erichahn

It's boring but usually everything is well-defined and hence well-understandable.

Not the case for most CS papers.

Wittgenstein had this effect on me too. But not the Tractatus. What calmed me down was the Philosophical Investigations and Kripkes Naming and Necessity.

It is both. A stimulant and a sedative (dumb categories).

Withdraw shrinks in only when you use tabacco regularly.

It is not the drug that is the problem but how you used / abused it.

Are they really doing this?

Mass anxiety is not a conspiracy theory but rather a measurable social effect. See social cooling.

I didn't say there is somebody controlling or forcing the population into technocracy. What I meant is that it is inevitable to some degree that tech will continue to shape our society in unexpected and unintended ways.

Even if I did. Nothing is wrong with good conspiracy theories. Sometimes they turn out to be true.

Tabacco is a magical plant that will grant you increased alertness, stress and anxiety release, decreased hunger, ...

Our problem is not smoking. Our problem is how and why we smoke. It has nothing to do with the drug itself but everything to do with us.

In germany at least... Maybe it is very different elsewhere... I hope so...

The same baseline problem holds for hospitalizations as well. Nobody knows how many hospital beds were occupied in the last 10 years. Evidence points to a strong decrease in hospital capacity in germany at least in the last 20 years. Talking about "collapse of hospital system" is very bad in my opinion. No data shows that this actually happened in the EU at least. Also keep in mind that there is always (year does not matter) some place on earth where a hospital may collapse. So bergamo (italy) for example is just anecdotal evidence.

On possibly reduced mortality by lockdown : OK, fair enough. I think you can see it in decreased car accidents in germany and also in seemingly low influenza associated deaths. But my theory is that influenza and sars2 are evolutionary competing for vulnerable individuals and sars2 seems to be winning right now. I saw a nice study showing this for several flu-like viruses a couple of years ago. But getting enough and clean data for problems like this is very hard.

I think the lockdown has almost nothing to do with the virus but is a strange reaction to increased pressure of globalization and digitalization on the societies as well as a global economic crisis. Maybe lockdown policies will stay for a long time. But not to protect us from a dangerous virus but to ease some kind of mass anxiety and to make us ready for a new kind of governance focused on ai, big data and techocracy.

1: This has nothing to do with excess mortality per se but rather with proper data analysis.

2-3: That's exactly what I mean but you should not measure excess mortality as absolute deaths / million people but you should adjust for demographics. Because if you have older population, you will have more expected deaths. And if you adjust for age the excess mortality vanishes for germany.

4: Yes, it is talked about but not to the extend that would be necessary. In reality most people experience the various negative consequences of the lockdown policies but very few the negative consequences of the virus.

5: This is certainly debatable. If you compare this to the HIV-pandemic (still ~1mio Deaths / yr), then Covid has a factor of 10^5 more public attention (but 2x the death toll). A similar number of people die of tuberculosis. Nobody knows how many people die of influenza or coronavirus each year. We suspect there are 100 Mio dengue virus infections each year. Nobody knows what a once-in-a-century pandemic is. It seems to be a matter of public attention and newness instead of careful data analysis.

Sadly the lies, skewed narratives and new kinds of ideologies are abundant in media, politics and counter-culture this day.

Here are a couple of key concerns that I have with the data and the anti-covid measures : 1) There is no baseline. How many people die with a postive PCR-test for influenza or other coronaviruses each year? Did we ever test millions of sick people? 2) How did the morbidity actually change after the spread of Sars-CoV-2? If you want to measure this you need to compute some age-adjusted probability that people will die in a given year. Data from germany suggests that although ~100k people died, the probability to die didn't change much or at least changed around a typical baseline for the last 10 years. 3) If age-adjusted morbidity didn't change then PCR-tests measure natural morbidity to a large extend. 4) Nobody is talking about the negative impact of the lockdown. To name a few : less births in several countries in the EU, inflation, +60% cases of anorexia in german youth, +10% increased tabacco usage in germany, no proper schooling for 2 years, social cooling in general, ... I could go on endlessly. 5) The language changed. There are hundrets of new words invented around a common frame of reality in the last year. Everybody uses these words that nobody would understand nor use <=2019. This is very worrysome in my opinion because it shows a radical change in ideology of the population that goes way beyond a pandemic and shows changes in mass psychology.

whats wrong with battery factory

Baader-Meinhof bias

Why not as its own file in the directory of the source code? Or in another directory...

No history actually tells you that only 1% of produced maths is useful. You only see the good stuff not the ugly stuff.

I think this is a good sign honestly. Historically many innovations came from military research.

Can you provide a paper or book on this?

I never said it is a paradox. I never said that PA or Q is inconsistent. I said that they cannot prove their own consistency by Gödels theorem. Hence we don't know if Gödels theorem was formalized in an inconsistent system.

Honestly it would be weird if the incompleteness theorems don't apply to themselves.

I think if Gödels proof needs Q then that is OK.

Q cannot prove its own consistency. Which means there is no way of telling that Gödels theorems are proved in a theory that is inconsistent (where everything is true).

"it does not rely on any particular axiomatisation"

-- OK, what does it rely on then?

Yes and then you can do self-referential statements again which leads you to the conclusion that Gödels theorems are provable only in a system that cannot prove its own consistency.

If you figure this out correctly you can claim the third incompleteness theorem. (If I am right)

Which one? I don't think so tbh because his proof uses PA.

What most ppl don't notice is that Gödels incompleteness theorems are themselves expressible only in a logic system capable of expressing Peano arithmetic. Now this means that they apply to themselves which means that we cannot know if they are made up from a axiomatic system that can prove anything.

How is entropy related to "male gaze". This approach seems to be unsupervised, I don't see the problem.

No, this does not solve the problem that he describes in the article. You can have a great crossvalidation score and still struggle on unseen data if the data is relatively dissimilar from your train set. Like X-Ray scans produced from a different machine. There are numerous other examples. CNNs on images for example are famously known to disintegrate on images + white noise (which look the same to a human).

"Denise Herzing, the founder and research director of the nonprofit Wild Dolphin Project, which studies dolphins in the Atlantic, is now using similar algorithms, coupled with underwater keyboards and computers, to try to decode dolphin"

- Why does she need underwater computers?

I mean yes, there is parametric ML (maximum likelihood, MAP, GMMs, ...) and there is non-parametric ML (everything neural network, SVM, GBM, random forrests, ...).

I'd argue that the latter had bigger success in the past since the prior on the data distribution is usually wrong in real life. Think about a prior for image data distributions or the same in nlp. Forget about it.

Isn't the point of ML exactly that you don't know the underlying distribution? How is this ever assumed in any way? ML is not parametric statistics.

Bad rules. My suggestions ask yourself the following all the time 1) Do I ~really~ understand what the definition / theorem is supposed to tell me. 2) A proof is just a reason why something is true in maths. Do you understand ~why~ a statement is true? 3) Exercise all the time. You won't learn math by memorizing definitions, theorems and proofs.