craziness increased at all. Instead we were seeing a breakdown in the ability to control the flow of this craziness.

What had been a vital element in the party’s success for the past quarter century has now become a serious threat. By purest coincidence, Republican officials and conservative outlets decided to take a principled stand shortly after these conspiracy theories started to bite the party in the ass.
Posted by Mark at 6:30 AM 0 comments Links to this post

Wednesday, May 27, 2020

Believe it or not, this is actually an improvement over the NYT's 2016 coverage.




Tuesday, May 3, 2016

Context only counts if it shows up in the first two dozen paragraphs

The New York Times has a good piece on the impact of voter ID laws but I do have a problem with a few parts (or at least with the way they're arranged).

Stricter Rules for Voter IDs Reshape Races

By MICHAEL WINES and MANNY FERNANDEZ MAY 1, 2016

SAN ANTONIO — In a state where everything is big, the 23rd Congressional District that hugs the border with Mexico is a monster: eight and a half hours by car across a stretch of land bigger than any state east of the Mississippi. In 2014, Representative Pete Gallego logged more than 70,000 miles there in his white Chevy Tahoe, campaigning for re-election to the House — and lost by a bare 2,422 votes.

So in his bid this year to retake the seat, Mr. Gallego, a Democrat, has made a crucial adjustment to his strategy. “We’re asking people if they have a driver’s license,” he said. “We’re having those basic conversations about IDs at the front end, right at our first meeting with voters.”

Since their inception a decade ago, voter identification laws have been the focus of fierce political and social debate. Proponents, largely Republican, argue that the regulations are essential tools to combat election fraud, while critics contend that they are mainly intended to suppress turnout of Democratic-leaning constituencies like minorities and students.
In the third paragraph, we have two conflicting claims that go to the foundation of the whole debate. If election fraud is a significant problem, you can make a case for voter ID laws. If not, it's difficult to see this as anything other than voter suppression. This paragraph pretty much demands some additional information to help the reader weigh the claims and the article provides it...

More than twenty paragraphs later.

Mr. Abbott, perhaps the law’s most ardent backer, has said that voter fraud “abounds” in Texas. A review of some 120 fraud charges in Texas between 2000 and 2015, about eight cases a year, turned up instances of buying votes and setting up fake residences to vote. Critics of the law note that no more than three or four infractions would have been prevented by the voter ID law.

Nationally, fraud that could be stopped by IDs is almost nonexistent, said Lorraine C. Minnite, author of the 2010 book “The Myth of Voter Fraud.” To sway an election, she said, it would require persuading perhaps thousands of people to commit felonies by misrepresenting themselves — and do it undetected.

“It’s ludicrous,” she said. “It’s not an effective way to try to corrupt an election.”

I shouldn't have to say this but, if a story contains claims that the reporter has reason to believe are false or misleading, he or she has an obligation to address the issue promptly. Putting the relevant information above the fold is likely to anger the people who made the false statements, but doing anything else is a disservice to the readers.
Posted by Mark at 6:30 AM 0 comments Links to this post

Tuesday, May 26, 2020

Pharmacoepidemiology in debate

This is Joseph

Andrew Gelman was discussing the Lancet article on the safety of hydroxychloroquine. He posited three possible outcomes:

1. The criticisms are mistaken. Actually the research in question adjusted just fine for pre-treatment covariates, and the apparent data anomalies are just misunderstandings. Or maybe there are some minor errors requiring minor corrections.
2. The criticisms are valid and the authors and journal publicly acknowledge their mistakes. I doubt this will happen. Retractions and corrections are rare. Even the most extreme cases are difficult to retract or correct. Consider the most notorious Lancet paper of all, the vaccines paper by Andrew Wakefield, which appeared in 1998, and was finally retracted . . . in 2010. If the worst paper ever took 12 years to be retracted, what can we expect for just run-of-the-mill bad papers?
3. The criticisms are valid, the authors dodge and do not fully grapple with the criticism, and the journal stays clear of the fray, content to rack up the citations and the publicity.
I wrote a comment that went into what I see as a fourth possibility: the data is real and the analysis was done as stated but the question, itself, is not fit for observational analysis
I think there is a fourth possibility. We know that studies of drugs for their primary indication are fraught with risk. This was clearly articulated in the literature by 1983 (https://onlinelibrary.wiley.com/doi/abs/10.1002/sim.4780020222) bu Olli Miettinen but was built on a long tradition of problematic cases. More recently we have Tobias Kurth’s paper showing implausible results in cases with medical desperation (https://academic.oup.com/aje/article/163/3/262/59818). Treatment is not random when physicians, patients, and families are desperate.
We also saw this with hormone replacement therapy and CVD; the theory that it would help with CVD ended up channeling the drug to high SES individuals. This is why drug people insist on trials — too many results reverse direction when channeling effects are removed.
There is no possibility that these medications were being given for an indication other than covid-19. There is just not that much co-infection with malaria.
So option #4 is the study is fine. The data (despite some alarming patterns) turns out to be okay. But the result is simply wrong because they are studying an intended effect. Or they could be correct. Sometimes you end up accounting for channeling (people do, in fact, occasionally win the lottery) and the results are consistent with the causal association posited by trials. My favorite paper was looking at tricks to try and do this with observational data.
The real issue is the strength of the conclusions. Of course, the data issues are a separate and concerning factor as well. But I wish we’d show more humility with observational drug research. I do a LOT of it and it annoys me when these basics of interpretation are glossed over with bullet text about observational research having limitations (unknown confounders — grrrrr) and not a recognition that this is why we do trials — because these estimates are inherently unreliable.
#EndofRant
I see this as different than Wakefield, who faked data. Or an analysis structure that is incorrect. And people publish these studies all of the time. But they do so with known failure points and the use of the word "associated" is a thin veneer:
We were unable to confirm a benefit of hydroxychloroquine or chloroquine, when used alone or with a macrolide, on in-hospital outcomes for COVID-19. Each of these drug regimens was associated with decreased in-hospital survival and an increased frequency of ventricular arrhythmias when used for treatment of COVID-19.
Obviously this paper is only of value for a general medical audience if we are supposed to conclude something about the effectiveness of hydroxychloroquine or chloroquine. I mean this stuff is kind of there in the limitations section, but the conclusion is written without these caveats. The first author, I note, is a cardiologist and not a pharmacoepidemiologist.

Now, this is a long rant but here is the other side of the  coin:  arrhythmia is not an intended effect. Yes, we can study the unintended drug effects and this paper brings up completely legitimate safety concerns, given that the issues with the data raised by Gelmen and others can be addressed. But that actual useful contribution is being buried under the far more "hot" mortality data.

James Watson even noted:
This caught my eye, as an effect size that big should have been picked up pretty quickly in the interim analyses of randomized trials that are currently happening. For example, the RECOVERY trial has a hydroxychloroquine arm and they have probably enrolled ~1500 patients into that arm (~10,000 + total already). They will have had multiple interim analyses so far and the trial hasn’t been stopped yet.
This is definitely a big point. It is totally plausible that arrhythmias could be missed (they are often in clinical practice) especially with such a sick patient population. But the mortality signal is awfully high for the trials not to have noticed (although the WHO is now checking their interim data, so maybe?). 

Anyway, I guess this is in my wheelhouse and I just wish we had a better interpretation of these studies in major journals.
Posted by Joseph at 9:19 AM 0 comments Links to this post
Subscribe to: Posts (Atom)