Friday, March 13, 2020

covid-19 digression

There's not a nation in the world where communication channels are more numerous than the US. We're a global communication pacesetter. How is it that, during a national crisis, communication is so bad here that people don't know what to believe about Covid-19 ("CV")? It seems impossible without willful meddling, and this is why people have lost trust: "Why are they f*cking with us during a life/death crisis?"

Maybe the past 35 years have already been every man for himself.

  • broadcast and internet video information sources tie viewership to profits, through ad revenues. Ads are not paragons of honesty.
  • political system requires continual fund-raising for re-election efforts. Politicians are not paragons of honesty.
  • bureaucratic offices protect agency turf or contracts and attempt to grow budgets. Bureaucratic offices are not paragons of transparent disclosure.
  • businesses, in addition to ad leverage, spend immense amounts on lobbying and electioneering. Lobbyests are not paragons of public information reliability.

But suppose we face a crisis as deadly as CV has become; don't these forces step aside and make room for the welfare of the public? Perhaps some do, in percentages. But it appears that, in the main, the clouding of public communication channels with disguised revenue or vote seeking appeals have continued. Meanwhile, look what's happened to the economy and public trust.

Is a percentage of these declines related to the forces above? If we agree that people need regular accurate information to make good decisions about their welfare then the answer must be "yes": those who don't work for, or who lack access to, organizations with inside information, are doomed to random luck with their decisions about safety and finances. How is that American?

summary of broadcast improvements needed now during CV

  1. provide government level comms to public Does a police officer responding to a call for a domestic dispute receive information from dispatch that there's an "amazing argument somewhere on the Northside"? Imagine anyone in such a safety situation making decisions with vague, sensationalized information. During times of crises, we need accurate information. Someone in CDC has projected mortality rates, best forms of self-care, proper decision trees for hospital visit, and so on. Interagency, (non-commercially) this information is readily available, undiluted. Placing such information in public hands is likely beneficial during non-crisis times, but it's absolutely critical in a pandemic.
  2. de-incentivize commercial elements during crises Would a police officer responding to a call hear it radio-ed to him with a proviso to "stay-tuned" through ads for carpeting, a cruise, and virus software? And would these ads be inserted b/c the department made money from them, regardless of the safety risk to the officer? If we must rely on commercial broadcast journalism, then the incentive to sensationalize, to keep us listening and generate more ad revenue, must be reduced or eliminated during crises.
  3. cease exploitative announcements and reports Reports filled with, eg. throw-away superlatives about the "amazing spread" of CV are more dross for the public to waste time sifting in a crisis. 1) Vagueness leads to guesswork and panic, 2) when many are losing their income, empty adjectives to compel viewers to watch past ad breaks is opportunism. Give information clearly and directly -- it's compelling enough without embellishments.
  4. interview survivors Show numbers and closed locales on slides or crawlers and get to interviewing. First person interviews with survivors are almost entirely not present in broadcasts. Without first-person interviews, the prospect of some unknown experience too dreadful to imagine is easy for the public to assume. Withholding what is natural -- first person interviews -- leaves viewers aware we're being managed by reporters instead of hearing from our fellow sufferers. We know it's widespread, where are all the interviews? (Edit 2020/04/03) Some edited accounts of survivors now on the news. Even YouTube will not show any unedited first hand accounts -- all are from news agencies reposted. Why?
  5. clarify the problem, do not cloud it Be honest. A simple, informative announcement should be broadcast, say, hourly, along the lines of the following with important dates, infection numbers, and a website.
    We're doing our best to assist with this pandemic. We do need some public cooperation. CV severity varies, but if we slow the spread, we can clean ventilators after use and treat additional patients, hopefully all the patients who might need them. Distancing is, so far, our only known weapon for slowing the spread, and is therefore extremely important. We understand the public is enduring an economic and social sacrifice to distance, but we must continue it longer to save lives. If we don't distance during peak phases of the outbreak, ventilator demand might exceed ventilator supply, leading to loss of life.

    For those few weeks' critical period which we will announce, we'd appreciate people staying away from all but necessary travel, and we might ask police to politely disperse groups of more than two persons.

1. communication exploitation examples (panic creation)

1A. broadcast exploitation

CV socio-economic effects are significant, so reporting is expected, sometimes interrupting regular programs. What most of us witness however is emergency flavored continuous coverage with multiple entangled panic producing elements. Each one is potent -- and when combined, panic is nearly certain. Thus viewers stay tuned-in, even through advertising breaks. Mission accomplished. $$$.
  1. The actual CV experience is minimized or excluded. First-person reports are anecdotal, and so might seem irrelevant, but they are exactly what's missing, as noted above. Instead reporters emote and gesticulate between themselves, creating panic.
  2. Related is how long does CV last? Will I be sick a week, a month? The regular omission of CV symptoms, timeline, and cure leads to an unsettled audience more prone to panic.
  3. Using "Covid 19" to obscure Stating the uninformative words "covid-19" over and over obscures what "ventilator shortage" means and how we might accurately frame the problem socially. The inventory of ventilators is simply less than the number patients who might eventually need one during a few days of their CV infection. Euphemizing leaves the public feeling discouraged or angry about transparency.
  4. "Social distancing": necessary, but exploited Social distancing importantly slows viral spread. Its use goes unexplained and the images are exploited to cause more fear (and thus, continual viewing). Social distancing should never be mentioned without the calming context of why its necessary: ventilator demand could be distributed over a few months instead of a few weeks. Ventilator availability (through re-use) saves lives, and the public should not be made to feel simply afraid of fellow citizens. Save lives by distributing ventilator use over time.

1B. political exploitation

There are problems also with some who create public policy or are speaking to the press. This is another contribution to the panic.
  1. CV truths are mostly excluded. We will not "defeat" CV by distancing -- CV will still exist like any other virus does. Any vaccine preventing outbreaks takes time. But this almost doesn't matter: CV is a slightly stronger flu virus, not an outlier. CV appears to fall within normal CDC flu season expectations, with normal or near-normal (20-60,000) loss of life. Still, public figures continue to proclaim our mission is to "defeat" the virus by distancing, again confusing the public about the real distancing goal, ventilator availability.
  2. public safety response By implementing first responders, public leaders have again caused public concern, perhaps inflaming panic. Bringing a heavy hand seems to validate panic, it's self-reinforcing: "if they need to bring first responders and martial law, this disease must be deadly because they need to bring first responders and martial law.". this Everyone loves overtime and hazard pay for our heroes, but can we pay for this? Involving the first responders should always be announced in concert with supporting available ventilator supplies, the reason for social distancing and enforcing social distancing.
  3. indeterminate duration some are wondering how long these restrictions will be in effect. That is, if an officer tickets me in a state park, how many more weeks does that persist? Where is an omnibus information center for these restrictions and/or closures?

2. why was panic manufactured?

As I say at the top, something stinks. We all know that's true. Certainly there is a virus spreading which needs coverage. It's flu season, after all. But why is the coverage continual, only vaguely helpful, and panic producing? This level of response doesn't correspond with the actual threat.

During manufactured panics, it's typically interesting to determine who's benefiting from the events, to what extent they are the main beneficiary, where is the money going, what are the contributing elements, and so on. This is very often impossible or nearly impossible to do, so we want to avoid wild speculation and seek educated guesses (if available).
  • what else is happening in the nation that would normally garner significant press yet is marginalized or buried during the panic. For example, what significant economic or political occurrences have we experienced in the last 2 months, and which normally would have received press, but which went nearly unreported? Were any interesting bills passed or committee reports published?
  • what are the viewership numbers (and thereby ad revenue) for press coverage during this press-induced panic?
  • what government agencies are gathering information during this quasi martial law event, and what will they share about their gathering and how it's being used? This brings to mind preparatory, exploratory, or proof-of-concept scenarios.

3. real information

3A. the experience and duration

I'm not a doctor, but the Covid-19 experience is reported to be a respiratory one with little or no GI tract issues: no diarrhea, loss of liquids or electrolytes. For most of us, the virus experience will range from unnoticeable to a coughing experience, maybe with elevated temperature and other viral flu aches, and for a week or two, as noted below. Some people, mostly elderly or other compromised respiratory, cannot tolerate the irritation in the lungs without hospitalization to increase oxygen transfer, and some in that group apparently cannot get enough oxygen to survive.

Medicalnewstoday.com :: description of symptoms

Secondary bacterial infections (eg, sinus, pneumonia) can develop during the viral phase but they outlast the viral phase and may worsen. They might require antibiotics or a ventilator, even after the viral phase.

Healthline.com :: normal flu/cold (virus) duration and risks.


3B.apparent severity

People can die from it. But this is typical during any flu season. The CDC chart of flu season effects is below.

CDC Influenza effects, yearly

Every flu season results in 20-50,000 deaths, and the same categories of people are at risk in other years as with Covid 19: those with compromised immune systems or underlying conditions decreasing ability to withstand an illness: elderly, chemotherapy, heart disease, young children.There is nothing new here. What's new is the extremely interesting response.

Sunday, March 1, 2020

toolbox - data

Statistics knowledge comes first, of course. Looking at Python apart from Jupyter, we can make some data-related assumptions about modules. Now of course, Google Colab is even easier than arranging Jupyter or Virtual Environments on my own system, so let's leave aside system setups and sandboxes which I cover in another post on environments and their variables.

Data Science, which I think of "dynamic statistics" is overtaking classical statistics. in Classical stats, we had to accurately create hypotheses before null-testing them. In dynamic statistics one must have accurate code to let the data bubble up its own conclusison. In software, Python is rapidly overtaking R, esp since Google made Colab and TensorFlow available via browsers, impossibly. On a single system, it's more complex, as noted above. For learning Python in a Data Science centric manner, practicals can be stock or derivatives market, climate, or epidemial information. To try models against wall street quants, one can play with models at quantopian.com.

Getting Started with Colab (7:17) ProgrammingKnowledge, 2020. Intro to what is essentially Jupyter notebook, cloud version, which Google now hosts.
TensorFlow in 10 minutes (9:00) edureka, 2019. Google recently began including TensorFlow into its Colab, so we now have a complete machine learning environment.
How I Would Learn Data Science (8:35) Ken Jee, 2020. Several websites and specific methods, 5 years for this guy. He emphasizes practical projects.
Practical Scraping (31:56) Computer Science, 2020. Colab project. How to work the practical on Python. He gets to his function around 15:00, prior is visualizing.

type note
Geanytext-based for nearly any code via plugins
Jupyterweb-integrated Python, designed to display output in a browser.
Eclipsejava-based, takes plugins for, eg RStudio
RStudioR-specific IDE
PSPPGNU version of SPSS. Does most. $ yay -S pspp. GUI psppire
gretleconometrics. $ yay -S gretl. GUI gretl
octaveGNU version of MATLAB $ pacman -S octave. GUI octave

Data and Statistics (code)

10 Python Tips (39:20) Corey Schafer, 2019
Python NumPy overview w/arrays (58:40) Keith Galli, 2019
Jupyter - Python Sales Analysis Example (1:26:07) Keith Galli, 2020
Pandas - Data Science Tutorial (1:00:27) Keith Galli, 2018 CSV reading, beautiful soup
Python Stock Prediction (49:48) Computer Science, 2020
Beautiful Soup stock prices (10:47) straight_code, 2019
Options analysis in Python (1:02:22) Engineers.SG, 2016 Black-Scholes (emotional volatility) in Pandas.
Derivative analytics in Python (1:29:27) O'Reilly, 2014 Data frames and Monte Carlo (brownian).

Data and Statistics (classic)

Combinations vs. Permutations (20:59) Brandon Foltz, 2012 For either finite math or stats
Linear Regression Playlist (multiple) Brandon Foltz, 2013
Covariance basics (26:22) Brandon Foltz, 2013 stock examples, vs correlation/linear regression
Pandas - Data Science Tutorial (1:00:27) Keith Galli, 2018 CSV reading, beautiful soup
Python Stock Prediction (49:48) Computer Science, 2020