Notes from 9/3–16/3

The world is being overwhelmed by COVID-19.

I try to take the better articles and read them to understand how people think about such situations.

The Psychology of Human Misjudgment, by Charlie Munger

24 Standard Causes of Human Misjudgment

1. First. Under recognition of the power of what psychologists call reinforcement and economists call incentives
2. My second factor is simple psychological denial
3. Third. Incentive-cause bias
4. Fourth, and this is a superpower in error-causing psychological tendency, bias from consistency and commitment tendency, including the tendency to avoid or promptly resolve cognitive dissonance
5. And all these things like painful qualifying and initiation rituals, all those things, pound in your commitments and your ideas
6. Sixth. Bias from Pavlovian association, misconstruing past correlation as a reliable basis for decision-making
7. Seventh. Bias from reciprocation tendency, including the tendency of one on a roll to act as other persons expect
8. Eight. Now, this is a lollapalooza, and Henry Kaufman wisely talked about this, bias from over-influence by social proof, that is, the conclusions of others, particularly under conditions of natural uncertainty and stress.
9. Nine. Bias from contrast caused distortions of sensation, perception, and cognition.
10. Bias from over-influence by authority
11. Bias from Deprival Super Reaction Syndrome, including bias caused by present or threatened scarcity, including threatened removal of something almost possessed but never possessed. Here I took the Munger dog, a lovely harmless dog.
12. Bias from envy/jealousy
13. Bias from chemical dependency
14. Bias from gambling compulsion
15. Bias from liking distortion, including the tendency to especially like oneself, one’s own kind, and one’s own idea structures, and the tendency to be especially susceptible to being misled by someone liked.
16. incentive caused bias. His professional reputation is all tied up with what he knows.
17. Bias from the non-mathematical nature of the human brain in its natural state as it deals with probabilities employing crude heuristics and is often mislead by mere contrast
18. Or if something is very vivid, which I’m going to come to next, that will really pound in.
19. Now we come bias from over-influence by extra vivid evidence
20. Stress-induced mental changes, small and large, temporary and permanent
21. the tendency to lose ability through disuse. Then I’ve got mental and organizational confusion from the say-something syndrome

This is the most important question in this whole talk. What happens when these standard psychological tendencies combine?

The second question. Isn’t this list of standard psychological tendencies improperly tautological compared with the system of Euclid? That is, aren’t there overlaps, and can’t some items on the list be derived from combinations of other items? The answer to that is, plainly, yes

Three. What good, in the practical world, is the thought system indicated by the list?

Fourth, what special knowledge problems lie buried in the thought system indicated by the list?

COVID-19

Computer Scientists Are Building Algorithms to Tackle COVID-19

Lung Infection Quantification of COVID-19 in CT Images with Deep Learning

Shanghai researchers have devised a system that, alongside a human checking the results, could reduce the analysis time of a CT image from hours down to about four minutes.

Rapid AI Development Cycle for the Coronavirus (COVID-19) Pandemic: Initial Results for Automated Detection & Patient Monitoring using Deep Learning CT Image Analysis

This paper also claims to detect the presence of the COVID-19, but also visualizes the virus’s effects on the lungs to track the progress of the illness over time.

Abnormal respiratory patterns classifier may contribute to large-scale screening of people infected with COVID-19 in an accurate and unobtrusive manner

Researchers here look for an auditory way of screening for COVID-19 by analyzing how fast a person is breathing. The research isn’t conclusive, but it’s a new idea for a less invasive way of testing for the virus.

Deep Learning System to Screen Coronavirus Disease 2019 Pneumonia

This work tries to differentiate the pneumonia suffered by patients with COVID-19 from the garden-variety flu.

Prediction of criticality in patients with severe Covid-19 infection using three clinical features: a machine learning-based prognostic model with clinical data in Wuhan

Using nearly 3,000 electronic health records from patients in Wuhan, China, researchers built an algorithm that could predict the rate of mortality for patients with more than 90% accuracy.

#78 Balaji Srinivasan: Exploring COVID-19

Very much a lot of basic facts.

LessWrong Coronavirus Link Database

Brilliant!

Coronavirus: Why You Must Act Now

The article is super long as it goes into each individual country’s situation

Don’t “Flatten the Curve”, stop it!

Thought there was something wrong with this analogy. Yes you can’t just flatten this curve. Spreading out cases over a long time means it will last a decade! Also how do you even know how many cases there will be in absolute terms?

Why outbreaks like coronavirus spread exponentially, and how to ‘flatten the curve’

Simple javascript infection simulator.

What US Government should do regarding Covid-19

Immediate basics

• Figure out who’s doing good work today. (UW virology lab; Stanford testing facility; etc.) Call them; learn what they need; solve those problems. (E.g. shortage of reagents.)
• Push very hard on social distancing now. Close all non-essential commercial establishments for some period of time.

But who wrote this article?

Taiwan has millions of visitors from China and only 45 coronavirus cases. Here’s how.

BE PROACTIVE

Coronavirus: The Black Swan of 2020

We suggest you question every assumption about your business, including:

1. Cash runway. Do you really have as much runway as you think? Could you withstand a few poor quarters if the economy sputters? Have you made contingency plans? Where could you trim expenses without fundamentally hurting the business? Ask these questions now to avoid potentially painful future consequences.
2. Fundraising. Private financings could soften significantly, as happened in 2001 and 2009. What would you do if fundraising on attractive terms proves difficult in 2020 and 2021? Could you turn a challenging situation into an opportunity to set yourself up for enduring success? Many of the most iconic companies were forged and shaped during difficult times. We partnered with Cisco shortly after Black Monday in 1987. Google and PayPal soldiered through the aftermath of the dot-com bust. More recently, Airbnb, Square, and Stripe were founded in the midst of the Global Financial Crisis. Constraints focus the mind and provide fertile ground for creativity.
3. Sales forecasts. Even if you don’t see any direct or immediate exposure for your company, anticipate that your customers may revise their spending habits. Deals that seemed certain may not close. The key is to not be caught flat-footed.
4. Marketing. With softening sales, you might find that your customer lifetime values have declined, in turn suggesting the need to rein in customer acquisition spending to maintain consistent returns on marketing spending. With greater economic and fundraising uncertainty, you might even want to consider raising the bar on ROI for marketing spend.
5. Headcount. Given all of the above stress points on your finances, this might be a time to evaluate critically whether you can do more with less and raise productivity.
6. Capital spending. Until you have charted a course to financial independence, examine whether your capital spending plans are sensible in a more uncertain environment. Perhaps there is no reason to change plans and, for all you know, changing circumstances may even present opportunities to accelerate. But these are decisions that should be deliberate.

AI

The Annotated GPT-2

Brilliantly done. doubt people will read the entire thing but THIS IS A MODEL FOR COURSES

On Semantic Search

Doesn’t give me something I can use. Provides a general overview of what semantic search is like: synonym generation, query autocompletion, alternative query generation,embeddings, contextualisation based on user history/session history, learning to rank (nextsentenceprediction on BERT), ensemble, multilingual search

Uber Open Sources Manifold to Visually Debug Machine Learning Programs

It looks amazing, would love to try this out over the weekend.

Data

Cultivating Algos, Stitchfix

made with
GPU.js and D3.js

Nicely done. Turns out, they’re a fashion company. Did not expect this at all