johbjo

The problem (pointed to in many comments) seems to be confusion about roles.

Economics is a form of science. Maybe not a hard science, but there are honest ambitions to understand the world. Much of it is historical, since it is based on narratives that are taken to explain observed phenomena. The models are then vague and unprecise.

However, government officials and their experts are not pursuing science. They are literally trying to change the state of a system by intervening. The use of academic language is more of an instrument in this, rather than anything to do with science.

I even find the phrasing "the economics profession" (referring to government officials) problematic since obscures policy making in language that sounds scientific.

It goes without saying that if you know (or guess) the model or function you are trying to implement, then you don't need machine learning.

Toy problems like this are still interesting because they demonstrate techniques that are applicable to bigger problems. I'm guessing the network in the car game doesn't need more than one layer and two outputs (acceleration vector), but that is besides the point.

The technique being demonstrated is the ability of GA (or maybe particle filters), to find "optimal" weights for a network given whatever simulator. This is always interesting, especially when done with graphics like this.

Installing rack servers and setting up services to run a site used to be a sort of rite-of-passage 15-20 years ago, but that time period of the web was different. Still, I would consider basic familiarity with the infrastructure necessary also today.

Increasing hardware performance relative to task load created the rational for virtualization. Virtualization also turned out to be rational with respect to consistency, convenience, maintenance, and so on. At that point, outsourcing to a cloud can be rational.

But fewer people get hands-on experience with the infrastructure, and it sounds like many consider it almost mythical. For example, realizing the amount of work that can be done in 4U today. What does amazon charge for 96 cores and 256GB?

No, the FT article is referring to excess deaths during the outbreaks, which of course is higher than the average for those months.

Mortality for the year in total is on average.

The graph in the link is badly chosen to demonstrate this, but there is no excess mortality during january-september relative to previous years.

It is clear from monthly data that from march to may there was excess mortality. From june to september the mortality was lower than previous years.

The main effect of covid in mortality statistics is that it shortens lives by about 6 to 12 months, which causes a bump in mortality followed by a slump.

Asset inflation is not the same as wage inflation.

Consider the relative size of new mortgages to wages. Since asset prices are typically leveraged on wages (through mortgages), assets will inflate faster than wages (unless the interest rate is set to counteract this.)

But inflation is only beneficial to /new/ buyers if wages inflate faster than assets. Otherwise the relative debt of all mortgage holders increases over time.

Whether asset inflation is beneficial to a home owner is debatable, since all properties appreciate at the same rate. "Moving up" still means increasing the relative size of the mortgage.

You're assuming inflation in wages, but not in prices. And there are relative differences in price inflation. Nowadays, assets typically inflate first, then wages and consumer prices. Prices can move faster than wages, so wages would inflate/deflate last.

So you have to imagine that your wage is fixed for the year, while your prices change every month or week.

The crazy idea behind Google was that they had no feasible plan for monetization.

There was an early and obvious insight that fast and simple search would win, but ads at the time were gif banners and flash animation.

They tried adwords, and it worked.

No. The reason there are few FANGS is that they are platform businesses, and the value of their services is proportional (or more?) to the size of their user bases.

This means winner-takes-all. There were others, but they did not grow users as fast, so they lost.

Another thing to point out is that while search and ads is Google's revenue, with chrome, gmail, android, etc they are achieving a form of lock-in. If anyone else "owned" these surfaces, replacing their search box would be much easier.

The spread risk investment strategy is also correct, since the value of any platform market is low before the winner emerges. After the winner emerges, the value of that platform is the value of the entire market.

Governments have the right to print money.

The reason it hurts the middle class is that the size of their mortgages are the first to be affected by inflation, whereas their wages are among the last.

Mortgage holders are stakeholders in inflation, hoping for appreciating property prices.

A zero-inflation policy would correspond to matching the money supply growth to productivity growth. This would, ideally, result in approximately stable prices. Relative prices would still vary. A more efficient candy-bar machine would lower prices of candy-bars, whereas hourly wages would be unchanged.

Defacto, there is a policy to aim for consumer price inflation of around 2%.

Many justifications can be thought of, but one is the tendency that prices and wages rarely adjust downwards. So instead of expecting the economy to lower prices due to productivity increases, the central banks depreciate the currency. This means flat wages in stagnating industries, instead of falling wages. One could argue there is some psychological benefit to this.

Another justification is because cash savings depreciate, there is need to invest them. In a textbook world, savings become investments, which creates growth.

One typical scenario; (rich people) take loans to buy assets. The loan itself causes inflation, and asset prices will rise. Then wages and cash savings have less relative value. Wages won't be affected until the firms can increase prices, which might happen much later.

How do you define inflation, and what benefits do you see with inflation free money?

Inflation itself is a policy, intended to generate incentives. In addition, the central bank mechanisms for inflation are themselves "biased" in various ways.

Mechanically, it would be trivial to have a zero-inflation policy, but there are just too many stakeholders for it to be possible.

There were rural regions in Sweden with no deaths and where the economy was essentially uninterrupted. The urban regions, however, were already in the "exponential wall" when covid became a news story.

The difference with Norway, Denmark and Finland is that their urban regions did not have early outbreaks.

"Dry tinder" hypothesis: the massive (but hidden) early spread exposed vulnerable people, whereas now they are protecting themselves and possibly receiving smaller inoculations.

The magnitude of the early spread in cities like NY is difficult to estimate. Antibodies does not tell the full story.

Know a guy who introduced SharedLaTeX at a security consulting firm. Apparently they all liked it.

The LyX editor for math is ridiculously competent. Even using plain LaTeX, LyX can be used just for the maths because it is so quick. I can't imagine the math for which LyX would be insufficient.

It doesn't render like tex in the editor, but it is logically correct.

LyX has a "tag" for plain tex, which allows you to insert any tex file into your document. That is a practical way of working with TikZ, for example.

Kind of. Pronouncing a_b as "a wrt b" should work though.

With f_uk, it could mean a model with parameters trained for UK - "theta uk".

In my view, it's nice to have subscript-like things in the constructors. Ex:

uk_price_model = PriceModel(uk_model_parameters)

If weaknesses can be known in advance, then they can "just" be incorporated into the training environment. Unless you can feed the sensors "false" data, there aren't necessarily that many possible exploits.

a_b in maths is pronounced "a of b" in English, and is often intended to be equivalent to a(b).

Superscripts in maths is closer to concatenated class names, as it separates it from being a function of something.