All posts from 2021.


2021 June 12


Scala & Kolacny Brothers - Creep (Radiohead cover)

Bo Burnham - Can’t handle this

Tish Hinojosa - Atlantico

Ratatat - Lex

The first guilty verdict was reached for an insurrectionist in the January 6 attack on the capitol. He was a lone actor who pled guilty to obstruction, and faces $2000 in restitution for damages, additional fines, and 15 to 21 months in prison.

Ever wondered why trains (depicted above), planes, and sometimes buses have a tiny fabric flap on the headrest? Today these are largely decorative, but these antimacassars were introduced in the 1800s in response to the widespread use of Macassar oil, a precursor of modern hair conditioner. Traditionally antimacassars would have been rather larger and also found on household furniture, including armrests, to protect them from damage. While antimacassars still have some utility today, modern hair products are far less oily.

The word cue is derived from the letter Q, an abbreviation for Latin quando (“when”), whereas the word queue is from Old French cue (“tail”) from Latin coda (“tail”).

Given two convex subsets of the plane whose boundaries are smooth, their Minkowski sum (the set formed by adding the points pairwise) has a boundary which must be C^4: i.e., it is four-times differentiable, with continuous fourth derivative. What is remarkable is that if we further require the boundaries to be real-analytic, then the boundary of their Minkwoski sum is C^{20/3}! That is, it is six-times differentiable, with the sixth derivative being Hoelder continuous of order 2/3. This bound is sharp.

Did you know that Berwick-upon-Tweed, a small town in northern England, was still in the Crimean war against Russia until the 1960s? Hopefully not, because it isn’t true. Berwick lies on the border of England and Scotland and had repeatedly changed hands between them, so much so that certain acts of parliament would specifically mention that they applied in Berwick. Apocryphally, Berwick was mentioned in the declaration of war on Russia but not in the peace treaty, leaving them arguably in a state of war until 1966 when the mayor declared peace, allegedly saying “Please tell the Russian people through your newspaper that they can sleep peacefully in their beds”. In truth Berwick was fully integrated into England by the time of the Crimean war.

The programming language Rust had in 2016 an obscure bug in which hash maps could in rare cirucmstances exhibit quadratic runtime behavior, depicted above. While the hash map implementation had been carefully designed to protect from all sorts of deliberate attacks and pathological behaviors, the quadratic behavior was triggered by remarkably simple means: reading elements from one hash map and adding them to another in iteration order.

In response to citizens of Mississippi passing a constitutional amendment legalizing medical marijuana with 69% approval, the Mississippi supreme court ruled that citizens may not amend the Mississippi constitution. The argument is that the constitution requires that proposed amendments must gather a certain number of signatures, with at most 1/5th of the requirement being fulfilled by signatures from each congressional district: however, Mississippi only has four congressional districts. 6 of the 9 justices in the Mississippi supreme court were unelected. See the linked podcast for a breakdown on the multiple levels of wrongness with this decision.

On the Japanese copy of the Instrument of Surrender marking the end of World War II, four of the signatories signed on the wrong line before anyone noticed: the representatives of Canada, France, the Netherlands, and New Zealand all shifted one line down, with the last signing in a blank area at the bottom. The scarcity of paper and access to printing meant a replacement could not be made.


“Tick tock Matt Gaetz”

My favorite Agatha Christie work, And Then There Were None, was made in 2015 into a superb 3 hour miniseries. It remains closely true to the book but for a few details it improves upon, with more vivid characters and excellent cinematography, writing, music, and acting.

Slow motion footage of a Saturn V launch taken from the launchpad, with detailed and informative voice over.

Lennart Green’s card magic


Poster from the 1980s Silence = Death Project, which promoted the gay rights movement by pushing coming out as a moral imperative. As discussed here, the AIDS epidemic and public disinterest in the survival of gay people made gay rights an urgent concern; the goal of a broad coming-out campaign was to humanize these victims.

The Privy seal of Japan is the Japanese Emperor’s seal, reading “Emperor” down the right and “Imperial seal” down the left. The seal has been used in some capacity for about 1200-1300 years. The current seal was manufactured in 1874 of pure gold, weighing 4.5 kilograms.

“He’s” is short for “he is”.

Trees in the Ancient Bristlecone Pine Forest, most of which are over 4000 years old.

Defeating influenza: lessons from covid

2021 May 27

A British World War II propaganda poster. The existential threat of war forced a reckoning of where inefficiencies lay, such as the cost that our society bears due to illness.

A victim of the seasonal flu being transported. Source

  1. Introduction
  2. Flu vs covid: the data
  3. Viruses hate this one simple trick to eradicate contagious respiratory disease
  4. How to defeat seasonal flu
    1. Public messaging
    2. Testing
    3. Masks
    4. Sick leave
    5. Travel screening
    6. Vaccination
    7. Contact tracing
    8. Non-measures: hand washing, deep cleaning, social distancing, and physical distancing
  5. Cost-benefit analysis of flu intervention in the US
    1. Costs of the flu
    2. Costs of proposed interventions
    3. Costs vs benefits
  6. Eradication of influenza
    1. Flu seasonality
    2. Animal-to-human transmission
  7. Preventing pandemics
  8. Mutations: the flu fights back
  9. When to start work against the flu?


In the background of the worldwide fight against covid-19, other respiratory disease has been struck a heavy blow. Incidence of seasonal flu has plummeted more than 99% globally: in the peak week of the 2020-2021 flu season, only 412 cases of the flu were laboratory-confirmed worldwide, with many countries not detecting the flu in the whole season. (Note that the WHO report has the data for northern and southern hemispheres reversed.)

As with all images, click to zoom.

This shows that the death and disease brought by the flu virus is preventable. With the average US flu victim having 16 years of life left, we should not resign ourselves to the inevitability of this illness. I believe in several decades we will regard deaths from the seasonal flu the way we today regard deaths from measles or cholera. Let us push to achieve that future now instead of waiting for it passively.

Although very few countries, such as New Zealand and China, have successfully been able to contain covid, this widespread accidental suppression of the flu is possible because the flu is much less contagious than covid. Evidently the global anti-covid efforts are commensurate with the effort needed to convincingly defeat the flu, but grossly insufficient for fighting covid. (Much as the mediocre efforts to mitigate climate change have incidentally led to modest improvements in air quality while being orders of magnitude less than needed for climate change.)

As the flu is less contagious than covid, intuitively one might suppose less effort should be expended in the fight: after this logic has been filtered through policy makers we end up with the present level of effort, to wit, middling vaccination campaigns, and some companies let people with the flu stay home. This is the wrong intuition. Rather, the flu being less contagious implies that there is more benefit from the same degree of effort and thus we should be looking for any useful avenue by which to fight the flu.

I propose that we can use the exact tools that we’ve been using to half-heartedly fight covid to defeat the flu. I believe this is possible without the invasive strategies needed to defeat covid, although such strategies should be held in reserve if necessary. And if we start right now, during the covid pandemic, we can make full use of the massive head start on defeating the flu before it slips away.

There are two components of this fight: specific government interventions against the flu, and a change in public perception of the flu as an accepted and inevitable component of daily life. I will focus on the former below.

Related articles:

Flu vs covid: the data

Seasonal flu Covid-19
Incidence estimated 800 million annually 145 million official cases as of 2021 April
Mortality estimated 500 thousand annually 3 million official deaths as of 2021 April
Mortality rate estimated 0.05% estimated 1% (counting only official cases gives 2%)
Incubation period 2 days (1 to 4) 5 days (2 to 14)
Start of infectious period 0 to 1 days before symptoms often 2 to 3 days before symptoms
Seasonality winter in temperate climes; year round / complicated in tropics year-round
R_0 1.19 to 1.37 2.0 to 4.0

I do not trust most scientific work I have read for estimating R_0, and would discourage the lay use of R_0 values; I only included it in the table to motivate writing this disclaimer. It is sufficient to observe that covid is much more contagious than the flu without trying to exactly quantify the relationship. Note that covid has become steadily more contagious as the pandemic has continued.

(If you were to use R_0 to quantifiably compare disease, the appropriate measure would be the ratio of the logarithms: this calculation comes out to covid being 2 to 8 times more contagious than seasonal flu.)

A significant fraction of infections of both flu and covid are asymptomatic. Research suggests that transmission of the flu from asymptomatic or presymptomatic people is rare. In contrast, it appears that asymptomatic or presymptomatic transmission of covid is responsible for about 50% of covid cases, and there are many known instances of superspreading events initiated by people without symptoms at the time.

Viruses hate this one simple trick to eradicate contagious respiratory disease

Lockdowns are the gold standard for eliminating covid. Very few countries enforced any lockdowns, and to my knowledge all of them now have almost zero community transmission of the disease. (Although many used the word “lockdown”, it turns out actual lockdowns are much more effective than saying the word.)

The facile solution to all contagious respiratory disease is simply, what if we all stayed home for two-ish weeks? Covid, the flu, colds, and other disease – all gone. The cost would be high, but the benefit far higher.

Number of detections per week of various respiratory disease in Austria following a closure of various public gathering spaces, notably including schools. Detections were made by testing samples taken from patients with influenza-like-illness through a national surveillance network. Only about 1/4th of the samples were tested for RSV, hMPV, and RHIV. The study does not say whether the number of samples tested varies week to week.

Of course, there is no half-lockdown. If one small town somewhere failed to fully commit, it could wholly ruin the project. If we can’t coordinate 8 billion people well enough to give everyone drinking water (according to the CDC, 700 million people lack an adequate source of drinking water and almost a billion have zero sanitation), then I rather suspect it might be somewhat beyond our ability to coordinate 8 billion people to lockdown for a few weeks.

Lesser lockdowns, limited to geographic regions suffering from an outbreak of the flu, may still be useful; certainly these have been successful against covid. While I think it would be wise to keep in reserve the option of local lockdowns, I doubt they would ever be necessary – I believe the flu could be defeated globally without the use of lockdowns or other large-scale interruptions to daily life for the general public.

How to defeat seasonal flu

I present the general structure of one strategy for defeating seasonal flu; it is meant as a starting point for discussion and comparison. Specialized expertise and research would be needed to develop a refined, actionable plan.

The components of the strategy are:

This should be very familiar as they are lifted directly from interventions against covid, with lockdowns and distancing omitted. I’ll elaborate some details about each intervention below.

The steps of the plan are:

  1. Vaccination, sick leave, and situational use of masks to drive down the incidence of flu
  2. Testing and contact tracing so that resilient clusters of cases are rapidly identified and contained
  3. Travel screening to protect from regions with endemic flu and reduce the chance of clusters spreading geographically.

Eventually the only noticeable burden for most people will be annual vaccination and regular travel screening; those developing any respiratory illness will use a home testing kit, and any rare cases of the flu will be followed by isolation and extensive testing of their contacts. Ill people will wear masks regardless of test results. If measures are loosened too far and the flu escapes containment, the outbreak is naturally reset in the summer and a fresh start can begin next flu season (for countries with flu seasons).

Public messaging

Defeating the flu is not possible unless people are onboard with the project. People need to know what the goal of the campaign is, the steps being taken, and their role. Most of all, there needs to be a change in the public attitude that the flu is merely an accepted, inevitable part of daily life, which just happens to kill a large fraction of elderly people each year – an attitude I held until very recently.

I expect such a change in attitude to largely precede any government action, so at least some of this messaging needs to be done by non-government actors, which is why I am writing this.


In many individuals, the flu creates symptoms easily confused with the common cold or other respiratory illnesses. Unless we plan on taking on all contagious respiratory illnesses at once (a strategy I did not analyze), testing will be necessary to identify cases of the flu against a background of lesser illnesses.

As will be a common theme in the following subsections, the appropriate level of testing depends on the contemporary incidence of the flu. Flu tests are readily administered at most hospitals, and this practice should be continued, but it is not sufficient to only test people who are seeking medical attention. In some cases it may be appropriate to screen individuals in advance of coming to some gathering, as discussed in the travel screening subsection. On demand testing at doctor’s offices, pharmacies, or specialized testing centers should be made available to the public.

The one essential change that would greatly increase flu testing is to make testing at home cheap and standard practice – this is also a big missed opportunity in covid testing. It takes initiative and investment to get in a car and drive to a testing site, wait in line, provide some identification and fill out a form, and then have to wait to speak with someone on the phone about your test results (or wait at the site, or dig up your login credentials for their website). If you are sick, fatigued, and not looking presentable that can be an additional obstacle. With home testing you don’t have to leave your bed, which is especially relevant to someone too ill to stand up.

Furthermore, home testing is faster than testing-on-site, which is necessary for successful contact tracing, as discussed below.

There are no technical obstacles to adapting existing flu tests for the home, as they have multi-year shelf life, require no lab equipment, and can be performed by unskilled and untrained users: if you can follow a 3-step Ikea pictographic instruction, you can perform a flu test. However no such test is approved or packaged for home use, and tests suitable for consumer use are still in development.

For covid, certain universities and companies practice population-scale screening by requiring biweekly tests for their community. This is key because of the long incubation period of covid and widespread asymptomatic transmission; I don’t expect such measures to be necessary or helpful for defeating the flu, and such screening would pose a significant burden to the public.


Sick people should wear a mask: this applies equally for any contagious, respiratory illness, and regardless of whether there is a larger fight against the flu. I hope that the creation of such an expectation in the West is a lasting consequence of the covid pandemic – I had not given it any consideration before 2020, but now it seems like it should have been obvious. A similar cultural change would be replacing handshakes with nodding or bowing the head, which is already common in the US and should be made the default practice.

The use of masks among people with no known illness can be a cheap though only modestly effective way to reduce transmissions in situations involving lots of disparate contact. There is already a precedent for certain occupations, notably medical workers, to have frequent mask usage. Occupations that involve lots of public contact are also good candidates for masks: ushers, waiters, cashiers, barbers, etc. Having a mask-by-default rule, but allowing the removal or lowering of a mask when inconvenient or away from the public, allows for minimal burden while retaining most of the benefit.

Requiring mask usage among the general public for certain large gatherings or while on public transit would also cut incidence of the flu, and may become necessary to address transmission events that cannot be reasonably traced. While such mask usage is cheap and not especially odious, it may be too much to expect the public to sustain for an extended period of time. I have tried to restrict myself to interventions that do not require daily action from a significant fraction of the public, which rules out requiring masks in public transit. I would only push for such measures if mandatory sick leave combined with aggressive vaccination failed to produce sufficient results.

Sick leave

Obviously letting sick people stay home from work reduces the spread of illness. Among developed countries this is largely a US- and Japan-centric problem. I don’t think more needs to be said here, but did you know that workers on the ancient Egyptian pyramids got paid sick leave?

Sick leave is especially pertinent to the flu as almost all transmission is from symptomatic individuals. It is estimated that during local flu outbreaks, 80% of otherwise healthy people with a cough or sore throat have the flu virus. These symptomatic people should stay home even if they feel well enough to work!

Travel screening

If we can invasively screen every airplane traveler’s luggage at great expense in search of the remotest chance of someone carrying a bomb, then we can handle flu tests and fever checks as well. Fever checks with infrared cameras are already standard practice in many countries. Existing rapid flu tests cost $5 to $10 and produce results in 15 minutes; all people entering the secure region of the airport should be screened, and arrivals from locations not performing screening should also be screened. Since you have to wait more than 15 minutes at your gate before boarding anyhow, in theory this could be designed so that it takes no more time than the actual administration of the test. While these tests have a noticeable false negative rate, particularly for people who are not contagious, they have a low false positive rate.

With economies of scale and/or adjusting the design of flu tests for this use case, the cost can likely be driven down relative to using existing off-the-shelf rapid flu tests. It may make sense to pool the samples from everyone on the same flight if interventions don’t depend on which passenger tests positive.

If we wanted to actually protect other people at the airport from catching the flu, we would need some kind of layered or staged system so that travelers are not admitted into the general airport population until their flu test has come back negative. Not coincidentally, these are the exact same changes to airport layout that would be necessary to protect against physical security threats, in case we wanted to upgrade from just security theater. However I don’t expect such changes to be worth the investment: a cheaper alternative would be to require travelers to wear a mask until they have confirmed negative, and mask policies can be easily changed to adapt to changing circumstances.

Possible responses to positive tests among travelers include:

The appropriate reaction to the receipt of positive tests depends on the existing incidence of flu in the people being tested, and in particular might vary with time of year. As community incidence of the flu decreases, a higher fraction of positive tests are false positives and follow-up interventions will be more capable of handling secondary cases, so less intervention is necessary. However, as the flu gets close to eradication (either locally or globally), the potential benefit of getting the rest of the way there greatly increases, and stronger interventions become favored again.

In the extreme, mandatory isolation of all people crossing an international border may be taken: probably 24 hours would be sufficient. However this would pose a much greater burden than the other measures I have proposed here, and I hope that nothing of the sort would be necessary.

Screening may also be appropriate for long-distance train, bus, and car travel, or before large gatherings.


Vaccination is the one measure that developed countries already expend a roughly commensurate effort on. In the US, almost 50% of adults were vaccinated for the 2019-2020 flu season, the highest rate of the decade:

Given that vaccination is on the order of 50% effective, if a randomly distributed 50% of Americans were vaccinated, then naively using an R_0 value of 1.33 for the flu suggests that flu outbreaks would be self-limiting in the US.

Observably this is not true, and the current level of vaccination is insufficient to defeat the flu. However there is a lot of room for improvement on 50% of adult Americans receiving vaccines each year. I believe that achieving 90% flu vaccination alongside technological improvements in vaccine effectiveness would nearly eliminate the flu in the US, with relatively little effort on other interventions needed.

To achieve a goal of 90% flu vaccination,

Currently two thirds of flu vaccines are administered in doctor offices. Scheduling an appointment with a doctor can be inconvenient, a financial burden, and carry negative mental associations; flu drives would address these issues, and incidentally raise public awareness of the necessity of flu vaccinations.

Children are a major vector of flu illness and must be vaccinated as well.

While 90% vaccination makes a simple public-facing goal, really the goal is that each community separately reaches an acceptable vaccination rate, say 60% or so. Without this, flu can remain endemic in such communities and will constantly resurface elsewhere. I expect most of the cost of vaccinations to go into outreach to communities with little engagement with the formal medical system. Poverty, religious fundamentalism, and science opposition are a hindrance here, along with everywhere else.

Vaccine technology has been advancing rapidly in the last decade, as anyone who has heard of mRNA is aware. Chicken eggs are no longer used in the US since 2019 (I have seen conflicting information: the CDC says 80% of flu vaccines for 2021 were made with chicken eggs). A variety of non-egg flu vaccine technologies have recently begun use in the US. The mRNA platform is not yet used for the flu but might lead to more efficacious and better targeted flu vaccines in the future.

Note that if the global number of flu cases is brought down significantly, the mutation rate declines, and vaccination becomes substantially more effective. This compounds with any technological improvements to the vaccine.

Contact tracing

Most of the interventions discussed, such as travel screening and vaccination, are non-targeted population-scale efforts. These are cheap ways to greatly reduce infection among the large majority of people, but each intervention by itself will miss a significant fraction of cases: very highly effective travel screening would pose a tremendous burden, as all travelers would need to quarantine for several days, and existing vaccines simply aren’t highly effective, even if we could reach 100% vaccination rates. The purpose of contact tracing is to clean up the outbreaks that escape the cheap, population-scale interventions.

Similarly, the cost of contact tracing scales approximately with the number of cases, whereas the cost of population-scale interventions scales with the total population. As incidence of the flu goes down, contact tracing becomes cheaper and more beneficial, while other interventions cost the same for less benefit.

With existing incidence of the flu contact tracing is nearly meaningless: during flu season, a large fraction of working adults will become possible-contacts of someone with the flu, so identifying which people are possible-contacts doesn’t add any new information. However, after the incidence of the flu is driven down by other interventions (or by it not being local flu season anymore), contact tracing is essential for identifying who is at risk and stopping further transmission. It is also essential for identifying existing outbreaks of the flu: health agencies only find out about a small fraction of flu cases through hospital reporting, and other cases must be found by working backwards from known cases. This is especially important for communities with little engagement with the formal medical system. Note that increasing reporting of the flu has a substantial impact on reducing the number of outbreaks that escape detection from contact tracing.

Of course, simply identifying who has the flu and their contacts does not serve a purpose unless some action is taken in response to that information. A variety of responses are possible:

The appropriate degree of intervention depends on the incidence of flu in the community, and the fraction of flu clusters being successfully identified. If most everyone is exposed to the flu anyways, what’s the harm if some cases escape from one of the known clusters? Conversely, if very few cases remain, then there is a very large benefit to stopping clusters from getting out of control into the general community.

A key component of contact tracing identified in the WHO report on China’s actions against covid is speed. China hired 10000 contact tracers to address an epidemic with only 80000 confirmed cases in total, and heavily emphasized rapid testing and tracing. Contact tracing must be very fast – symptoms of the flu begins an average of two days after exposure, usually 1 to 4. A delay of 24 hours could be the difference of a whole level of indirection of contacts that need to be identified, and the number of people involved grows exponentially with the number of levels of indirection of contact. Therefore successful contact tracing of flu cases requires

As contact tracing is the lynchpin for the closing fight to defeat the flu, each of the above are essential components of the larger project.

Non-measures: hand washing, deep cleaning, social distancing, and physical distancing

Some interventions often discussed in context of covid are absent from my flu plan because I don’t think they have a significant effect, either for the flu or covid.

Hand washing with soap was a great invention, and something more people should take advantage of. It is very effective for digestive ailments, diseases found on raw foods, and disease with fecal-oral transmission, among others; for respiratory illnesses, direct inhalation of droplets is generally the greater danger. Certainly hand washing should be promoted but it will not have a measurable impact on the flu. Furthermore, a public health campaign can only influence the behavior of so many people, and it is difficult to enforce hand washing mandates outside of very narrow contexts.

Whatever “deep cleaning” really is, it doesn’t help reduce spread of respiratory disease.

“Social distancing” refers to reducing the number of social contacts you have, and how frequently you interact with them, to reduce the connectivity of the graph of close contacts. This can be effective if a small number of people enter a mutual agreement to form a “quarantine circle” that carefully restricts contacts outside the group. However, I imagine unilateral spontaneous cuts in contacts to be totally useless:

“Physical distancing” refers to maintaining a physical separation from others, especially from members of the general public who are not social contacts. Under laboratory conditions, the quantity of respiratory drops from someone rapidly declines with increasing distance. However, I am skeptical about the effectiveness of physical distance in practice, as rooms with closed circulation that do not rapidly exchange air with the outside seem to be the main vector for respiratory transmission, and after an extended period of time in a small such room the location of the people within it becomes irrelevant. I still practice physical distancing for whatever marginal benefit it may provide. Regardless, what I’ve observed is that people do not practice physical distancing no matter how many signs demanding it are posted or how often it is repeated in the media. The only place I have seen it successfully enforced in the general public is at large gatherings, but if you are faced with a disease so dire that physical distancing becomes worth the effort, then large gatherings should be banned long before.

Cost-benefit analysis of flu intervention in the US

The cost of the interventions described above varies greatly depending on how far along the fight against the flu has progressed: with local eradication, most of these interventions are unnecessary or can be scaled back, and with global eradication the cost becomes zero. A full cost-benefit analysis would require far more work and domain-specific knowledge than I can contribute. However, the costs of each of the interventions separately can reasonably be estimated, and can be compared with the existing costs of the flu to see what level of intervention is break even. Near our present level of intervention, I believe the benefits of intervention should scale roughly linearly with the level of intervention; as the amount of flu cases are driven down, non-linearities will develop, until eventually at eradication the ongoing cost becomes zero.

Costs of the flu

Extensive scientific work on the burden of seasonal flu has been conducted, and it is infeasible for me to do a critical analysis of this body of knowledge, so I will uncritically cite some studies selected at random.

In the years 2010 - 2020, the CDC estimates that in the average year there 30 million symptomatic flu cases, 14 million medical visits, 440 thousand hospitalizations, and 36 thousand deaths. In this time interval the average US population was 320 million, so 10% became ill, 5% had a medical visit, 0.1% were hospitalized, and 0.01% died. Young children are more likely to become ill than adults.

How many is 36000 deaths? The average American has a 1.3% chance of eventually dying of the flu. Like many diseases, deaths occur disproportionately among older people, but that is not to suggest victims of the flu are already on death’s door: a 2007 study finds that the flu is responsible for 611 thousand life-years lost annually. As this study also estimated 36000 deaths annually, the average victim of the flu had 16 years more life left; and overall, the average American loses more than two months of life to the flu.

Lacking better data, let us imagine the typical person with the flu spends four days convalescing before they have sufficiently recovered to be productive or engage in enjoyable activities (which is a form of productivity, in that it increases human welfare, just not in a way that you can exchange for money: e.g., you can’t make someone happy by selling them your happiness, but you can sell them your chocolate, which is close enough).

In Canada, the average hospitalization lasts 11 days and costs about $10000 USD, has a 14% chance of going to the ICU; it appears US numbers are comparable. In the US, direct medical expenses come to around $10 billion annually. When including “indirect” expenses, studies give “total” costs of the flu at around $90 billion annually, mostly from loss of wages and productivity; however I would be wary of taking such a number too literally, as it will be very sensitive to how it is calculated.

From 2010 to 2020, the CDC estimates that the existing level of flu vaccinations have annually averted 5.1 million symptomatic illnesses, 2.5 million medical visits, 72 thousand hospitalizations, and 6000 deaths. This is about one sixth of the non-averted harm. Note that these benefits are disproportionately felt by those who get vaccinated.

Combining the above figures, and linearly extrapolating the averted harm, each year the average American:

The above costs, including those already being averted by existing interventions, are approximately the budget for defeating the flu while unambiguously being net beneficial. Additional benefits, such as those discussed below in the section on preventing pandemics, may be far larger but are harder to quantify without sophisticated analysis.

Costs of proposed interventions

Let us begin with the cost of vaccinations. Currently, half of all Americans are vaccinated each year, at an administrative cost of $20 per shot; vaccinated adults below 65 spend an additional $14 in time spent receiving the vaccine – this compares favorably to the averted cost of $90 per vaccinated person. While there may be some cost savings to be found in vaccinating all Americans or in technological improvements in vaccines, more realistically I expect the difficulty of reaching the most hesitant or inaccessible Americans to increase administrative costs, say to $30 per person plus time loss of one hour.

Existing off-the-shelf rapid flu tests cost about $5 to $10 each. As there are about 700 million security screenings at airports per year, if all passengers are tested for the flu then the average American will receive such tests twice a year. This additional expense would be included in ticket prices, as other screening expenses already are: most tickets include a $5.60 security fee, which offsets the $10 the TSA spends on each screening. I suspect most travelers will not notice the price increase (did you already know how much the security fee is?), and as the people paying are those who most directly stand to benefit from the screening, this would increase the palatability of such testing to the public.

While in theory the 15 minutes of waiting for test results could be parallelized with other waiting time in the airport, due to inevitable inefficiencies I will suppose travelers lose 10 minutes to the testing process. Home tests perhaps take 20 minutes due to lack of familiarity with the procedure, although if performed by the sick person that time would probably be lost anyhow.

In addition to general airport screening, we will strongly encourage the use of home testing kits. The average adult has 1 to 2 colds a year along with the flu 0.1 times; ideally all of these illnesses would be tested. Let us suppose the average adult usefully uses 1 home testing kit a year and wastes another 1.

Finally, if one additional test is performed each year for miscellaneous purposes, such as screening at large events or other modes of transport, this makes 5 tests per year for the average American. About 1 of those would be due to air travel outside of flu season, which could be omitted. If the flu were locally eradicated it would only be necessary to test international arrivals, which is less than 0.5 per American per year. Most of those international arrivals are air travelers.

If economies of scale allow the TSA to screen people at only $10 per person, it seems only fair to suppose similar savings would occur with flu screenings as well. As flu tests currently cost $5 to $10, I will imagine that economies of scale will conservatively bring the cost to at most $5 per test. The cost of adding infrared cameras to identify fevers among travelers is negligible.

In all, the average American will be tested 5 times at a cost of $25 and one hour per year.

Disposable masks are currently available to end consumers at 20 cents each. The cost of using one per day of illness and per flight is negligible, but let us add a time loss of one hour to cover the inconvenience of getting and dealing with masks. During the covid pandemic, the US planned to distribute 5 re-usable masks to every household, in total 650 million masks, at a cost of $1 per mask; this would only be a one-time expense rather than repeating every year and only modestly more expensive than disposable masks, but still negligible.

Contact tracing would not be performed, and thus have no cost, until the incidence of flu is reduced by several orders of magnitude. For the eradication of the last, intransigent cases, contact tracing will become greatly expensive per case due to the effort to reliably identify almost every transmission event, although the total cost would be manageable. Precedent for contact tracing of respiratory diseases include tuberculosis, measles, and the handful of countries that were able to eliminate covid. I have no data on the cost of these contact tracing precedents and would find it difficult to extrapolate to the flu.

Costs vs benefits

In total, the estimated annual cost of the above interventions works out to about $60 per person, plus three hours of time, plus the cost of contact tracing. Of that, about $10 and half an hour is already being spent on flu vaccinations. These costs would go down if the flu were locally or globally eradicated.

The total cost of the seasonal flu each year is about 26 hours sick at home or dead and $30 in direct medical expenses, plus 5 more hours and $5 averted by current flu vaccinations. (There is an additional $240 (and $40 averted) in social / productivity costs, but “social costs” includes time spent sick or dead converted to money.)

Additionally, the costs of the flu are not equally borne by all people. Most people will never suffer from a serious case of the flu: only an unlucky few flu victims will be hospitalized, be permanently injured, or die. Flu interventions are a form of insurance, with a guaranteed small loss protecting against a small chance of a very large loss. The risk-averse option is generally preferred even if average costs were equal.

Furthermore, the unequal costs of the flu are correlated with other inequalities, as poverty increases one’s exposure to illness and susceptibility to illness if exposed, and decreases access to medical care, including very basic care like staying home from work when ill.

Finally, there are indirect benefits to defeating the flu, as elaborated in the section on pandemics. I expect there to be many other ancillary benefits beyond those specifically mentioned.

In balance I find interventions against the flu to be overwhelmingly preferred.

Eradication of influenza

I have left ambiguous what exactly it means to “defeat” the flu. It makes little difference to human welfare whether the flu is merely defeated, by which I mean its incidence is reduced many orders of magnitude like polio, or whether the flu is eradicated, like smallpox. However, different strategies are suitable for these goals, and it is not immediately apparent which goal is more appropriate.

Likewise, a strategy needs to be designed around either the goal of local eradication (formally the term is “elimination”) or global eradication (formally, just “eradication”).

Is eradication of the flu plausible? Based on our track record of disease eradication so far, the safe bet is that the flu will not be eradicated in the coming decades – although measles, which is much more contagious than the flu or covid, is on the shortlist for eradication. None-the-less I believe global eradication of the flu is within our capability (though see animal-to-human transmission, below), and should be explicitly taken as a primary or secondary goal.

Flu seasonality

Understanding in what way seasonal flu is seasonal is essential to deciding whether to pursue local or global eradication. Scientists still do not fully know why the flu is seasonal or where it “goes” in the off-season.

In temperature zones, the flu nearly disappears in summer, and consistently resurges each winter. It is often suggested that warm weather causes conditions to be less favorable to the flu, but this immediately appears implausible as the flu is endemic year-round in parts of the tropics.

This contradiction remains unresolved. However, it appears that the key factor is humidity, which is closely correlated to temperature: the flu spreads best either when it is cold and dry, with an increase in cases observed after dry spells, or warm and humid, or possibly warm and rainy in specific. For example in India, the primary flu season begins in spring / summer, concurrent with the monsoon season.

In tropical regions, the flu is often observed to have two distinct seasons, or to otherwise be significant throughout the year.

So where does the flu come from in temperate regions? It appears most likely that the flu regularly spreads from south and east Asia, where it is endemic, to other parts of the world; such seed cases only grow to form large outbreaks in the local flu season, when conditions are favorable enough for the flu to be self-sustaining. The obvious question then is what flu seasons were like before the industrial era, but I could not find relevant information. Alternative hypotheses are that the flu alternates between northern and southern hemispheres, spreading directly from one temperate region to the opposite temperate region, or that it remains endemic at low levels within each region even within the off-season; this last hypothesis seems self-evidently improbable.

For a temperate country seeking to defeat the flu, this suggests that it would be viable to forego all but the most cost-effective of internal measures and focus on preventing flu from crossing the border. This is more feasible than with covid, as there is very little non-symptomatic transmission of flu, and flu cases tend to have a short incubation period of only two days, limiting the period of surveillance necessary. However only a few cases of flu escaping detection at the border would render the strategy useless without arduous barriers to international travel.

Alternatively, as part of a global campaign to defeat the flu, one could similarly forego expensive interventions in parts of the world without endemic flu, and focus everything on south and east Asia. India and China each have flu vaccination rates below 5%; it is hard to imagine much forwards progress against the flu without improvement there. If we were able to break the cycle of flu transmission in Asia, perhaps we would get the rest of the world for free. However this is not the huge savings one might hope for: half the world population lives in these endemic regions, and obstacles to universal vaccination are high there.

I think neither of these risky strategies are a good gamble. Instead, either a country can unilaterally fight the flu locally, employing both strong internal interventions and robust border protections from the flu, or the world can aim to globally defeat the flu, with strong internal interventions in all countries whether temperate or not, and only weak border protections in use.

Animal-to-human transmission

The number of animal-to-human transmissions of the flu is totally negligible compared with human-to-human transmissions, but is large enough to pose an obstacle to eradication of influenza A. (Influenza B only spreads significantly among humans.)

Active measures are taken in the US to reduce incidence of flu in domestic pig herds. Pigs are believed to pose a specific danger of pandemic flu because of the ability for avian flu, swine flu, and human flu to co-infect pigs.

Avian flu is common in domestic bird flocks. Certain types of avian flu are naturally found in wild waterfowl, from which they can spread to domestic birds and humans unmonitored, making the eradication of those strains essentially impossible for current techniques.

Horse flu, dog flu, bat flu, and others are not thought to pose a particular risk to humans.

I can’t usefully speculate on a strategy for addressing animal-to-human transmission. Perhaps the only answer is to concede eradication of certain strains as infeasible and accept low levels of flu among those who directly work with animals, with active measures to prevent further spread.

Preventing pandemics

With influenza eliminated or eradicated from the human population, human-to-animal transmission of the flu would likewise end, and human-specific adaptions to the flu virus circulating among animal hosts would gradually die out. This decreases the danger that future animal-to-human flu transmissions carry the mutations necessary to be contagious among humans, and particularly reduces the rare chance of a flu pandemic event where such a virus is especially well adapted to humans.

Besides reducing the risk of such events, reducing incidence of the flu makes it much easier to spot incipient pandemics in the early stages when they can be most easily stopped. That place where 100 people suddenly have the flu stands out a lot more when noone else has the flu. This applies for other pandemics as well: covid-19 was first identified as a novel disease at the end of December 2019, by which time 174 cases of covid were already known. This early virus was far less contagious than the later variants, and genetic analysis suggests that it circulated for perhaps one to four months among human hosts prior to its identification – this delay being necessary for it adapt to human hosts. During this same time interval, Wuhan was suffering from a particularly severe flu season, with one surveillance hospital alone detecting about 300 cases of the flu each day. It is remarkable that against this background of approximately one million people becoming ill with the flu, it was possible to detect covid at all!

But what if there were no flu? Covid could have been identified much earlier, and testing kits and vaccines similarly designed earlier. No specialized infrastructure for contact tracing would need to be rushed to completion, as contact tracing is much easier when there isn’t a similar but much more common disease circulating at the same time – and the infrastructure would already have been built and tested for dealing with the flu. Even if the initial outbreak wasn’t stopped and covid began to slip out of China, the head start on making test kits and lack of an ongoing flu season would make identification of incoming cases more reliable. It’s hard to imagine covid growing into a pandemic in such a environment.

Mutations: the flu fights back

It is possible that in response to a plan such as the one I have laid out, the flu will evolve to defeat the interventions under taken; for example, it could evolve to become more asymptomatic, as covid appears to have done. If only one country undertook such an effort, maybe they produce a local variant adapted to the circumstances that escapes to other countries before the local eradication program can be successful. However, I expect this locally adapted variant to be outcompeted by the wild type in other countries, and so not contribute to the flu gene pool.

While I don’t foresee flu mutations posing a serious problem to the above plan, a well-designed plan should properly consider such an issue and prepare contingencies before the plan is begun.

When to start work against the flu?

Now! As with most progressive causes, the earlier we make progress the sooner we begin to collect benefits. But with defeating the flu, we have an absolutely golden opportunity during the covid pandemic. Interventions against the flu and against covid largely overlap, and working on one benefits the other. Cases of flu have plummeted to almost non-detectable levels worldwide, and defeating the flu has never been cheaper than right now. New Zealand has already locally eliminated the flu by accident. Eradicating the flu almost comes for free with eradicating covid, it’d be an agonizing failure to do the latter without taking up the “two for the price of one” offer on sale right now. In a few years the flu could be right back to where it was before the covid pandemic, and we’re back to paying full price – though still a price worth taking.

Fighting the flu is a fight we can win, as the last year has ably demonstrated. Currently the flu is on its last legs, we just need to finish the job.

What is gravitoelectromagnetism?

2021 May 08

In Newtonian gravity, the acceleration towards a massive particle with mass M and distance d is

\frac M {d^2}

times some constant. What happens if the particle is moving? We know from relativity that information cannot travel faster than the speed of light c, so any movements of the particle in the last d / c time cannot influence the current gravitational acceleration.

One might guess that a test particle is simply accelerated towards where the massive particle was d / c ago. However, in relativity, it is impossible to measure your absolute velocity – only the relative velocity between particles is observable. Therefore if the test and massive particles are moving at the same velocity, then the acceleration must be directly towards where the massive particle is and not towards where it was, in contradiction to our guess.

Reality is more complicated, and we find that the gravitational force depends on the velocities of the particles. This is analogous to electromagnetism: while both the electric and magnetic forces were discovered before relativity, with the discovery of relativity it was realized that the magnetic force is simply the velocity-dependent component of the electric force.

One observable consequence of the gravity’s velocity-dependence is frame-dragging, in which a rotating massive object causes nearby objects to counter-rotate.

Assuming that gravity, like electromagnetism, is Lorentz invariant we can work out the full non-static gravitational equations – although see the caveat at the end. The resulting theory is called gravitoelectromagnetism, and was suggested in 1893 by Oliver Heaviside.

Maxwell’s equations for electromagnetism are

    \nabla \cdot E &= \frac {\rho}{\varepsilon_0} \\
    \nabla \cdot B &= 0 \\
    \nabla \times E &= - \frac {\partial B}{\partial t} \\
    \nabla \times B &= \frac 1{\varepsilo...

where E and B are the electric and magnetic fields, \rho is the charge density, and J is the charge flux; the other terms being physical constants.

Then the equations for gravitoelectromagnetism are almost identical:

    \nabla \cdot E_g &= -4 \pi G \rho_g \\
    \nabla \cdot B_g &= 0 \\
    \nabla \times E_g &= - \frac {\partial B_g}{\partial t} \\
    \nabla \times B_g &= - \frac {4 \pi G}{c^...

Here \rho_g is mass density, J_g is mass flux, and E_g is the “gravitoelectric” field (i.e., the conventional gravitational field) and B_g the “gravitomagnetic” field (i.e., the velocity-dependent component).

These equations show how the electromagnetic fields E, B and gravitoelectromagnetic fields E_g, B_g are defined by a distribution of charges (\rho or \rho_g) and their velocities (J or J_g). Having defined the fields, the force exerted on a slow test particle of charge q or mass m and velocity v is

    F &= q(E + v \times B) \\
    F_g &= m(E_g + v \times 4B_g)

After adjusting constants, the electromagnetic equations and gravitoelectromagnetic equations are almost the same as each other: the factor of 4 in the gravitomagnetic force does not go away. This may seem unimportant, but in fact arises from differences in the fundamental nature of gravity and electromagnetism, and has a crucial impact in the usefulness of gravitoelectromagnetism as a theory. (Heaviside’s 1893 formulation omits the B_g contribution to force entirely.) Recall of course that gravitoelectromagnetism is only an approximation of general relativity, which gives a fully accurate theory of gravity. When performing this approximation, the factor of 4 inevitably arises. While general relativity is rather beyond me and I have only the dimmest understanding of why there is a 4, I will present what I believe the root cause to be.

In special relativity, Maxwell’s equations can be condensed into a single equation

 \partial_\alpha F^{\alpha \beta} = \mu_0 J^\beta

where F^{\alpha \beta} is the electromagnetic tensor and contains all the information of the electric and magnetic fields E and B, and J^\beta is the four current, which has four dimensions. The time-like component of the four current is the charge \rho, and the space-like coordinates are the charge flux:

 J = (\rho, J^1, J^2, J^3)

In contrast, in general relativity the source of gravity is the stress-energy tensor (“tensor” is physicist-speak for a matrix, and a distant relative of the mathematical notion of tensor). The time-time component of the stress-energy tensor is the mass density \rho_g, the time-space components are the momentum density, and the space-space components are the momentum flux:

T = \begin{pmatrix}
\rho_g & p^1 & p^2 & p^3 \\
p^1 & T^{11} & T^{12} & T^{13} \\
p^2 & T^{21} & T^{22} & T^{23} \\
p^3 & T^{31} & T^{32} & T^{33}

The momentum flux is the “stress” part of the stress-energy tensor, with the diagonal components being the pressure, and the off-diagonal components being the shear stress. These stress terms have no analogue in electromagnetism, so if we ignore them we are left with

T \approx \begin{pmatrix}
\rho_g & p^1 & p^2 & p^3 \\
p^1 & 0 & 0 & 0 \\
p^2 & 0 & 0 & 0 \\
p^3 & 0 & 0 & 0

The momentum terms, which are labelled p^\alpha here, become the gravitomagnetic terms J_g in gravitomagnetism. However, unlike in actual electromagnetism, there are two copies of each of these terms. As a consequence, a particle with gravitoelectric charge (i.e., mass) m will effectively have a gravitomagnetic charge of 2m. Since the strength of the gravitomagnetic force is proportional to the product of the gravitomagnetic charges of the particles involved, and both of those charges are doubled, the gravitomagnetic force is 4 times stronger than the analogous magnetic force.

(This above explanation could well be wrong, it was the best I was able to infer from staring at the linked wikipedia articles.)

This distinction between electromagnetism and gravity is sometimes described by calling the electromagnetic field spin 1, meaning it is a vector field with 4^1 components, and the stress-energy tensor field spin 2, meaning it is a tensor field with 4^2 components.

Note a critical consequence of arbitrarily zero’ing out the stress terms in T: the resulting tensor field is no longer Lorentzian, so gravitoelectromagnetism is not invariant under Lorentz transformations. The choice of which terms in T are zero’d out depends on which inertial frame we measure the coordinates in – changing frames causes the terms to mix with each other. This result somewhat defeats our initial purpose in introducing gravitomagnetic terms to salvage Newtonian gravity from being non-Lorentzian. Arguably it is even “less” Lorentzian than Newtonian gravity, as 4 is further from 1 than 0. We could achieve true Lorentz invariance by replacing the factor of 4 with 1 in the gravitomagnetic force, but at the cost of no longer approximating the true gravitational force anymore.

Icelandic archaeologists rush to record ruins in path of lava

2021 April 29

Iceland has a well-recorded oral history, with a mixture of legend and verified fact, which dates back to its settlement hundreds of years before literacy reached Iceland. The recent volcanic eruption in Iceland threatened to bury in lava one such minor site, which was referred to in oral history but whose contents are still unknown. Below, an Icelandic-language article from 2021 March 20 describing the situation has been run through google translate; I have lightly cleaned up the translation in parts.

Oddgeir Isaksen, an archaeologist at Minjastofnun, is on his way by helicopter to the eruption site in Geldingadal [“castrated horse valley”] in Fagradalsfjall [“beautiful-valley mountain”], to investigate whether there are any real traces of the settler Ísólfur’s cairn. This is the last chance, as the alleged cairn is on its way under lava.

Ísólfur’s cairns at Ísólfsskáli are mentioned in old place name registers and it is said that Ísólfur wanted to be buried in the valley. “He wanted to be buried where his geldings [castrated horses] had the best”, it says.

Oddgeir says in a conversation with that these archeological remains are in great danger due to the lava flow, which is heading towards the alleged cairn. What Oddgeir wants to do is measure the cairn, photograph it and try to assess whether there really was a grave there or not.

In the picture, the cairn is marked and everything looks like it will eventually go under lava. Kristinn Magnússon

Did not get to go to Geldingadal before eruption

It is common for place names which refer to graves and are named after certain settlers to turn out to be natural creations rather than actual dungeons or burial mounds. However, it is common for [pre-Christian] graves and cairn place names to turn out to contain real relics - which could be the case here.

Judging by lava flow forecasts, the lava is heading for the cairn. “Of course it’s sad, but it just shows the importance of archeology, because there is a cultural heritage that can just disappear at any moment. This is happening in many places, for example where it disappears due to a landslide by the sea,” says Oddgeir.

During this earthquake cycle, experts from Minjastofnun have traveled around the area and listed listed monuments in the most dangerous area. Before the eruption, however, no permission was obtained to enter Geldingadal due to the danger of the eruption, so now they have to go at the last minute.

Ara Gíslason’s place name register says about the cairn in question: “On the east side of Fagradalsfjall there is a hill, called Stórhóll, just west of Nátthagaskarð. North of it is a bend and then deep valleys with lawns, called Geldingadalir. There is a thicket on the surface, and it is said that Ísólfur á Skála is buried there […] He wanted to be buried where his geldings had the best.”


2021 April 26


Talos Principle OST - When in Rome

Philip Glass - Koyaanisqatsi

Moon OST - Welcome to Lunar Industries

Ayreon - Ride the comet

Scrope v Grosvenor was a 1389 dispute between two English families using the same coat of arms, azure, a bend or (a gold band across a blue background). The motifs and traditions developed over the previous century had only recently been solidified into heraldic law; both coats of arms had already been in use for some time. 700 years later a Grosvenor descendent named their son Bendor Grosvenor in recognition of the coat of arms they had lost the authority to use. Another member of the family owned the champion racehorse Bend Or.

Tennessee was the last state to ratify the 15th amendment, which protected the right to vote from denial “on account of race, color, or previous condition of servitude”… in 1997.

Stabbed while crocheting, “unspecified” spacecraft accident, and other bizarre causes of death recently given official codes in ICD-10. The stock photo of a shocked doctor is appropriate.

The plastic number

\rho = \sqrt[3]{\frac {9 + \sqrt{69}}{18}} + \sqrt[3]{\frac {9 - \sqrt{69}}{18}} = 1.3247\ldots

is the smallest number bigger than one whose exponents get closer to being integral (in the sense that the difference between \alpha^n and its nearest integer is \ell_2); such numbers are Pisot numbers. Specifically, \rho is a root of x^3 - x - 1, so if you construct the sequence r_0 = 1, r_1 = 1, r_2 = 2 with the “sub-Fibonacci” recurrence r_n = r_{n - 2} + r_{n - 3}, then \rho^n - r_n goes to zero as n becomes large. As \rho is quite close to 1, its conjugates are also close to 1, so the convergence \rho^n - r_n \to 0 is slow, but still exponential. Surprisingly \rho is not an accumulation point of the Pisot numbers: in fact, their smallest accumulation point is the golden ratio \phi = \frac {1 + \sqrt{5}}2 = 1.618\ldots, which is a root of x^2 - x - 1 and so whose powers approach the Fibonacci sequence. The fourth-smallest Pisot number is the “supergolden ratio” \psi, a root of x^3 - x^2 - 1, again with a corresponding recurrence. The term plastic number was coined by a Dutch architect who used it in designing the proportions of a church.

Covid articles: A fascinating, in-depth breakdown of the supply chains in the manufacture of the covid vaccines gives insight into the globalization of manufacturing. A summary of a study on evolution of spike proteins in a common cold coronavirus demonstrating a tendency towards immune escape on the time scale of decades. A primer on basic facts about covid vaccines written by a doctor. A potential adverse side effect of covid vaccines: one patient developed a heart arrhythmia after receiving a covid vaccine and being struck by lightning.

A Brief, Incomplete, and Mostly Wrong History of Programming Languages

Kasparov versus the World: current world chess champion Garry Kasparov played a game of chess in 1999 against the “World Team”, whose moves were decided by majority vote from all 50 000 people who chose to participate. Each ply took 24 hours. World Team was guided by four skilled chess players, and discussion was facilitated by an online forum, bringing with it every bit as much drama and controversy that one could expect. Kasparov called it the best and most-analyzed game of chess ever played.

Svg is far more featureful and complex than one might expect… remarkably, here is a website written entirely in svg. (There’s nothing particularly of note in the content in the website, and I never would have guessed that it is secretly just an svg file.) In fact, here is the full source code of the home page:

<!DOCTYPE html>
<html lang=en-US>
      <link rel="icon" href="favicon.ico">
      <meta charset="UTF-8">
      <meta name = "description" content = "The latest news and blog posts from High Tower Games">
      <link rel="stylesheet" type="text/css" href="common/common.css">
      <title>High Tower Games ♜ News &amp; Blog ♜</title>
      <div class = "content">
         <div class = "svg_container">
            <object type="image/svg+xml" data="news.svg">


Violinist breaks a string during a concert performance, immediately switches to playing the same note on a different string and the violin gets passed down the violinist hierarchy to be restrung.

A heat pump is a device that spends energy to move heat from a cold side to a warm side: air conditioners, fridges, and freezers are all common examples of heat pumps. By simply turning an air conditioner around, one obtains a heat pump that heats a home instead of cooling it. Such consumer heat pumps tend to be use about one-fifth the energy of standard resistive heating (or natural gas heating), but are very rarely used for home heating in the US, whereas they are common in Europe and parts of Asia. Technology Connections has an introductory run down on heat pumps in a home setting, and their advantages and limitations. There is also a follow up discussing ground-sink heat pumps.

Oscillations in the frequency of the eastern US’s power grid caused by a blackout that took place in Florida.

Garson Hampfield, crossword inker.


Near the southern-most point of Iceland.

The pompadour was named after Madame de Pompadour (above), who never wore one.

An adversarial attack on an image classifier.

A vantage point tree, from here.

WHO report on the origins of covid-19

2021 April 21

Last year I read the WHO’s initial report on the covid-19 outbreak, which gave early warnings on the nature of covid and laid out specific steps to prevent it from becoming a pandemic. Only a handful of countries followed the recommendations, leading to our present situation.

Now the WHO has published another report sharing the results of their investigation into the origins of covid. I’ve read the read the report and summarized it below – although being much longer and with fewer surprises I skimmed much of it.

  1. Population-scale data shows no evidence of covid in Wuhan prior to mid-January 2020, at which point there is a sudden increase in several indicators. This is consistent with the sharp increase in confirmed cases of covid around mid-January.

  2. Retrospective analysis of medical records and biological samples discovered no new cases of covid prior to 2020 January. (None of the illnesses examined were consistent and the samples were all negative, except a few samples for which were too small to test.) Of course, some cases of covid may have left no records, and mild cases might have been filtered out in the retrospective search without careful analysis. Many early cases were connected to the Huanan market, some were connected to other markets, and for a significant number no known connection could be established.

  3. The earliest known case continues to be 2019 December 8 (onset) and was laboratory-confirmed. The first known family cluster was from 2019 December 26. There are 174 known cases of covid in Wuhan in 2019 December, most of which are laboratory-confirmed.

  4. The report mentions but has no comment on studies claiming that covid was circulating in other countries variously 2019 Sept - Dec. I have read some of these studies for myself and I continue to believe these are exceptionally unlikely to be true positives.

  5. The outbreak in Wuhan preceded the outbreak in the other parts of Hubei province.

  6. Genomic etc. analysis suggested that the most recent common ancestor of the early Wuhan cases of covid to be sequenced is 2019 December 11, with a stddev of several weeks. A wide variety of studies give various dates for the MRCA in November or December.

  7. The report is highly confident that laboratory escape is not involved in the origin of covid.

  8. Contaminated frozen goods have caused several outbreaks of covid in regions with no community transmission of covid. The report speculates that frozen goods may have played a role in the initial outbreak of covid, but does not find significant evidence in favor of that theory.

  9. On the basis of genomic comparisons, it is highly probable that sars-cov-2 passed through intermediate animal hosts between bats and humans, although there was little evidence to help identify exactly which species were involved. Direct transmission from bats is still a possibility.

MIT MYSTery Hunt puzzle stats

2021 January 23

Last weekend I participated in the 2021 MIT MYSTery hunt on the team I’m not a planet either. During the hunt, I recorded the time each puzzle was unlocked by the team and then later solved, so I’ve decided to assemble some statistics on the team’s puzzle solving. Click on the graphs to make them full screen.

Number of unlocked but not yet solved puzzles at each point of time.

In these graphs I’ve specially separated out the puzzles that were in the Infinite Corridor round (except the Infinite Corridor meta itself). This round contained hundreds of thousands of randomly generated puzzles; solving the meta required only solving specifically chosen puzzles from the round, and there was no need to unlock or solve the others once you have figured out how the meta works. In darker colors the graphs show only the ordinary puzzles, while in lighter colors are shown all puzzles including the puzzles from the infinite corridor.

Distribution of when puzzles were unlocked and solved. Smoothed with a standard deviation of one hour.

Red shows when puzzles were unlocked, and green shows when puzzles were solved. The large burst of puzzles from the infinite corridor were when the meta was being worked on. Surprisingly I don’t see a clear daily cycle in when puzzles were solved.

Distribution of times to solve each puzzle. Smoothed with a standard deviation of 20 minutes.

In total, the team unlocked 330 puzzles and solved 144; among finite puzzles, we unlocked 191 and solved 103. Below is every puzzle solved by the team, and how long they took to solve, with a resolution of about 1 minute.

Rule of Three was the fastest solved puzzle, as it was automatically solved at the same time as it was unlocked. I believe Puzzle 23 was solved immediately after unlocking it by redeeming a token earned in a different Cafe Five. While we never figured out how to solve Library of Images puzzles, we were able to backsolve several of them from information in the Infinite Corridor meta (which then allowed us to solve other Library of Images puzzles, whose answers were then used to solve the meta). As the pair working on Infinite Corridor Simulator puzzles got experienced with the necessary steps, they were ultimately able to solve them in less than three minutes each. While I’m not familiar with how ‘A Cheat Code’ was solved, it was presumably the puzzle solved from scratch in the fastest time.

Puzzle Round Solve time
Rule of Three ⊥IW.giga 0s
Puzzle 23: Cafe Five Infinite Corridor 0s
Puzzle 57: Infinite Corridor Simulator Infinite Corridor 57s
Puzzle 359: Library of Images Infinite Corridor 59s
Puzzle 427: Library of Images Infinite Corridor 1m 3s
Puzzle 53: Infinite Corridor Simulator Infinite Corridor 2m 3s
Puzzle 486: Library of Images Infinite Corridor 2m 6s
A Cheat Code Athletics 4m 13s
At A Loss For Words Students 9m 24s
Puzzle 254: Cafe Five Infinite Corridor 11m 19s
Things Students 11m 37s
Puzzle 73858: Infinite Corridor Simulator Infinite Corridor 14m 57s
Puzzle 73848: Infinite Corridor Simulator Infinite Corridor 16m 55s
⊥IWPE: Applied Sailing ⛵ Charles River 18m 43s
Crossed Out Students 20m 27s
Puzzle 73838: Infinite Corridor Simulator Infinite Corridor 21m 2s
Puzzle 73718: Infinite Corridor Simulator Infinite Corridor 23m 12s
Namesake Students 23m 53s
Triangles Students 24m 46s
Form Students 28m 3s
Doctor’s Orders Students 31m 7s
Look at What We Drew Students 32m 32s
Don’t Let Me Down Yew Labs 39m 40s
That Sinking Feeling Students 39m 44s
Common Knowledge Yew Labs 42m 6s
Enter the Perpendicular Universe Yew Labs 42m 58s
MIT Confessions Yew Labs 43m 5s
Puzzle 83: Cafe Five Infinite Corridor 44m 43s
Thank You to Our Sponsors Students 45m 26s
Not Again! Yew Labs 45m 54s
Relitasti Students 46m 31s
Stud Finder Athletics 47m 29s
Can You Deliver 60 Eggs Athletics 47m 32s
Yes or No Yew Labs 50m 31s
For Better or For Worse Yew Labs 55m 4s
Debate Quotes Students 55m 14s
Let’s Get the Ball Rolling Students 55m 40s
Better Bridges Yew Labs 55m 42s
Escape! Hayden Library Green Building 56m 8s
Unraveling the Mystery Students 58m 56s
Lime Sand Season Students 59m 7s
Who’s That Pokémon Students 1h 0m
Magic Words Students 1h 0m
Puzzle 75: Cafe Five Infinite Corridor 1h 1m
The Lobster Network Charles River 1h 4m
Just Index Yew Labs 1h 5m
Plus or Minus Students 1h 7m
Attack of the Lobsters Charles River 1h 12m
Puzzle 167: Infinite Corridor Simulator Infinite Corridor 1h 18m
Puzzle 10: Library of Images Infinite Corridor 1h 19m
Puzzle 172: Infinite Corridor Simulator Infinite Corridor 1h 22m
Puzzle 28: Cafe Five Infinite Corridor 1h 23m
Untitled Goose Goals Athletics 1h 26m
Got Milk Students 1h 27m
The Game Students 1h 31m
Puzzle 176: Infinite Corridor Simulator Infinite Corridor 1h 32m
Puzzle 182: Infinite Corridor Simulator Infinite Corridor 1h 36m
Unmasked Yew Labs 1h 37m
✏️✉️➡️3️⃣5️⃣1️⃣➖6️⃣6️⃣6️⃣➖6️⃣6️⃣5️⃣5️⃣ Yew Labs 1h 40m
Puzzle 186: Infinite Corridor Simulator Infinite Corridor 1h 40m
Beverage Blunders Students 1h 42m
Questionable Answers ⊥IW.giga 1h 42m
Puzzle 191: Infinite Corridor Simulator Infinite Corridor 1h 43m
Stay Hydrated Students 1h 46m
Puzzle 255: Library of Images Infinite Corridor 1h 49m
Puzzle 196: Infinite Corridor Simulator Infinite Corridor 1h 49m
Film Students 1h 50m
Puzzle 202: Infinite Corridor Simulator Infinite Corridor 1h 52m
So You Think You Can Count Athletics 1h 55m
Puzzle 208: Infinite Corridor Simulator Infinite Corridor 1h 55m
An American in Paris Green Building 1h 55m
Puzzle 213: Infinite Corridor Simulator Infinite Corridor 1h 58m
Boggle Battle Athletics 2h 6m
Puzzle 275: Library of Images Infinite Corridor 2h 15m
Express Yourself Students 2h 27m
Back School To Yew Labs 2h 28m
Zip it Mr. Resetti! Athletics 2h 35m
Building Hacks Athletics 2h 36m
Hey Can You Give Me A Hand With This Puzzle Yew Labs 2h 53m
Bad Air Day Green Building 3h 23m
Puzzle 46: Cafe Five Infinite Corridor 3h 27m
Violet Coins In Space Athletics 3h 38m
Title of the Puzzle Tunnels 3h 40m
Green Tee Green Building 3h 52m
Successively More Abundant in Verbiage Green Building 4h 2m
The Lexicographer Looks After His Own Students 4h 8m
Puzzle 78: Cafe Five Infinite Corridor 4h 15m
Puzzle 88: Cafe Five Infinite Corridor 4h 17m
Bake Off Athletics 4h 23m
X Marks the Spot Athletics 4h 30m
Cooperation Students 4h 38m
Musical Theatre Guild Students 4h 45m
Whodunit Students 5h 9m
⊥IWPE: Sailing (Event 1) Charles River 5h 36m
People Wear Clothes Students 5h 36m
Yonder Athletics 5h 44m
Puzzle 6: Make Your Own Word Search Infinite Corridor 5h 57m
Puzzle 19: Cafe Five Infinite Corridor 6h 7m
Help I’m Trapped in a Computer Athletics 6h 31m
Squee Squee Athletics 6h 46m
Fish Hybridization Students 7h 2m
For Your Eyes Only Tunnels 7h 46m
Puzzle 1: Cafe Five Infinite Corridor 7h 50m
Cryptic Transmission Students 7h 55m
Football Athletics 8h 16m
Puzzle 12: Cafe Five Infinite Corridor 8h 40m
When All is Lost ⊥IW.giga 9h 13m
Circular Reasoning ⊥IW.giga 9h 48m
Exactly Students 10h 11m
Tic Tac Toe Students 10h 26m
Puzzle 37: Cafe Five Infinite Corridor 10h 41m
Students 10h 54m
You Will Explode If You Stop Talking Athletics 11h 27m
MIT/⊥IW Experimental Evidence Athletics 11h 48m
Hockfield Court Yew Labs 12h 10m
Puzzle 32: Cafe Five Infinite Corridor 12h 10m
15×15 Students 12h 39m
Puzzle 42: Cafe Five Infinite Corridor 14h 17m
Heart of the Cards Students 14h 40m
Illiterate Programming Students 15h 47m
Catch and Release Athletics 15h 52m
World Fisherman Athletics 16h 27m
⊥IWPE: Archery (Event 2) 🏹 Charles River 17h 35m
Bombs Students 18h 14m
Love at 150 km/h Green Building 19h 38m
Recursion Students 20h 20m
The Emperor’s New Kitchen Stata Center 20h 58m
Paint Students 21h 38m
Water Bottle ⊥IW.giga 21h 50m
Puzzle 5: Infinite Corridor Simulator Infinite Corridor 22h 34m
Infinite Corridor Infinite Corridor 23h 29m
⊥IWPE: Fencing (Event 3) ⚔️ Charles River 27h 21m
Sketchy Site Students 29h 16m
MacGregor House Students 29h 19m
Divided is Us Athletics 30h 5m
EAsT camPUS Students 31h 3m
⊥IWPE: Pistol (Event 4) 🔫 Charles River 32h 33m
Numbers Students 33h 20m
Puzzle 2: Library of Images Infinite Corridor 36h 41m
Puzzle 4: Library of Images Infinite Corridor 37h 48m
Button Press Athletics 38h 0m
Nutrition Facts Students 38h 35m
Voltage-Controlled Green Building 43h 55m
Puzzle 65547: Cafe Five Infinite Corridor 44h 46m

At the long end, there are several events and metas which are naturally unlocked much before they can be worked on. ‘Voltage-Controlled’ was a greatly intimidating puzzle which a few people worked on occasionally for most of the hunt. ‘Water Bottle’ was a much anticipated puzzle in which we inadvertently broke the automated system, leaving the puzzle unsolvable without the organizers’ help.

As always, I greatly enjoyed mystery hunt this year (and needed a few days to get enough sleep to fully recover) and hope those who also participated found something to recollect from the weekend in the above!

DC 51

2021 January 21

“…we shall be completely disfranchised in respect to the national government, while we retain no security for participating in the formation of even the most minute local regulations by which we are to be affected. We shall be reduced to that deprecated condition of which we pathetically complained in our charges against Great Britain, of being taxed without representation.” -a letter to Congress from the residents of DC, shortly before they lost their voting rights in 1801.

“Other rights, even the most basic, are illusory if the right to vote is undermined.” -Justice Black, writing for the majority in Wesberry v Sanders, 1964.

The City of Washington was founded by US Congress on 1790 July 16, making it the first national capital created by an act of legislation. Construction of the city began in 1793, with the seat of government transferred in 1800 as most of the public buildings were sufficiently complete to be usable, although the city was described as “raw and unfinished” by president John Adams. In 1801 Congress passed an act formally organizing the District of Columbia, whose territory included Georgetown and the City of Washington, and placing it under control of Congress, thus depriving its residents of federal representation. Georgetown and the City of Washington were dissolved in 1871. DC would continue to be deprived the right to elect its own municipal government until 1973, and even today Congress still has the authority to block municipal legislation. Until 2008, the municipal government of DC was prohibited by Congress from making any expenditures related to seeking representation. DC joined the Unrepresented Nations and Peoples Organization in 2015. Today the US is the only democracy whose capital lacks representation in the legislature.

Why DC should become a state

The issue of statehood for DC is a simple question of whether DC residents should have equal rights as American citizens, or be relegated to second-class status.

At the time of DC’s founding, the future of the capital and of the country was uncertain, and the temporary loss of equal rights for a handful of residents – many of whom were elite members of the federal government – was less pressing than the wholesale creation of an administrative apparatus from scratch. Now, 200 years later, “temporary” has gone on long enough. In the intervening centuries the right of suffrage has been recognized for non-landholders, for black Americans, and for women. It is long overdue that the last few gaps in suffrage be remedied. Furthermore, it cannot escape notice that these remaining gaps disproportionately affect black Americans: in fact, DC has more black residents than 20 states.

DC has grown enormously to become not just an administrative headquarters but a large and culturally-significant city in itself, and a representative for American culture as a whole; four of the six most-visited museums in the US are in DC. DC should be the pride of the US and a symbol of American democracy, not a glaring failure to recognize equal rights.

Progress towards equality for DC residents has been slow and hard-fought. A major milestone was reached with the bipartisan passage of the 23rd amendment in 1961, which extended to DC the right to send electors to the electoral college (but not “more than the least populous State”). However, residents in DC continued to have no representation in the Senate, the House and in the process of amending the constitution. What possible justification is there for the idea that DC residents have the right to be represented in the executive branch, but not in the legislative? Where in the constitution, outside of the 23rd amendment, can be found support for US citizens having such an arbitrary subset of rights, when it so plainly guarantees equal rights before the law for all?

The lack of recognition of the right to vote is not the only inequality facing DC; DC continues to lack home rule, the authority to govern itself. In 1973, Congress finally granted DC residents permission to elect their municipal government; previously, the officials were directly appointed by Congress. In 1975 DC would elect Walter Washington as mayor, one of the first black mayors in the US. However, DC judges are still appointed directly by the president, and DC legislation must be approved by Congress before becoming law. This congressional review continues to hamstring DC governance: in recent decades, it has removed protections for gay citizens, measurably increased the spread of HIV (by perhaps 5000 cases; at the time, DC had the highest rate of HIV infection of any US city) by delaying a clean needle program for 8 years, denied access to abortion, and blocked marijuana legalization. Bills that would end progressive DC programs are regularly introduced into Congress. DC’s secondary status has hampered its covid response and access to covid relief funds. Restrictions on abortion access remain in effect today.

Rather than continuing to piecemeal restore rights to DC residents one at a time through a patchwork of legislation, resulting in a separate-but-equal doctrine echoing segregation and civil unions, there is a simple and logical way to achieve equality: statehood for DC.

Why statehood is urgent

Restoring the constitutionally guaranteed rights to a people is always an urgent matter! The best time to restore justice is yesteryear, the second best time is now.

However it doesn’t evade our attention that the present political environment makes DC statehood an especially salient matter. With multiple members of the House and Senate openly supporting fascist extremists, control of the Congress balances on the very edge of democracy. Voters, many with desperate financial needs, are depending on the Biden administration to deliver enormously on its promises in the next two years. But restoration of the civil service is a task that will take decades, and is just one step in undoing the legacy of the last four years and beginning to advance towards a more progressive society. Sadly we don’t have decades; we may only have two years, with every step forwards held hostage to a single defection within the Democratic caucus. DC statehood is necessary itself to right a prolonged wrong, but is also the greatest enabler for every other step towards justice we seek at the federal level.

Among the many urgent issues facing this incoming Congress, few are more pressing than climate change. Effective, large-scale action to address climate change is already many decades too late; the human and economic cost of another two year delay at this stage is inconceivable. Unfortunately some members of the senate Democratic caucus have their interests deeply tied to coal and will not support any action – not to mention the many senators attached to the petroleum industry. It is hard to imagine any action on climate change in the next two years with the present balance of the Senate.

While the more progressive members of the Democratic caucus in the Senate have been the most vocal in supporting DC statehood, it is the senators in less liberal states who stand to gain the most! Rather than having their vote needed for every act of progressive legislation, they will be at liberty to vote against the most contentious legislation and not endanger their seat. The sooner DC gains statehood, the sooner they benefit.

Who is in favor

President Biden stated in 2015 that he had supported DC statehood for 20 years.

President Obama, 2014 July 21:

Folks in DC pay taxes like everybody else. They contribute to the overall well-being of the country like everybody else. They should be treated like everybody else.

President Bill Clinton, 1992:

The failure to grant statehood to the men and women of the District of Columbia undercuts America’s greatest promise – that the power flows from the people and not the other way around.

Senator Hillary Clinton, 2016 May 11: (secondary reporting)

…enfranchisement isn’t solely a matter of individual rights. In the case of our nation’s capital, we have an entire populace that is routinely denied a voice in its own democracy. Washington, D.C., is home to nearly 700,000 Americans – more than the entire population of several states. Washingtonians serve in the military, serve on juries and pay taxes just like everyone else. And yet they don’t even have a vote in Congress.

In 2019 Senator Sanders and 41 Democratic senators cosponsored S.631, which would grant statehood to DC. In 1993 he spoke on the congressional floor:

How could I in good conscience say that it is appropriate for Vermont to have two seats in the Senate, which we do, to have a congressman who can vote on all of the issues, which we do, to have a governor and a state legislature which deals with all of the problems facing our people, which we do, and then say that the people of the District of Columbia, with a population larger than Vermont’s and larger than some other states should not be able to enjoy the same rights. […] This debate is about one thing and one thing alone, and that is whether the people of Washington DC are entitled to be full citizens of this country or whether they are not entitled to be full citizens.

In 1999, the American Bar Association, representing more than 400000 lawyers, passed a resolution

that citizens of the District of Columbia shall no longer be denied the fundamental right belonging to other American citizens to vote for voting members of the Congress which governs them.

In 2006 the ABA submitted a statement to Congress fully supporting restoration of voting rights to DC on the basis of the 5th amendment:

It falls to this Congress to restore the voting rights lost by a previous Congress’ omission more than 200 years ago. Not only is there a moral obligation for Congress to restore such rights, there is also a constitutional obligation for Congress to ensure the right of D.C. residents to the equal protection of the laws. […] Under Fourteenth Amendment standards, if a State legislature were to deny to residents of the state’s capital city the right to vote for members of the Legislature, it would be depriving those residents of the equal protection of the laws which is guaranteed to them by the Fourteenth Amendment. Similarly, Congress’ elimination of D.C. residents’ voting representation in the Congress by its adoption of the Organic Act of 1801, may be seen in retrospect as having deprived D.C. residents of the equal protection of the laws guaranteed to them by the Fifth Amendment due process clause.

In 2020, all 26 Democratic candidates for president interviewed by the Washington Post supported statehood for DC.

On 2020 June 26, legislation admitting DC to the US as a state passed the House with 232 votes in favor.

In a 2020 June poll, 48% of registered voters stated they supported DC statehood.

Addendum. While I was writing this, I was informed that terrorists had placed bombs in DC and invaded the capitol building in an attempted coup of the US government. At this time it appears that the security of DC was deliberately undermined by Trump, who had direct authority over the DC national guard, unlike in any state. The very delayed order to activate the DC national guard was given by Pence, not Trump. As a longer form of sabotage, several members of Congress, including Ted Cruz, Marco Rubio, and Jim Jordan, have introduced failed legislation that would remove DC gun laws. Were DC a state or had it had home rule, DC would have had the authority to adequately protect the capitol from armed insurrection.

Follow RSS/Atom feed or twitter for updates.