Election reform, the right way

There is always a great deal of interest in reforming the electoral college, particularly replacing it with a nationwide popular election for president. A constitutional amendment is not strictly needed for this because there is the National Popular Vote Interstate Compact. States which are members of the compact agree to pledge their electors to the presidential candidate that receives the most votes nationwide. The compact only takes effect once enough states have become members that the compact will determine the outcome.

But this election, and the last, revealed a more important threat that needs to be addressed independently. Bush v Gore exposed a surprising partisanship on the Supreme Court. In a close election the Supreme Court voted to intervene along party lines. There is no reason to believe this will not happen again. Indeed there is a remarkable new legal theory promoted by some members of the Supreme Court. In the words of Justice Alito, there should be strict adherence to “the provisions of the Federal Constitution conferring on state legislatures, not state courts, the authority to make rules governing federal elections". Much has been written about the absurdity of this, but the end result if this doctrine takes hold is SCOTUS would get to decide close elections.

So how close is close? Perhaps it does not matter. The party with control of SCOTUS need only file lawsuits challenging key states. SCOTUS could then hand them victories. Our system has a flaw which allows for a bloodless coup using two branches of government. This problem becomes more severe with a national popular vote because the Supreme Court becomes directly involved in contested elections. Actual vote counts are never cut and dried. There are always hanging chads, smeared postmarks, mismarked ballots, and rules that don't cover every contingency. A really close national popular vote would be a legal nightmare. Imagine a lawsuit in every single county in America. That's 3,141 lawyers for each side, and 3,141 judges making rulings, appeals all over the place, etc.

There is a straightforward way to mitigate this risk. A constitutional amendment that denies SCOTUS any jurisdiction over state selection of electors for current presidential elections. SCOTUS could still strike down state rules, for example to protect the voting rights of minorities, but only in hindsight. Temporary disenfranchisement is a risk, but it beats the permanent disenfranchisement of a coup.

[Edit: as it turned out one of the branches declined to join the coup. We're safe for now]

Proof of insanity

From the New York Post:

Now, travelers must show proof of a negative test taken within three days of arriving in the Empire State, and must also quarantine for three more days — and get a new COVID-19 test on the fourth.

And then later in the very same article:

New York’s statewide positivity rate is 1.49 percent.

 

1.49 percent is roughly 300,000 infected people already inside their borders. What exactly are they hoping to accomplish?

More NPR nonsense

Another grossly misleading article on NPR. This one is called "Americans Are Dying In The Pandemic At Rates Far Higher Than In Other Countries" Here is a link. 

Let's pick this one apart piece by piece. 

Overall deaths in the United States this year are more than 85% higher than in places such as Germany, Israel and Denmark after adjusting for population size. Deaths in the U.S. are 29% higher than even in Sweden

According to the CDC, 2,188,554 people have died in the US in 2020 up to October 13. That is 10% more than normal. If that is 85% higher than those other countries, they should have 40% fewer deaths than normal so far this year. That is obviously ridiculous, but just to be sure, here is some available data from those countries:

Here is a graph from Germany's Federal Statistical Office. 2020 is the red curve. The blue region is the range for the last few years. Looks like this is a fairly typical year there.

The website of Israel's Central Bureau of Statistics is not working, but here is an excerpt from an article in Haaretz: "A summary of Israeli deaths from March through July [2020] shows a similar trend, even slightly lower compared to last year. The figures show that 18,476 Israelis died during this period, 211 fewer than during the same period in 2019". That's a 1% decline.

According to the Statistics Denmark, 40858 Danes died from 1/1/2019 through 10/4/2020, and 40775 died from 1/1/2020 through 10/4/2020, a 0.3% decrease.

According to Statistics Sweden, there were 9696 deaths in the first 40 weeks of 2020 compared to an average of 8262 for the first 40 weeks of the previous 5 years. This is a 17% increase. So Sweden is actually doing worse than the US.

Even looking just at confirmed COVID-19 deaths, the number of people dying since May 10th — again after adjusting for population size — is on average 50% higher than every other country in the study. In addition the rate people are dying in the U.S. has stayed far above everywhere else. Emanuel says the current elevated mortality rates are important because they eliminate the chaotic early months of the pandemic when testing, treatment and reporting varied dramatically around the globe.

This is an unfair comparison. Many parts of the US had their epidemic at different times. For example, Arizona, Florida, and Texas didn't really get started until July. Europe was done by June.

If the U.S. had managed to keep its per capita death rate at the level of Italy's, 79,120 fewer Americans would have died.

That means Italy had 36% fewer deaths per capita. Actually Italy had 574 deaths per 1M population while the US had 660. That's 13% less. Not that that's good for the US, but at least get your facts straight.

Log log per capita

For fun I decided to recreate the log log chart I showed in my last post, but modify it using per capita deaths. Here is the result:

The highest country is tiny San Marino with 41 deaths in March and April and one death in May. The second highest is the US. What I find interesting about this graph is, no country in the world is getting over 1/1000 per capita deaths without curving down.

This suggests one way to think about the effect of Covid: out of every thousand people, it kills one (usually with underlying health issues).

Media bias in the age of Covid

Media bias has always been a problem, but I find it particularly brazen with regard to Covid. They do this partly for sensationalism, partly to responsibly promote behaviors they believe are beneficial, and occasionally out of partisanship. Here are a few examples.

Flattening the curve

Remember "flattening the curve"? The idea was to suppress the spread of the virus so that the health care system is not overwhelmed by simultaneous patients. Ideally you engineer social restrictions so that ICU beds are near but below 100% capacity until herd immunity takes over. The media loved this because it promoted correct behavior. But we don't hear this much anymore. That's because most locales are now at the far end of that curve, and the media does not want people to ease up on social distancing.

Log log plots

For a while, the media switched to log-log curves. They are really the same data, but they exaggerate the plunge when the curve starts downwards. This made the US look really bad compared to some other countries, encouraging social distancing. But when countries such as Australia started a second wave of infections, it ruined the effect.

Cases vs deaths

Here are graphs from google of cases over time and deaths over time. Notice cases are way up and deaths are way down. That's because there are 90 times as many tests available. Cases are a meaningless statistic. But they continue to be reported as a "resurgence".

Case fatality ratios

Early on the media would report dramatic "case fatality ratios". You can read them off of the google graphs above. They started around 10% and are now closer to 2%. Because it uses case numbers, it's another meaningless statistic. The interesting part is, the media doesn't report this statistic anymore. I wonder why.

The fate of New Zealand

New Zealand's response to Covid has been widely praised in the media. They controlled the outbreak with early aggressive quarantines, testing, contact tracing, and physical distancing. Borders were closed and returning New Zealanders were required to undergo managed isolation.

The policies were entirely effective. By early August, New Zealand went over 100 days with no community spread. The country went to it's lowest alert level (contact tracing and border restrictions only) in early June. Everyday life was nearly normal, but the country was living in a bubble.

The long stretch without community transmission was broken on August 11, when 4 new cases arose. The country has returned to higher alert levels.

In seemingly unrelated news, there's a good chance the pandemic of 1890, which killed over a million people, was not influenza as has long been assumed, but rather a novel coronavirus called hCoV-OC43, which jumped from cows to humans. What makes this especially interesting is that OC43 is very common today, and is one of several virus species that cause the common cold. The implication is that we have developed a sort of "herd immunity" to the more severe effects of OC43. Those whose immune systems were vulnerable died off, and new generations are immunized naturally at a young age when there is little risk.

This presents an interesting conundrum to New Zealand and other countries with similar harsh approaches to Covid. If Covid-19 develops the same way OC43 supposedly did, then those countries will never develop herd immunity, and they will need to continue living in a bubble for at least 130 years.

An open letter to Mark Zuckerberg

Mr. Zuckerberg,

You have said that "Facebook shouldn't be the arbiter of truth". I wholeheartedly agree. But after considerable pressure from advertisers, Facebook has finally capitulated and blocked one of Trump's posts about Covid. The post contained a video which included Trump saying that children are "almost immune from this disease."

Facebook justified the removal, saying "this video includes false claims that a group of people is immune from COVID-19 which is a violation of our policies around harmful COVID misinformation"

Of course "immune" and "almost immune" are not the same. In the absence of an obvious definition of "almost immune", let's just examine if the President's statement is reasonable from a common sense perspective.

As of July 11 in the US, 20 kids aged 5-14 and 6838 people aged 45-54 have died of Covid. The populations of these groups are about the same. This means school age kids have a 99.7% lower mortality from Covid than the older group. Or we can turn it around and say the older group's mortality from Covid is 34,090% higher. It may be hard to say if Trump's imprecise statement is false, but it is certainly not misleading.

The President does not seem fazed by his posts being censored. It just gains him more free exposure. I worry more about others such as myself, who attempt to expose ways in which our conventional wisdom has lost its way. We are dependent on social media companies, which operate as monopolies. Corporate censorship of alternative views is as clear a threat to our civil liberties as a state controlled media.

One way or another, this will not end well.

Covid and the parable of bear island

Imagine you are a member of a large stone age tribe on a large island. The island is inhabited by 6 bears. You go hunting every day. About once a year, you hear that one person was killed by a bear. Yet the people continue to hunt every day.

In April, you hear that 2 bears have swum over from the mainland. The tribe is terrified. It is decided to not go hunting and just gather berries near each cave. In May you hear that one bear swam back to the mainland. Still there is one unfamiliar bear and the people continue to avoid hunting out of fear.

Does these people seem silly to you? They should. And yet this is a pretty good analogy for Covid. For every 6 deaths from all causes among people over 35 in April 2019, there were 8 in April 2020. The extra 2 were from Covid. Now the death rate is lower. For every 6 deaths from all causes in July 2019, there were 7 in July 2020. These ratios are pretty consistent across age groups over 35 (younger people are much less affected by Covid).

So why so much panic about the 7th bear, but not the other 6? You might expect that people accept their mortality and the risks of everyday living. But if that were true, they would quickly accept the 7th bear. Apparently it is not acceptance, but denial. If we live in denial of the 6 bears, a bear swimming over from the mainland could indeed be terrifying.

To get a feel for this effect, it helps to imagine the island contains only people over 85. In that case the island population is only 5. In other words, 1 in 5 people over 85 die each year. That's pretty scary indeed. Yet in normal years, they continue to go about their lives without fear despite the 6 bears.

Not one of us will live forever. If we can come to acceptance that death is part of life, maybe we can avoid futile attempts to escape death at any cost.

Cost effectiveness of Covid-19 response

Shutting down major sectors of the economy was a drastic measure to try and prevent the spread of Covid-19. But was it worth the expense in terms of lives saved? Let's do an admittedly crude comparison. The leading causes of death this year, in rough decreasing order, will be heart disease, cancer, accidents, lung disease, Covid, stroke, and Alzheimers. For a fair comparison, we need to know the expected number of deaths and the cost to society of measures taken to prevent statistical deaths. We do not include individual treatments, as most of those costs are borne by the afflicted families after the fact, or through insurance payments, and are not as relevant for public policy. For most of the leading causes, it is difficult to estimate the costs, but we have a nice metric for cancer and Alzheimers, where we can use annual research and deaths from previous years. For Covid, an accurate measure of lost productivity due to the lockdown would be the loss of GDP. This has been estimated at $15T or more, but it is hard to pin down. We will be extra conservative and use the $2T stimulus for a rough order of magnitude.

Cancer deaths: 600k
Cancer research expense: $50B (private big pharma)
Cancer research expense: $5B (federal research)
Cost per death: $90k

Alzheimer's deaths 120k
Alzheimers research expense $3B
Cost per death: $25k

Covid deaths 145k (to date)
Covid response expense $2T
Cost per death: $13M

Here it is in a chart:

Covid-19

Demonstrations have sprung up in many states by people who believe our response to Covid-19 is overblown. They may have a point. With better testing of silent infections, New York City announced that 20% of residents tested positive for antibodies. Combining that with the number of deaths in the same time frame provides a more reliable death rate than has been reported. Out of every 1000 people infected, roughly 7 have died, with a heavier burden among those over 80, or with heart disease, lung disease, or diabetes. For comparison, in any given year, out of 1000 people, 8 will die from all causes, with a heavier burden among those over 80, or with heart disease, lung disease, or diabetes.

So being infected with Covid-19 is only as risky as a year of living, or perhaps a year of life is just as risky as a Covid-19 infection. Your perception may depend on your tolerance for risk.

Number of deaths attributed to Covid-19 to date in all states: 78,000
Number of deaths from all causes in 2017 (reported by CDC): 2,813,503