nhaliday : estimate   124

 « earlier
exponential function - Feynman's Trick for Approximating $e^x$ - Mathematics Stack Exchange
1. e^2.3 ~ 10
2. e^.7 ~ 2
3. e^x ~ 1+x
e = 2.71828...

errors (absolute, relative):
1. +0.0258, 0.26%
2. -0.0138, -0.68%
3. 1 + x approximates e^x on [-.3, .3] with absolute error < .05, and relative error < 5.6% (3.7% for [0, .3]).
nibble  q-n-a  overflow  math  feynman  giants  mental-math  calculation  multiplicative  AMT  identity  objektbuch  explanation  howto  estimate  street-fighting  stories  approximation  data  trivia  nitty-gritty
october 2019 by nhaliday
CakeML
some interesting job openings in Sydney listed here
programming  pls  plt  functional  ocaml-sml  formal-methods  rigor  compilers  types  numerics  accuracy  estimate  research-program  homepage  anglo  jobs  tech  cool
august 2019 by nhaliday
Anti-hash test. - Codeforces
- Thue-Morse sequence
In general, polynomial string hashing is a useful technique in construction of efficient string algorithms. One simply needs to remember to carefully select the modulus M and the variable of the polynomial p depending on the application. A good rule of thumb is to pick both values as prime numbers with M as large as possible so that no integer overflow occurs and p being at least the size of the alphabet.
2.2. Upper Bound on M
[stuff about 32- and 64-bit integers]
2.3. Lower Bound on M
On the other side Mis bounded due to the well-known birthday paradox: if we consider a collection of m keys with m ≥ 1.2√M then the chance of a collision to occur within this collection is at least 50% (assuming that the distribution of fingerprints is close to uniform on the set of all strings). Thus if the birthday paradox applies then one needs to choose M=ω(m^2)to have a fair chance to avoid a collision. However, one should note that not always the birthday paradox applies. As a benchmark consider the following two problems.

I generally prefer to use Schwartz-Zippel to reason about collision probabilities w/ this kind of thing, eg, https://people.eecs.berkeley.edu/~sinclair/cs271/n3.pdf.

A good way to get more accurate results: just use multiple primes and the Chinese remainder theorem to get as large an M as you need w/o going beyond 64-bit arithmetic.

more on this: https://codeforces.com/blog/entry/60442
oly  oly-programming  gotchas  howto  hashing  algorithms  strings  random  best-practices  counterexample  multi  pdf  papers  nibble  examples  fields  polynomials  lecture-notes  yoga  probability  estimate  magnitude  hacker  adversarial  CAS  lattice  discrete
august 2019 by nhaliday
Laurence Tratt: What Challenges and Trade-Offs do Optimising Compilers Face?
Summary
It's important to be realistic: most people don't care about program performance most of the time. Modern computers are so fast that most programs run fast enough even with very slow language implementations. In that sense, I agree with Daniel's premise: optimising compilers are often unimportant. But “often” is often unsatisfying, as it is here. Users find themselves transitioning from not caring at all about performance to suddenly really caring, often in the space of a single day.

This, to me, is where optimising compilers come into their own: they mean that even fewer people need care about program performance. And I don't mean that they get us from, say, 98 to 99 people out of 100 not needing to care: it's probably more like going from 80 to 99 people out of 100 not needing to care. This is, I suspect, more significant than it seems: it means that many people can go through an entire career without worrying about performance. Martin Berger reminded me of A N Whitehead’s wonderful line that “civilization advances by extending the number of important operations which we can perform without thinking about them” and this seems a classic example of that at work. Even better, optimising compilers are widely tested and thus generally much more reliable than the equivalent optimisations performed manually.

But I think that those of us who work on optimising compilers need to be honest with ourselves, and with users, about what performance improvement one can expect to see on a typical program. We have a tendency to pick the maximum possible improvement and talk about it as if it's the mean, when there's often a huge difference between the two. There are many good reasons for that gap, and I hope in this blog post I've at least made you think about some of the challenges and trade-offs that optimising compilers are subject to.

[1]
Most readers will be familiar with Knuth’s quip that “premature optimisation is the root of all evil.” However, I doubt that any of us have any real idea what proportion of time is spent in the average part of the average program. In such cases, I tend to assume that Pareto’s principle won't be far too wrong (i.e. that 80% of execution time is spent in 20% of code). In 1971 a study by Knuth and others of Fortran programs, found that 50% of execution time was spent in 4% of code. I don't know of modern equivalents of this study, and for them to be truly useful, they'd have to be rather big. If anyone knows of something along these lines, please let me know!
techtariat  programming  compilers  performance  tradeoffs  cost-benefit  engineering  yak-shaving  pareto  plt  c(pp)  rust  golang  trivia  data  objektbuch  street-fighting  estimate  distribution  pro-rata
july 2019 by nhaliday
The Existential Risk of Math Errors - Gwern.net
How big is this upper bound? Mathematicians have often made errors in proofs. But it’s rarer for ideas to be accepted for a long time and then rejected. But we can divide errors into 2 basic cases corresponding to type I and type II errors:

1. Mistakes where the theorem is still true, but the proof was incorrect (type I)
2. Mistakes where the theorem was false, and the proof was also necessarily incorrect (type II)

Before someone comes up with a final answer, a mathematician may have many levels of intuition in formulating & working on the problem, but we’ll consider the final end-product where the mathematician feels satisfied that he has solved it. Case 1 is perhaps the most common case, with innumerable examples; this is sometimes due to mistakes in the proof that anyone would accept is a mistake, but many of these cases are due to changing standards of proof. For example, when David Hilbert discovered errors in Euclid’s proofs which no one noticed before, the theorems were still true, and the gaps more due to Hilbert being a modern mathematician thinking in terms of formal systems (which of course Euclid did not think in). (David Hilbert himself turns out to be a useful example of the other kind of error: his famous list of 23 problems was accompanied by definite opinions on the outcome of each problem and sometimes timings, several of which were wrong or questionable5.) Similarly, early calculus used ‘infinitesimals’ which were sometimes treated as being 0 and sometimes treated as an indefinitely small non-zero number; this was incoherent and strictly speaking, practically all of the calculus results were wrong because they relied on an incoherent concept - but of course the results were some of the greatest mathematical work ever conducted6 and when later mathematicians put calculus on a more rigorous footing, they immediately re-derived those results (sometimes with important qualifications), and doubtless as modern math evolves other fields have sometimes needed to go back and clean up the foundations and will in the future.7

...

Isaac Newton, incidentally, gave two proofs of the same solution to a problem in probability, one via enumeration and the other more abstract; the enumeration was correct, but the other proof totally wrong and this was not noticed for a long time, leading Stigler to remark:

...

TYPE I > TYPE II?
“Lefschetz was a purely intuitive mathematician. It was said of him that he had never given a completely correct proof, but had never made a wrong guess either.”
- Gian-Carlo Rota13

Case 2 is disturbing, since it is a case in which we wind up with false beliefs and also false beliefs about our beliefs (we no longer know that we don’t know). Case 2 could lead to extinction.

...

Except, errors do not seem to be evenly & randomly distributed between case 1 and case 2. There seem to be far more case 1s than case 2s, as already mentioned in the early calculus example: far more than 50% of the early calculus results were correct when checked more rigorously. Richard Hamming attributes to Ralph Boas a comment that while editing Mathematical Reviews that “of the new results in the papers reviewed most are true but the corresponding proofs are perhaps half the time plain wrong”.

...

Gian-Carlo Rota gives us an example with Hilbert:

...

Olga labored for three years; it turned out that all mistakes could be corrected without any major changes in the statement of the theorems. There was one exception, a paper Hilbert wrote in his old age, which could not be fixed; it was a purported proof of the continuum hypothesis, you will find it in a volume of the Mathematische Annalen of the early thirties.

...

Leslie Lamport advocates for machine-checked proofs and a more rigorous style of proofs similar to natural deduction, noting a mathematician acquaintance guesses at a broad error rate of 1/329 and that he routinely found mistakes in his own proofs and, worse, believed false conjectures30.

[more on these "structured proofs":
https://mathoverflow.net/questions/35727/community-experiences-writing-lamports-structured-proofs
]

We can probably add software to that list: early software engineering work found that, dismayingly, bug rates seem to be simply a function of lines of code, and one would expect diseconomies of scale. So one would expect that in going from the ~4,000 lines of code of the Microsoft DOS operating system kernel to the ~50,000,000 lines of code in Windows Server 2003 (with full systems of applications and libraries being even larger: the comprehensive Debian repository in 2007 contained ~323,551,126 lines of code) that the number of active bugs at any time would be… fairly large. Mathematical software is hopefully better, but practitioners still run into issues (eg Durán et al 2014, Fonseca et al 2017) and I don’t know of any research pinning down how buggy key mathematical systems like Mathematica are or how much published mathematics may be erroneous due to bugs. This general problem led to predictions of doom and spurred much research into automated proof-checking, static analysis, and functional languages31.

[related:
https://mathoverflow.net/questions/11517/computer-algebra-errors
I don't know any interesting bugs in symbolic algebra packages but I know a true, enlightening and entertaining story about something that looked like a bug but wasn't.

Define sinc𝑥=(sin𝑥)/𝑥.

Someone found the following result in an algebra package: ∫∞0𝑑𝑥sinc𝑥=𝜋/2
They then found the following results:

...

So of course when they got:

∫∞0𝑑𝑥sinc𝑥sinc(𝑥/3)sinc(𝑥/5)⋯sinc(𝑥/15)=(467807924713440738696537864469/935615849440640907310521750000)𝜋

hmm:
Which means that nobody knows Fourier analysis nowdays. Very sad and discouraging story... – fedja Jan 29 '10 at 18:47

--

Because the most popular systems are all commercial, they tend to guard their bug database rather closely -- making them public would seriously cut their sales. For example, for the open source project Sage (which is quite young), you can get a list of all the known bugs from this page. 1582 known issues on Feb.16th 2010 (which includes feature requests, problems with documentation, etc).

That is an order of magnitude less than the commercial systems. And it's not because it is better, it is because it is younger and smaller. It might be better, but until SAGE does a lot of analysis (about 40% of CAS bugs are there) and a fancy user interface (another 40%), it is too hard to compare.

I once ran a graduate course whose core topic was studying the fundamental disconnect between the algebraic nature of CAS and the analytic nature of the what it is mostly used for. There are issues of logic -- CASes work more or less in an intensional logic, while most of analysis is stated in a purely extensional fashion. There is no well-defined 'denotational semantics' for expressions-as-functions, which strongly contributes to the deeper bugs in CASes.]

...

Should such widely-believed conjectures as P≠NP or the Riemann hypothesis turn out be false, then because they are assumed by so many existing proofs, a far larger math holocaust would ensue38 - and our previous estimates of error rates will turn out to have been substantial underestimates. But it may be a cloud with a silver lining, if it doesn’t come at a time of danger.

https://mathoverflow.net/questions/338607/why-doesnt-mathematics-collapse-down-even-though-humans-quite-often-make-mista

more on formal methods in programming:
https://www.quantamagazine.org/formal-verification-creates-hacker-proof-code-20160920/
https://intelligence.org/2014/03/02/bob-constable/

Update: measured effort
In the October 2018 issue of Communications of the ACM there is an interesting article about Formally verified software in the real world with some estimates of the effort.

Interestingly (based on OS development for military equipment), it seems that producing formally proved software requires 3.3 times more effort than with traditional engineering techniques. So it's really costly.

On the other hand, it requires 2.3 times less effort to get high security software this way than with traditionally engineered software if you add the effort to make such software certified at a high security level (EAL 7). So if you have high reliability or security requirements there is definitively a business case for going formal.

WHY DON'T PEOPLE USE FORMAL METHODS?: https://www.hillelwayne.com/post/why-dont-people-use-formal-methods/
You can see examples of how all of these look at Let’s Prove Leftpad. HOL4 and Isabelle are good examples of “independent theorem” specs, SPARK and Dafny have “embedded assertion” specs, and Coq and Agda have “dependent type” specs.6

If you squint a bit it looks like these three forms of code spec map to the three main domains of automated correctness checking: tests, contracts, and types. This is not a coincidence. Correctness is a spectrum, and formal verification is one extreme of that spectrum. As we reduce the rigour (and effort) of our verification we get simpler and narrower checks, whether that means limiting the explored state space, using weaker types, or pushing verification to the runtime. Any means of total specification then becomes a means of partial specification, and vice versa: many consider Cleanroom a formal verification technique, which primarily works by pushing code review far beyond what’s humanly possible.

...

The question, then: “is 90/95/99% correct significantly cheaper than 100% correct?” The answer is very yes. We all are comfortable saying that a codebase we’ve well-tested and well-typed is mostly correct modulo a few fixes in prod, and we’re even writing more than four lines of code a day. In fact, the vast… [more]
ratty  gwern  analysis  essay  realness  truth  correctness  reason  philosophy  math  proofs  formal-methods  cs  programming  engineering  worse-is-better/the-right-thing  intuition  giants  old-anglo  error  street-fighting  heuristic  zooming  risk  threat-modeling  software  lens  logic  inference  physics  differential  geometry  estimate  distribution  robust  speculation  nonlinearity  cost-benefit  convexity-curvature  measure  scale  trivia  cocktail  history  early-modern  europe  math.CA  rigor  news  org:mag  org:sci  miri-cfar  pdf  thesis  comparison  examples  org:junk  q-n-a  stackex  pragmatic  tradeoffs  cracker-prog  techtariat  invariance  DSL  chart  ecosystem  grokkability  heavyweights  CAS  static-dynamic  lower-bounds  complexity  tcs  open-problems  big-surf  ideas  certificates-recognition  proof-systems  PCP  mediterranean  SDP  meta:prediction  epistemic  questions  guessing  distributed  overflow  nibble  soft-question  track-record  big-list  hmm  frontier  state-of-art  move-fast-(and-break-things)  grokkability-clarity  technical-writing  trust
july 2019 by nhaliday
Why is Software Engineering so difficult? - James Miller
basic message: No silver bullet!

most interesting nuggets:
Scale and Complexity
- Windows 7 > 50 million LOC
Expect a staggering number of bugs.

Bugs?
- Well-written C and C++ code contains some 5 to 10 errors per 100 LOC after a clean compile, but before inspection and testing.
- At a 5% rate any 50 MLOC program will start off with some 2.5 million bugs.

Bug removal
- Testing typically exercises only half the code.

Better bug removal?
- There are better ways to do testing that do produce fantastic programs.”
* No, its only an opinion!
* In general Software Engineering has ....
NO FACTS!

So why not do this?
- The costs are unbelievable.
- It’s not unusual for the qualification process to produce a half page of documentation for each line of code.
pdf  slides  engineering  nitty-gritty  programming  best-practices  roots  comparison  cost-benefit  software  systematic-ad-hoc  structure  error  frontier  debugging  checking  formal-methods  context  detail-architecture  intricacy  big-picture  system-design  correctness  scale  scaling-tech  shipping  money  data  stylized-facts  street-fighting  objektbuch  pro-rata  estimate  pessimism  degrees-of-freedom  volo-avolo  no-go  things  thinking  summary  quality  density  methodology
may 2019 by nhaliday
quality - Is the average number of bugs per loc the same for different programming languages? - Software Engineering Stack Exchange
Contrary to intuition, the number of errors per 1000 lines of does seem to be relatively constant, reguardless of the specific language involved. Steve McConnell, author of Code Complete and Software Estimation: Demystifying the Black Art goes over this area in some detail.

I don't have my copies readily to hand - they're sitting on my bookshelf at work - but a quick Google found a relevant quote:

Industry Average: "about 15 - 50 errors per 1000 lines of delivered code."
(Steve) further says this is usually representative of code that has some level of structured programming behind it, but probably includes a mix of coding techniques.

Quoted from Code Complete, found here: http://mayerdan.com/ruby/2012/11/11/bugs-per-line-of-code-ratio/

If memory serves correctly, Steve goes into a thorough discussion of this, showing that the figures are constant across languages (C, C++, Java, Assembly and so on) and despite difficulties (such as defining what "line of code" means).

Most importantly he has lots of citations for his sources - he's not offering unsubstantiated opinions, but has the references to back them up.

[ed.: I think this is delivered code? So after testing, debugging, etc. I'm more interested in the metric for the moment after you've gotten something to compile.

edit: cf https://pinboard.in/u:nhaliday/b:0a6eb68166e6]
q-n-a  stackex  programming  engineering  nitty-gritty  error  flux-stasis  books  recommendations  software  checking  debugging  pro-rata  pls  comparison  parsimony  measure  data  objektbuch  speculation  accuracy  density  correctness  estimate  street-fighting  multi  quality  stylized-facts  methodology
april 2019 by nhaliday
An adaptability limit to climate change due to heat stress
Despite the uncertainty in future climate-change impacts, it is often assumed that humans would be able to adapt to any possible warming. Here we argue that heat stress imposes a robust upper limit to such adaptation. Peak heat stress, quantified by the wet-bulb temperature TW, is surprisingly similar across diverse climates today. TW never exceeds 31 °C. Any exceedence of 35 °C for extended periods should induce hyperthermia in humans and other mammals, as dissipation of metabolic heat becomes impossible. While this never happens now, it would begin to occur with global-mean warming of about 7 °C, calling the habitability of some regions into question. With 11–12 °C warming, such regions would spread to encompass the majority of the human population as currently distributed. Eventual warmings of 12 °C are possible from fossil fuel burning. One implication is that recent estimates of the costs of unmitigated climate change are too low unless the range of possible warming can somehow be narrowed. Heat stress also may help explain trends in the mammalian fossil record.

Trajectories of the Earth System in the Anthropocene: http://www.pnas.org/content/early/2018/07/31/1810141115
We explore the risk that self-reinforcing feedbacks could push the Earth System toward a planetary threshold that, if crossed, could prevent stabilization of the climate at intermediate temperature rises and cause continued warming on a “Hothouse Earth” pathway even as human emissions are reduced. Crossing the threshold would lead to a much higher global average temperature than any interglacial in the past 1.2 million years and to sea levels significantly higher than at any time in the Holocene. We examine the evidence that such a threshold might exist and where it might be.
study  org:nat  environment  climate-change  humanity  existence  risk  futurism  estimate  physics  thermo  prediction  temperature  nature  walls  civilization  flexibility  rigidity  embodied  multi  manifolds  plots  equilibrium  phase-transition  oscillation  comparison  complex-systems  earth
august 2018 by nhaliday
Theory of Self-Reproducing Automata - John von Neumann
Fourth Lecture: THE ROLE OF HIGH AND OF EXTREMELY HIGH COMPLICATION

Comparisons between computing machines and the nervous systems. Estimates of size for computing machines, present and near future.

Estimates for size for the human central nervous system. Excursus about the “mixed” character of living organisms. Analog and digital elements. Observations about the “mixed” character of all componentry, artificial as well as natural. Interpretation of the position to be taken with respect to these.

Evaluation of the discrepancy in size between artificial and natural automata. Interpretation of this discrepancy in terms of physical factors. Nature of the materials used.

The probability of the presence of other intellectual factors. The role of complication and the theoretical penetration that it requires.

Questions of reliability and errors reconsidered. Probability of individual errors and length of procedure. Typical lengths of procedure for computing machines and for living organisms--that is, for artificial and for natural automata. Upper limits on acceptable probability of error in individual operations. Compensation by checking and self-correcting features.

Differences of principle in the way in which errors are dealt with in artificial and in natural automata. The “single error” principle in artificial automata. Crudeness of our approach in this case, due to the lack of adequate theory. More sophisticated treatment of this problem in natural automata: The role of the autonomy of parts. Connections between this autonomy and evolution.

- 10^10 neurons in brain, 10^4 vacuum tubes in largest computer at time
- machines faster: 5 ms from neuron potential to neuron potential, 10^-3 ms for vacuum tubes

https://en.wikipedia.org/wiki/John_von_Neumann#Computing
pdf  article  papers  essay  nibble  math  cs  computation  bio  neuro  neuro-nitgrit  scale  magnitude  comparison  acm  von-neumann  giants  thermo  phys-energy  speed  performance  time  density  frequency  hardware  ems  efficiency  dirty-hands  street-fighting  fermi  estimate  retention  physics  interdisciplinary  multi  wiki  links  people  🔬  atoms  duplication  iteration-recursion  turing  complexity  measure  nature  technology  complex-systems  bits  information-theory  circuits  robust  structure  composition-decomposition  evolution  mutation  axioms  analogy  thinking  input-output  hi-order-bits  coding-theory  flexibility  rigidity  automata-languages
april 2018 by nhaliday
Existential Risks: Analyzing Human Extinction Scenarios
https://archive.is/dUTD5
Would you endorse choosing policy to max the expected duration of civilization, at least as a good first approximation?
Can anyone suggest a different first approximation that would get more votes?

https://archive.is/RpygO
How useful would it be to agree on a relatively-simple first-approximation observable-after-the-fact metric for what we want from the future universe, such as total life years experienced, or civilization duration?

We're Underestimating the Risk of Human Extinction: https://www.theatlantic.com/technology/archive/2012/03/were-underestimating-the-risk-of-human-extinction/253821/
An Oxford philosopher argues that we are not adequately accounting for technology's risks—but his solution to the problem is not for Luddites.

Anderson: You have argued that we underrate existential risks because of a particular kind of bias called observation selection effect. Can you explain a bit more about that?

Bostrom: The idea of an observation selection effect is maybe best explained by first considering the simpler concept of a selection effect. Let's say you're trying to estimate how large the largest fish in a given pond is, and you use a net to catch a hundred fish and the biggest fish you find is three inches long. You might be tempted to infer that the biggest fish in this pond is not much bigger than three inches, because you've caught a hundred of them and none of them are bigger than three inches. But if it turns out that your net could only catch fish up to a certain length, then the measuring instrument that you used would introduce a selection effect: it would only select from a subset of the domain you were trying to sample.

Now that's a kind of standard fact of statistics, and there are methods for trying to correct for it and you obviously have to take that into account when considering the fish distribution in your pond. An observation selection effect is a selection effect introduced not by limitations in our measurement instrument, but rather by the fact that all observations require the existence of an observer. This becomes important, for instance, in evolutionary biology. For instance, we know that intelligent life evolved on Earth. Naively, one might think that this piece of evidence suggests that life is likely to evolve on most Earth-like planets. But that would be to overlook an observation selection effect. For no matter how small the proportion of all Earth-like planets that evolve intelligent life, we will find ourselves on a planet that did. Our data point-that intelligent life arose on our planet-is predicted equally well by the hypothesis that intelligent life is very improbable even on Earth-like planets as by the hypothesis that intelligent life is highly probable on Earth-like planets. When it comes to human extinction and existential risk, there are certain controversial ways that observation selection effects might be relevant.
bostrom  ratty  miri-cfar  skunkworks  philosophy  org:junk  list  top-n  frontier  speedometer  risk  futurism  local-global  scale  death  nihil  technology  simulation  anthropic  nuclear  deterrence  environment  climate-change  arms  competition  ai  ai-control  genetics  genomics  biotech  parasites-microbiome  disease  offense-defense  physics  tails  network-structure  epidemiology  space  geoengineering  dysgenics  ems  authoritarianism  government  values  formal-values  moloch  enhancement  property-rights  coordination  cooperate-defect  flux-stasis  ideas  prediction  speculation  humanity  singularity  existence  cybernetics  study  article  letters  eden-heaven  gedanken  multi  twitter  social  discussion  backup  hanson  metrics  optimization  time  long-short-run  janus  telos-atelos  poll  forms-instances  threat-modeling  selection  interview  expert-experience  malthus  volo-avolo  intel  leviathan  drugs  pharma  data  estimate  nature  longevity  expansionism  homo-hetero  utopia-dystopia
march 2018 by nhaliday
Stein's example - Wikipedia
Stein's example (or phenomenon or paradox), in decision theory and estimation theory, is the phenomenon that when three or more parameters are estimated simultaneously, there exist combined estimators more accurate on average (that is, having lower expected mean squared error) than any method that handles the parameters separately. It is named after Charles Stein of Stanford University, who discovered the phenomenon in 1955.[1]

An intuitive explanation is that optimizing for the mean-squared error of a combined estimator is not the same as optimizing for the errors of separate estimators of the individual parameters. In practical terms, if the combined error is in fact of interest, then a combined estimator should be used, even if the underlying parameters are independent; this occurs in channel estimation in telecommunications, for instance (different factors affect overall channel performance). On the other hand, if one is instead interested in estimating an individual parameter, then using a combined estimator does not help and is in fact worse.

...

Many simple, practical estimators achieve better performance than the ordinary estimator. The best-known example is the James–Stein estimator, which works by starting at X and moving towards a particular point (such as the origin) by an amount inversely proportional to the distance of X from that point.
nibble  concept  levers  wiki  reference  acm  stats  probability  decision-theory  estimate  distribution  atoms
february 2018 by nhaliday
Information Processing: US Needs a National AI Strategy: A Sputnik Moment?
FT podcasts on US-China competition and AI: http://infoproc.blogspot.com/2018/05/ft-podcasts-on-us-china-competition-and.html

A new recommended career path for effective altruists: China specialist: https://80000hours.org/articles/china-careers/
Our rough guess is that it would be useful for there to be at least ten people in the community with good knowledge in this area within the next few years.

By “good knowledge” we mean they’ve spent at least 3 years studying these topics and/or living in China.

We chose ten because that would be enough for several people to cover each of the major areas listed (e.g. 4 within AI, 2 within biorisk, 2 within foreign relations, 1 in another area).

AI Policy and Governance Internship: https://www.fhi.ox.ac.uk/ai-policy-governance-internship/

https://www.fhi.ox.ac.uk/deciphering-chinas-ai-dream/
Deciphering China’s AI Dream
The context, components, capabilities, and consequences of
China’s strategy to lead the world in AI

Europe’s AI delusion: https://www.politico.eu/article/opinion-europes-ai-delusion/
Brussels is failing to grasp threats and opportunities of artificial intelligence.
By BRUNO MAÇÃES

When the computer program AlphaGo beat the Chinese professional Go player Ke Jie in a three-part match, it didn’t take long for Beijing to realize the implications.

If algorithms can already surpass the abilities of a master Go player, it can’t be long before they will be similarly supreme in the activity to which the classic board game has always been compared: war.

As I’ve written before, the great conflict of our time is about who can control the next wave of technological development: the widespread application of artificial intelligence in the economic and military spheres.

...

If China’s ambitions sound plausible, that’s because the country’s achievements in deep learning are so impressive already. After Microsoft announced that its speech recognition software surpassed human-level language recognition in October 2016, Andrew Ng, then head of research at Baidu, tweeted: “We had surpassed human-level Chinese recognition in 2015; happy to see Microsoft also get there for English less than a year later.”

...

One obvious advantage China enjoys is access to almost unlimited pools of data. The machine-learning technologies boosting the current wave of AI expansion are as good as the amount of data they can use. That could be the number of people driving cars, photos labeled on the internet or voice samples for translation apps. With 700 or 800 million Chinese internet users and fewer data protection rules, China is as rich in data as the Gulf States are in oil.

How can Europe and the United States compete? They will have to be commensurately better in developing algorithms and computer power. Sadly, Europe is falling behind in these areas as well.

...

Chinese commentators have embraced the idea of a coming singularity: the moment when AI surpasses human ability. At that point a number of interesting things happen. First, future AI development will be conducted by AI itself, creating exponential feedback loops. Second, humans will become useless for waging war. At that point, the human mind will be unable to keep pace with robotized warfare. With advanced image recognition, data analytics, prediction systems, military brain science and unmanned systems, devastating wars might be waged and won in a matter of minutes.

...

The argument in the new strategy is fully defensive. It first considers how AI raises new threats and then goes on to discuss the opportunities. The EU and Chinese strategies follow opposite logics. Already on its second page, the text frets about the legal and ethical problems raised by AI and discusses the “legitimate concerns” the technology generates.

The EU’s strategy is organized around three concerns: the need to boost Europe’s AI capacity, ethical issues and social challenges. Unfortunately, even the first dimension quickly turns out to be about “European values” and the need to place “the human” at the center of AI — forgetting that the first word in AI is not “human” but “artificial.”

https://archive.is/m3Njh
US military: "LOL, China thinks it's going to be a major player in AI, but we've got all the top AI researchers. You guys will help us develop weapons, right?"

US AI researchers: "No."

US military: "But... maybe just a computer vision app."

US AI researchers: "NO."

https://www.theverge.com/2018/4/4/17196818/ai-boycot-killer-robots-kaist-university-hanwha
https://archive.is/3wbHm
AI-risk was a mistake.
hsu  scitariat  commentary  video  presentation  comparison  usa  china  asia  sinosphere  frontier  technology  science  ai  speedometer  innovation  google  barons  deepgoog  stories  white-paper  strategy  migration  iran  human-capital  corporation  creative  alien-character  military  human-ml  nationalism-globalism  security  investing  government  games  deterrence  defense  nuclear  arms  competition  risk  ai-control  musk  optimism  multi  news  org:mag  europe  EU  80000-hours  effective-altruism  proposal  article  realness  offense-defense  war  biotech  altruism  language  foreign-lang  philosophy  the-great-west-whale  enhancement  foreign-policy  geopolitics  anglo  jobs  career  planning  hmm  travel  charity  tech  intel  media  teaching  tutoring  russia  india  miri-cfar  pdf  automation  class  labor  polisci  society  trust  n-factor  corruption  leviathan  ethics  authoritarianism  individualism-collectivism  revolution  economics  inequality  civic  law  regulation  data  scale  pro-rata  capital  zero-positive-sum  cooperate-defect  distribution  time-series  tre
february 2018 by nhaliday
Sex, Drugs, and Bitcoin: How Much Illegal Activity Is Financed Through Cryptocurrencies? by Sean Foley, Jonathan R. Karlsen, Tālis J. Putniņš :: SSRN
Cryptocurrencies are among the largest unregulated markets in the world. We find that approximately one-quarter of bitcoin users and one-half of bitcoin transactions are associated with illegal activity. Around $72 billion of illegal activity per year involves bitcoin, which is close to the scale of the US and European markets for illegal drugs. The illegal share of bitcoin activity declines with mainstream interest in bitcoin and with the emergence of more opaque cryptocurrencies. The techniques developed in this paper have applications in cryptocurrency surveillance. Our findings suggest that cryptocurrencies are transforming the way black markets operate by enabling “black e-commerce.” study economics law leviathan bitcoin cryptocurrency crypto impetus scale markets civil-liberty randy-ayndy crime criminology measurement estimate pro-rata money monetary-fiscal crypto-anarchy drugs internet tradecraft opsec security intel february 2018 by nhaliday Team *Decorations Until Epiphany* on Twitter: "@RoundSqrCupola maybe just C https://t.co/SFPXb3qrAE" https://archive.is/k0fsS Remember ‘BRICs’? Now it’s just ICs. -- maybe just C Solow predicts that if 2 countries have the same TFP, then the poorer nation should grow faster. But poorer India grows more slowly than China. Solow thinking leads one to suspect India has substantially lower TFP. Recent growth is great news, but alas 5 years isn't the long run! FWIW under Solow conditional convergence assumptions--historically robust--the fact that a country as poor as India grows only a few % faster than the world average is a sign they'll end up poorer than S Europe. see his spreadsheet here: http://mason.gmu.edu/~gjonesb/SolowForecast.xlsx spearhead econotariat garett-jones unaffiliated twitter social discussion india asia china economics macro growth-econ econ-metrics wealth wealth-of-nations convergence world developing-world trends time-series cjones-like prediction multi backup the-bones long-short-run europe mediterranean comparison simulation econ-productivity great-powers thucydides broad-econ pop-diff microfoundations 🎩 marginal hive-mind rindermann-thompson hari-seldon tools calculator estimate december 2017 by nhaliday galaxy - How do astronomers estimate the total mass of dust in clouds and galaxies? - Astronomy Stack Exchange Dust absorbs stellar light (primarily in the ultraviolet), and is heated up. Subsequently it cools by emitting infrared, "thermal" radiation. Assuming a dust composition and grain size distribution, the amount of emitted IR light per unit dust mass can be calculated as a function of temperature. Observing the object at several different IR wavelengths, a Planck curve can be fitted to the data points, yielding the dust temperature. The more UV light incident on the dust, the higher the temperature. The result is somewhat sensitive to the assumptions, and thus the uncertainties are sometimes quite large. The more IR data points obtained, the better. If only one IR point is available, the temperature cannot be calculated. Then there's a degeneracy between incident UV light and the amount of dust, and the mass can only be estimated to within some orders of magnitude (I think). nibble q-n-a overflow space measurement measure estimate physics electromag visuo methodology december 2017 by nhaliday Tax Evasion and Inequality This paper attempts to estimate the size and distribution of tax evasion in rich countries. We combine stratified random audits—the key source used to study tax evasion so far—with new micro-data leaked from two large offshore financial institutions, HSBC Switzerland (“Swiss leaks”) and Mossack Fonseca (“Panama Papers”). We match these data to population-wide wealth records in Norway, Sweden, and Denmark. We find that tax evasion rises sharply with wealth, a phenomenon that random audits fail to capture. On average about 3% of personal taxes are evaded in Scandinavia, but this figure rises to about 30% in the top 0.01% of the wealth distribution, a group that includes households with more than$40 million in net wealth. A simple model of the supply of tax evasion services can explain why evasion rises steeply with wealth. Taking tax evasion into account increases the rise in inequality seen in tax data since the 1970s markedly, highlighting the need to move beyond tax data to capture income and wealth at the top, even in countries where tax compliance is generally high. We also find that after reducing tax evasion—by using tax amnesties—tax evaders do not legally avoid taxes more. This result suggests that fighting tax evasion can be an effective way to collect more tax revenue from the ultra-wealthy.

Figure 1

America’s unreported economy: measuring the size, growth and determinants of income tax evasion in the U.S.: https://link.springer.com/article/10.1007/s10611-011-9346-x
This study empirically investigates the extent of noncompliance with the tax code and examines the determinants of federal income tax evasion in the U.S. Employing a refined version of Feige’s (Staff Papers, International Monetary Fund 33(4):768–881, 1986, 1989) General Currency Ratio (GCR) model to estimate a time series of unreported income as our measure of tax evasion, we find that 18–23% of total reportable income may not properly be reported to the IRS. This gives rise to a 2009 “tax gap” in the range of $390–$540 billion. As regards the determinants of tax noncompliance, we find that federal income tax evasion is an increasing function of the average effective federal income tax rate, the unemployment rate, the nominal interest rate, and per capita real GDP, and a decreasing function of the IRS audit rate. Despite important refinements of the traditional currency ratio approach for estimating the aggregate size and growth of unreported economies, we conclude that the sensitivity of the results to different benchmarks, imperfect data sources and alternative specifying assumptions precludes obtaining results of sufficient accuracy and reliability to serve as effective policy guides.
pdf  study  economics  micro  evidence-based  data  europe  nordic  scale  class  compensation  money  monetary-fiscal  political-econ  redistribution  taxes  madisonian  inequality  history  mostly-modern  natural-experiment  empirical  🎩  cocktail  correlation  models  supply-demand  GT-101  crooked  elite  vampire-squid  nationalism-globalism  multi  pro-rata  usa  time-series  trends  world-war  cold-war  government  todo  planning  long-term  trivia  law  crime  criminology  estimate  speculation  measurement  labor  macro  econ-metrics  wealth  stock-flow  time  density  criminal-justice  frequency  dark-arts  traces  evidence
october 2017 by nhaliday
Caught in the act | West Hunter
The fossil record is sparse. Let me try to explain that. We have at most a few hundred Neanderthal skeletons, most in pretty poor shape. How many Neanderthals ever lived? I think their population varied in size quite a bit – lowest during glacial maxima, probably highest in interglacials. Their degree of genetic diversity suggests an effective population size of ~1000, but that would be dominated by the low points (harmonic average). So let’s say 50,000 on average, over their whole range (Europe, central Asia, the Levant, perhaps more). Say they were around for 300,000 years, with a generation time of 30 years – 10,000 generations, for a total of five hundred million Neanderthals over all time. So one in a million Neanderthals ends up in a museum: one every 20 generations. Low time resolution!

So if anatomically modern humans rapidly wiped out Neanderthals, we probably couldn’t tell. In much the same way, you don’t expect to find the remains of many dinosaurs killed by the Cretaceous meteor impact (at most one millionth of one generation, right?), or of Columbian mammoths killed by a wave of Amerindian hunters. Sometimes invaders leave a bigger footprint: a bunch of cities burning down with no rebuilding tells you something. But even when you know that population A completely replaced population B, it can be hard to prove that just how it happened. After all, population A could have all committed suicide just before B showed up. Stranger things have happened – but not often.
west-hunter  scitariat  discussion  ideas  data  objektbuch  scale  magnitude  estimate  population  sapiens  archaics  archaeology  pro-rata  history  antiquity  methodology  volo-avolo  measurement  pop-structure  density  time  frequency  apollonian-dionysian  traces  evidence
september 2017 by nhaliday
Atrocity statistics from the Roman Era
Gibbon, Decline & Fall v.2 ch.XVI: < 2,000 k. under Roman persecution.
Ludwig Hertling ("Die Zahl de Märtyrer bis 313", 1944) estimated 100,000 Christians killed between 30 and 313 CE. (cited -- unfavorably -- by David Henige, Numbers From Nowhere, 1998)
Catholic Encyclopedia, "Martyr": number of Christian martyrs under the Romans unknown, unknowable. Origen says not many. Eusebius says thousands.

...

General population decline during The Fall of Rome: 7,000,000 [make link]
- Colin McEvedy, The New Penguin Atlas of Medieval History (1992)
- From 2nd Century CE to 4th Century CE: Empire's population declined from 45M to 36M [i.e. 9M]
- From 400 CE to 600 CE: Empire's population declined by 20% [i.e. 7.2M]
- Paul Bairoch, Cities and economic development: from the dawn of history to the present, p.111
- "The population of Europe except Russia, then, having apparently reached a high point of some 40-55 million people by the start of the third century [ca.200 C.E.], seems to have fallen by the year 500 to about 30-40 million, bottoming out at about 20-35 million around 600." [i.e. ca.20M]
- Francois Crouzet, A History of the European Economy, 1000-2000 (University Press of Virginia: 2001) p.1.
- "The population of Europe (west of the Urals) in c. AD 200 has been estimated at 36 million; by 600, it had fallen to 26 million; another estimate (excluding ‘Russia’) gives a more drastic fall, from 44 to 22 million." [i.e. 10M or 22M]

also:
The geometric mean of these two extremes would come to 4½ per day, which is a credible daily rate for the really bad years.

why geometric mean? can you get it as the MLE given min{X1, ..., Xn} and max{X1, ..., Xn} for {X_i} iid Poissons? some kinda limit? think it might just be a rule of thumb.

yeah, it's a rule of thumb. found it it his book (epub).
org:junk  data  let-me-see  scale  history  iron-age  mediterranean  the-classics  death  nihil  conquest-empire  war  peace-violence  gibbon  trivia  multi  todo  AMT  expectancy  heuristic  stats  ML-MAP-E  data-science  estimate  magnitude  population  demographics  database  list  religion  christianity  leviathan
september 2017 by nhaliday
Medicine as a pseudoscience | West Hunter
The idea that venesection was a good thing, or at least not so bad, on the grounds that one in a few hundred people have hemochromatosis (in Northern Europe) reminds me of the people who don’t wear a seatbelt, since it would keep them from being thrown out of their convertible into a waiting haystack, complete with nubile farmer’s daughter. Daughters. It could happen. But it’s not the way to bet.

Back in the good old days, Charles II, age 53, had a fit one Sunday evening, while fondling two of his mistresses.

Monday they bled him (cupping and scarifying) of eight ounces of blood. Followed by an antimony emetic, vitriol in peony water, purgative pills, and a clyster. Followed by another clyster after two hours. Then syrup of blackthorn, more antimony, and rock salt. Next, more laxatives, white hellebore root up the nostrils. Powdered cowslip flowers. More purgatives. Then Spanish Fly. They shaved his head and stuck blistering plasters all over it, plastered the soles of his feet with tar and pigeon-dung, then said good-night.

...

Friday. The king was worse. He tells them not to let poor Nelly starve. They try the Oriental Bezoar Stone, and more bleeding. Dies at noon.

Most people didn’t suffer this kind of problem with doctors, since they never saw one. Charles had six. Now Bach and Handel saw the same eye surgeon, John Taylor – who blinded both of them. Not everyone can put that on his resume!

You may wonder how medicine continued to exist, if it had a negative effect, on the whole. There’s always the placebo effect – at least there would be, if it existed. Any real placebo effect is very small: I’d guess exactly zero. But there is regression to the mean. You see the doctor when you’re feeling worse than average – and afterwards, if he doesn’t kill you outright, you’re likely to feel better. Which would have happened whether you’d seen him or not, but they didn’t often do RCTs back in the day – I think James Lind was the first (1747).

Back in the late 19th century, Christian Scientists did better than others when sick, because they didn’t believe in medicine. For reasons I think mistaken, because Mary Baker Eddy rejected the reality of the entire material world, but hey, it worked. Parenthetically, what triggered all that New Age nonsense in 19th century New England? Hash?

This did not change until fairly recently. Sometime in the early 20th medicine, clinical medicine, what doctors do, hit break-even. Now we can’t do without it. I wonder if there are, or will be, other examples of such a pile of crap turning (mostly) into a real science.

The brilliant GP I've had for 35+ years has retired. How can I find another one who meets my requirements?

1 is overweight
2 drinks more than officially recommended amounts
3 has an amused, tolerant atitude to human failings
4 is well aware that we're all going to die anyway, & there are better or worse ways to die
5 has a healthy skeptical attitude to mainstream medical science
6 is wholly dismissive of "a|ternative” medicine
7 believes in evolution
8 thinks most diseases get better without intervention, & knows the dangers of false positives
9 understands the base rate fallacy

EconPapers: Was Civil War Surgery Effective?: http://econpapers.repec.org/paper/htrhcecon/444.htm
contra Greg Cochran:
To shed light on the subject, I analyze a data set created by Dr. Edmund Andrews, a Civil war surgeon with the 1st Illinois Light Artillery. Dr. Andrews’s data can be rendered into an observational data set on surgical intervention and recovery, with controls for wound location and severity. The data also admits instruments for the surgical decision. My analysis suggests that Civil War surgery was effective, and increased the probability of survival of the typical wounded soldier, with average treatment effect of 0.25-0.28.

Medical Prehistory: https://westhunt.wordpress.com/2016/03/14/medical-prehistory/
What ancient medical treatments worked?

https://westhunt.wordpress.com/2016/03/14/medical-prehistory/#comment-76878
In some very, very limited conditions, bleeding?
--
Bad for you 99% of the time.

https://westhunt.wordpress.com/2016/03/14/medical-prehistory/#comment-76947
Colchicine – used to treat gout – discovered by the Ancient Greeks.

https://westhunt.wordpress.com/2016/03/14/medical-prehistory/#comment-76973
Dracunculiasis (Guinea worm)
Wrap the emerging end of the worm around a stick and slowly pull it out.
(3,500 years later, this remains the standard treatment.)
https://en.wikipedia.org/wiki/Ebers_Papyrus

https://westhunt.wordpress.com/2016/03/14/medical-prehistory/#comment-76971
Some of the progress is from formal medicine, most is from civil engineering, better nutrition ( ag science and physical chemistry), less crowded housing.

Nurses vs doctors: https://westhunt.wordpress.com/2014/10/01/nurses-vs-doctors/
Medicine, the things that doctors do, was an ineffective pseudoscience until fairly recently. Until 1800 or so, they were wrong about almost everything. Bleeding, cupping, purging, the four humors – useless. In the 1800s, some began to realize that they were wrong, and became medical nihilists that improved outcomes by doing less. Some patients themselves came to this realization, as when Civil War casualties hid from the surgeons and had better outcomes. Sometime in the early 20th century, MDs reached break-even, and became an increasingly positive influence on human health. As Lewis Thomas said, medicine is the youngest science.

Nursing, on the other hand, has always been useful. Just making sure that a patient is warm and nourished when too sick to take care of himself has helped many survive. In fact, some of the truly crushing epidemics have been greatly exacerbated when there were too few healthy people to take care of the sick.

Nursing must be old, but it can’t have existed forever. Whenever it came into existence, it must have changed the selective forces acting on the human immune system. Before nursing, being sufficiently incapacitated would have been uniformly fatal – afterwards, immune responses that involved a period of incapacitation (with eventual recovery) could have been selectively favored.

when MDs broke even: https://westhunt.wordpress.com/2014/10/01/nurses-vs-doctors/#comment-58981
I’d guess the 1930s. Lewis Thomas thought that he was living through big changes. They had a working serum therapy for lobar pneumonia ( antibody-based). They had many new vaccines ( diphtheria in 1923, whopping cough in 1926, BCG and tetanus in 1927, yellow fever in 1935, typhus in 1937.) Vitamins had been mostly worked out. Insulin was discovered in 1929. Blood transfusions. The sulfa drugs, first broad-spectrum antibiotics, showed up in 1935.

DALYs per doctor: https://westhunt.wordpress.com/2018/01/22/dalys-per-doctor/
The disability-adjusted life year (DALY) is a measure of overall disease burden – the number of years lost. I’m wondering just much harm premodern medicine did, per doctor. How many healthy years of life did a typical doctor destroy (net) in past times?

...

It looks as if the average doctor (in Western medicine) killed a bunch of people over his career ( when contrasted with doing nothing). In the Charles Manson class.

Eventually the market saw through this illusion. Only took a couple of thousand years.

https://westhunt.wordpress.com/2018/01/22/dalys-per-doctor/#comment-100741
That a very large part of healthcare spending is done for non-health reasons. He has a chapter on this in his new book, also check out his paper “Showing That You Care: The Evolution of Health Altruism” http://mason.gmu.edu/~rhanson/showcare.pdf
--
I ran into too much stupidity to finish the article. Hanson’s a loon. For example when he talks about the paradox of blacks being more sentenced on drug offenses than whites although they use drugs at similar rate. No paradox: guys go to the big house for dealing, not for using. Where does he live – Mars?

I had the same reaction when Hanson parroted some dipshit anthropologist arguing that the stupid things people do while drunk are due to social expectations, not really the alcohol.
Horseshit.

I don’t think that being totally unable to understand everybody around you necessarily leads to deep insights.

https://westhunt.wordpress.com/2018/01/22/dalys-per-doctor/#comment-100744
What I’ve wondered is if there was anything that doctors did that actually was helpful and if perhaps that little bit of success helped them fool people into thinking the rest of it helped.
--
Setting bones. extracting arrows: spoon of Diocles. Colchicine for gout. Extracting the Guinea worm. Sometimes they got away with removing the stone. There must be others.
--
Quinine is relatively recent: post-1500. Obstetrical forceps also. Caesarean deliveries were almost always fatal to the mother until fairly recently.

Opium has been around for a long while : it works.

https://westhunt.wordpress.com/2018/01/22/dalys-per-doctor/#comment-100839
If pre-modern medicine was indeed worse than useless – how do you explain no one noticing that patients who get expensive treatments are worse off than those who didn’t?
--
were worse off. People are kinda dumb – you’ve noticed?
--
My impression is that while people may be “kinda dumb”, ancient customs typically aren’t.
Even if we assume that all people who lived prior to the 19th century were too dumb to make the rational observation, wouldn’t you expect this ancient practice to be subject to selective pressure?
--
Your impression is wrong. Do you think that there some slick reason for Carthaginians incinerating their first-born?

details on blood-letting and hemochromatosis: https://westhunt.wordpress.com/2018/01/22/dalys-per-doctor/#comment-100746

Starting Over: https://westhunt.wordpress.com/2018/01/23/starting-over/
Looking back on it, human health would have … [more]
west-hunter  scitariat  discussion  ideas  medicine  meta:medicine  science  realness  cost-benefit  the-trenches  info-dynamics  europe  the-great-west-whale  history  iron-age  the-classics  mediterranean  medieval  early-modern  mostly-modern  🌞  harvard  aphorism  rant  healthcare  regression-to-mean  illusion  public-health  multi  usa  northeast  pre-ww2  checklists  twitter  social  albion  ability-competence  study  cliometrics  war  trivia  evidence-based  data  intervention  effect-size  revolution  speculation  sapiens  drugs  antiquity  lived-experience  list  survey  questions  housing  population  density  nutrition  wiki  embodied  immune  evolution  poast  chart  markets  civil-liberty  randy-ayndy  market-failure  impact  scale  pro-rata  estimate  street-fighting  fermi  marginal  truth  recruiting  alt-inst  academia  social-science  space  physics  interdisciplinary  ratty  lesswrong  autism  👽  subculture  hanson  people  track-record  crime  criminal-justice  criminology  race  ethanol  error  video  lol  comedy  tradition  institutions  iq  intelligence  MENA  impetus  legacy
august 2017 by nhaliday
Constitutive equation - Wikipedia
In physics and engineering, a constitutive equation or constitutive relation is a relation between two physical quantities (especially kinetic quantities as related to kinematic quantities) that is specific to a material or substance, and approximates the response of that material to external stimuli, usually as applied fields or forces. They are combined with other equations governing physical laws to solve physical problems; for example in fluid mechanics the flow of a fluid in a pipe, in solid state physics the response of a crystal to an electric field, or in structural analysis, the connection between applied stresses or forces to strains or deformations.

Some constitutive equations are simply phenomenological; others are derived from first principles. A common approximate constitutive equation frequently is expressed as a simple proportionality using a parameter taken to be a property of the material, such as electrical conductivity or a spring constant. However, it is often necessary to account for the directional dependence of the material, and the scalar parameter is generalized to a tensor. Constitutive relations are also modified to account for the rate of response of materials and their non-linear behavior.[1] See the article Linear response function.
nibble  wiki  reference  article  physics  mechanics  electromag  identity  estimate  approximation  empirical  stylized-facts  list  dirty-hands  fluid  logos
august 2017 by nhaliday
Demography of the Roman Empire - Wikipedia
There are few recorded population numbers for the whole of antiquity, and those that exist are often rhetorical or symbolic. Unlike the contemporaneous Han Dynasty, no general census survives for the Roman Empire. The late period of the Roman Republic provides a small exception to this general rule: serial statistics for Roman citizen numbers, taken from census returns, survive for the early Republic through the 1st century CE.[41] Only the figures for periods after the mid-3rd century BCE are reliable, however. Fourteen figures are available for the 2nd century BCE (from 258,318 to 394,736). Only four figures are available for the 1st century BCE, and are feature a large break between 70/69 BCE (910,000) and 28 BCE (4,063,000). The interpretation of the later figures—the Augustan censuses of 28 BCE, 8 BCE, and 14 CE—is therefore controversial.[42] Alternate interpretations of the Augustan censuses (such as those of E. Lo Cascio[43]) produce divergent population histories across the whole imperial period.[44]

Roman population size: the logic of the debate: https://www.princeton.edu/~pswpc/pdfs/scheidel/070706.pdf
- Walter Scheidel (cited in book by Vaclav Smil, "Why America is Not a New Rome")

Our ignorance of ancient population numbers is one of the biggest obstacles to our understanding of Roman history. After generations of prolific scholarship, we still do not know how many people inhabited Roman Italy and the Mediterranean at any given point in time. When I say ‘we do not know’ I do not simply mean that we lack numbers that are both precise and safely known to be accurate: that would surely be an unreasonably high standard to apply to any pre-modern society. What I mean is that even the appropriate order of magnitude remains a matter of intense dispute.

Historical urban community sizes: https://en.wikipedia.org/wiki/Historical_urban_community_sizes

World population estimates: https://en.wikipedia.org/wiki/World_population_estimates
As a general rule, the confidence of estimates on historical world population decreases for the more distant past. Robust population data only exists for the last two or three centuries. Until the late 18th century, few governments had ever performed an accurate census. In many early attempts, such as in Ancient Egypt and the Persian Empire, the focus was on counting merely a subset of the population for purposes of taxation or military service.[3] Published estimates for the 1st century ("AD 1") suggest an uncertainty of the order of 50% (estimates range between 150 and 330 million). Some estimates extend their timeline into deep prehistory, to "10,000 BC", i.e. the early Holocene, when world population estimates range roughly between one and ten million (with an uncertainty of up to an order of magnitude).[4][5]

Estimates for yet deeper prehistory, into the Paleolithic, are of a different nature. At this time human populations consisted entirely of non-sedentary hunter-gatherer populations, with anatomically modern humans existing alongside archaic human varieties, some of which are still ancestral to the modern human population due to interbreeding with modern humans during the Upper Paleolithic. Estimates of the size of these populations are a topic of paleoanthropology. A late human population bottleneck is postulated by some scholars at approximately 70,000 years ago, during the Toba catastrophe, when Homo sapiens population may have dropped to as low as between 1,000 and 10,000 individuals.[6][7] For the time of speciation of Homo sapiens, some 200,000 years ago, an effective population size of the order of 10,000 to 30,000 individuals has been estimated, with an actual "census population" of early Homo sapiens of roughly 100,000 to 300,000 individuals.[8]
history  iron-age  mediterranean  the-classics  demographics  fertility  data  europe  population  measurement  volo-avolo  estimate  wiki  reference  article  conquest-empire  migration  canon  scale  archaeology  multi  broad-econ  pdf  study  survey  debate  uncertainty  walter-scheidel  vaclav-smil  urban  military  economics  labor  time-series  embodied  health  density  malthus  letters  urban-rural  database  list  antiquity  medieval  early-modern  mostly-modern  time  sequential  MENA  the-great-west-whale  china  asia  sinosphere  occident  orient  japan  britain  germanic  gallic  summary  big-picture  objektbuch  confidence  sapiens  anthropology  methodology  farmers-and-foragers  genetics  genomics  chart
august 2017 by nhaliday
Introduction to Scaling Laws
http://galileo.phys.virginia.edu/classes/304/scaling.pdf

Galileo’s Discovery of Scaling Laws: https://www.mtholyoke.edu/~mpeterso/classes/galileo/scaling8.pdf
Days 1 and 2 of Two New Sciences

An example of such an insight is “the surface of a small solid is comparatively greater than that of a large one” because the surface goes like the square of a linear dimension, but the volume goes like the cube.5 Thus as one scales down macroscopic objects, forces on their surfaces like viscous drag become relatively more important, and bulk forces like weight become relatively less important. Galileo uses this idea on the First Day in the context of resistance in free fall, as an explanation for why similar objects of different size do not fall exactly together, but the smaller one lags behind.
nibble  org:junk  exposition  lecture-notes  physics  mechanics  street-fighting  problem-solving  scale  magnitude  estimate  fermi  mental-math  calculation  nitty-gritty  multi  scitariat  org:bleg  lens  tutorial  guide  ground-up  tricki  skeleton  list  cheatsheet  identity  levers  hi-order-bits  yoga  metabuch  pdf  article  essay  history  early-modern  europe  the-great-west-whale  science  the-trenches  discovery  fluid  architecture  oceans  giants  tidbits  elegance
august 2017 by nhaliday
Diophantine approximation - Wikipedia
- rationals perfectly approximated by themselves, badly approximated (eps>1/bq) by other rationals
- irrationals well-approximated (eps~1/q^2) by rationals:
https://en.wikipedia.org/wiki/Dirichlet%27s_approximation_theorem
nibble  wiki  reference  math  math.NT  approximation  accuracy  levers  pigeonhole-markov  multi  tidbits  discrete  rounding  estimate  tightness  algebra
august 2017 by nhaliday
Subgradients - S. Boyd and L. Vandenberghe
If f is convex and x ∈ int dom f, then ∂f(x) is nonempty and bounded. To establish that ∂f(x) ≠ ∅, we apply the supporting hyperplane theorem to the convex set epi f at the boundary point (x, f(x)), ...
pdf  nibble  lecture-notes  acm  optimization  curvature  math.CA  estimate  linearity  differential  existence  proofs  exposition  atoms  math  marginal  convexity-curvature
august 2017 by nhaliday
How to estimate distance using your finger | Outdoor Herbivore Blog
1. Hold your right arm out directly in front of you, elbow straight, thumb upright.
2. Align your thumb with one eye closed so that it covers (or aligns) the distant object. Point marked X in the drawing.
3. Do not move your head, arm or thumb, but switch eyes, so that your open eye is now closed and the other eye is open. Observe closely where the object now appears with the other open eye. Your thumb should appear to have moved to some other point: no longer in front of the object. This new point is marked as Y in the drawing.
4. Estimate this displacement XY, by equating it to the estimated size of something you are familiar with (height of tree, building width, length of a car, power line poles, distance between nearby objects). In this case, the distant barn is estimated to be 100′ wide. It appears 5 barn widths could fit this displacement, or 500 feet. Now multiply that figure by 10 (the ratio of the length of your arm to the distance between your eyes), and you get the distance between you and the thicket of blueberry bushes — 5000 feet away(about 1 mile).

- Basically uses parallax (similar triangles) with each eye.
- When they say to compare apparent shift to known distance, won't that scale with the unknown distance? The example uses width of an object at the point whose distance is being estimated.

per here: https://www.trails.com/how_26316_estimate-distances-outdoors.html
Select a distant object that the width can be accurately determined. For example, use a large rock outcropping. Estimate the width of the rock. Use 200 feet wide as an example here.
outdoors  human-bean  embodied  embodied-pack  visuo  spatial  measurement  lifehack  howto  navigation  prepping  survival  objektbuch  multi  measure  estimate
august 2017 by nhaliday
How accurate are population forecasts?
2 The Accuracy of Past Projections: https://www.nap.edu/read/9828/chapter/4
good ebook:
Beyond Six Billion: Forecasting the World's Population (2000)
Appendix A: Computer Software Packages for Projecting Population
PDE Population Projections looks most relevant for my interests but it's also *ancient*
https://applieddemogtoolbox.github.io/Toolbox/
This Applied Demography Toolbox is a collection of applied demography computer programs, scripts, spreadsheets, databases and texts.

How Accurate Are the United Nations World Population Projections?: http://pages.stern.nyu.edu/~dbackus/BCH/demography/Keilman_JDR_98.pdf

cf. Razib on this: https://pinboard.in/u:nhaliday/b:d63e6df859e8
news  org:lite  prediction  meta:prediction  tetlock  demographics  population  demographic-transition  fertility  islam  world  developing-world  africa  europe  multi  track-record  accuracy  org:ngo  pdf  study  sociology  measurement  volo-avolo  methodology  estimate  data-science  error  wire-guided  priors-posteriors  books  guide  howto  software  tools  recommendations  libraries  gnxp  scitariat  org:gov  org:davos
july 2017 by nhaliday
Harmonic mean - Wikipedia
The harmonic mean is a Schur-concave function, and dominated by the minimum of its arguments, in the sense that for any positive set of arguments, {\displaystyle \min(x_{1}\ldots x_{n})\leq H(x_{1}\ldots x_{n})\leq n\min(x_{1}\ldots x_{n})} . Thus, the harmonic mean cannot be made arbitrarily large by changing some values to bigger ones (while having at least one value unchanged).

more generally, for the weighted mean w/ Pr(x_i)=t_i, H(x1,...,xn) <= x_i/t_i
nibble  math  properties  estimate  concept  definition  wiki  reference  extrema  magnitude  expectancy  metrics  ground-up
july 2017 by nhaliday
Educational Romanticism & Economic Development | pseudoerasmus
deleeted

https://archive.is/p5hRA

Did Nations that Boosted Education Grow Faster?: http://econlog.econlib.org/archives/2012/10/did_nations_tha.html
On average, no relationship. The trendline points down slightly, but for the time being let's just call it a draw. It's a well-known fact that countries that started the 1960's with high education levels grew faster (example), but this graph is about something different. This graph shows that countries that increased their education levels did not grow faster.

https://archive.is/kjxqp

https://archive.is/3YPic

http://hanushek.stanford.edu/publications/schooling-educational-achievement-and-latin-american-growth-puzzle

The Case Against Education: What's Taking So Long, Bryan Caplan: http://econlog.econlib.org/archives/2015/03/the_case_agains_9.html

The World Might Be Better Off Without College for Everyone: https://www.theatlantic.com/magazine/archive/2018/01/whats-college-good-for/546590/
Students don't seem to be getting much out of higher education.
- Bryan Caplan

College: Capital or Signal?: http://www.economicmanblog.com/2017/02/25/college-capital-or-signal/
After his review of the literature, Caplan concludes that roughly 80% of the earnings effect from college comes from signalling, with only 20% the result of skill building. Put this together with his earlier observations about the private returns to college education, along with its exploding cost, and Caplan thinks that the social returns are negative. The policy implications of this will come as very bitter medicine for friends of Bernie Sanders.

Doubting the Null Hypothesis: http://www.arnoldkling.com/blog/doubting-the-null-hypothesis/

ballpark: 50% signaling, 30% selection, 20% addition to human capital
more signaling in art history, more human capital in engineering, more selection in philosophy

Econ Duel! Is Education Signaling or Skill Building?: http://marginalrevolution.com/marginalrevolution/2016/03/econ-duel-is-education-signaling-or-skill-building.html
Marginal Revolution University has a brand new feature, Econ Duel! Our first Econ Duel features Tyler and me debating the question, Is education more about signaling or skill building?

Against Tulip Subsidies: https://slatestarcodex.com/2015/06/06/against-tulip-subsidies/

https://nintil.com/2018/02/05/notes-on-the-case-against-education/

https://www.nationalreview.com/magazine/2018-02-19-0000/bryan-caplan-case-against-education-review

Most American public school kids are low-income; about half are non-white; most are fairly low skilled academically. For most American kids, the majority of the waking hours they spend not engaged with electronic media are at school; the majority of their in-person relationships are at school; the most important relationships they have with an adult who is not their parent is with their teacher. For their parents, the most important in-person source of community is also their kids’ school. Young people need adult mirrors, models, mentors, and in an earlier era these might have been provided by extended families, but in our own era this all falls upon schools.

Caplan gestures towards work and earlier labor force participation as alternatives to school for many if not all kids. And I empathize: the years that I would point to as making me who I am were ones where I was working, not studying. But they were years spent working in schools, as a teacher or assistant. If schools did not exist, is there an alternative that we genuinely believe would arise to draw young people into the life of their community?

...

It is not an accident that the state that spends the least on education is Utah, where the LDS church can take up some of the slack for schools, while next door Wyoming spends almost the most of any state at \$16,000 per student. Education is now the one surviving binding principle of the society as a whole, the one black box everyone will agree to, and so while you can press for less subsidization of education by government, and for privatization of costs, as Caplan does, there’s really nothing people can substitute for it. This is partially about signaling, sure, but it’s also because outside of schools and a few religious enclaves our society is but a darkling plain beset by winds.

This doesn’t mean that we should leave Caplan’s critique on the shelf. Much of education is focused on an insane, zero-sum race for finite rewards. Much of schooling does push kids, parents, schools, and school systems towards a solution ad absurdum, where anything less than 100 percent of kids headed to a doctorate and the big coding job in the sky is a sign of failure of everyone concerned.

But let’s approach this with an eye towards the limits of the possible and the reality of diminishing returns.

https://westhunt.wordpress.com/2018/01/27/poison-ivy-halls/
https://westhunt.wordpress.com/2018/01/27/poison-ivy-halls/#comment-101293
The real reason the left would support Moander: the usual reason. because he’s an enemy.

https://westhunt.wordpress.com/2018/02/01/bright-college-days-part-i/
I have a problem in thinking about education, since my preferences and personal educational experience are atypical, so I can’t just gut it out. On the other hand, knowing that puts me ahead of a lot of people that seem convinced that all real people, including all Arab cabdrivers, think and feel just as they do.

One important fact, relevant to this review. I don’t like Caplan. I think he doesn’t understand – can’t understand – human nature, and although that sometimes confers a different and interesting perspective, it’s not a royal road to truth. Nor would I want to share a foxhole with him: I don’t trust him. So if I say that I agree with some parts of this book, you should believe me.

...

Caplan doesn’t talk about possible ways of improving knowledge acquisition and retention. Maybe he thinks that’s impossible, and he may be right, at least within a conventional universe of possibilities. That’s a bit outside of his thesis, anyhow. Me it interests.

He dismisses objections from educational psychologists who claim that studying a subject improves you in subtle ways even after you forget all of it. I too find that hard to believe. On the other hand, it looks to me as if poorly-digested fragments of information picked up in college have some effect on public policy later in life: it is no coincidence that most prominent people in public life (at a given moment) share a lot of the same ideas. People are vaguely remembering the same crap from the same sources, or related sources. It’s correlated crap, which has a much stronger effect than random crap.

These widespread new ideas are usually wrong. They come from somewhere – in part, from higher education. Along this line, Caplan thinks that college has only a weak ideological effect on students. I don’t believe he is correct. In part, this is because most people use a shifting standard: what’s liberal or conservative gets redefined over time. At any given time a population is roughly half left and half right – but the content of those labels changes a lot. There’s a shift.

https://westhunt.wordpress.com/2018/02/01/bright-college-days-part-i/#comment-101492
I put it this way, a while ago: “When you think about it, falsehoods, stupid crap, make the best group identifiers, because anyone might agree with you when you’re obviously right. Signing up to clear nonsense is a better test of group loyalty. A true friend is with you when you’re wrong. Ideally, not just wrong, but barking mad, rolling around in your own vomit wrong.”
--
You just explained the Credo quia absurdum doctrine. I always wondered if it was nonsense. It is not.
--
Someone on twitter caught it first – got all the way to “sliding down the razor blade of life”. Which I explained is now called “transitioning”

What Catholics believe: https://theweek.com/articles/781925/what-catholics-believe
We believe all of these things, fantastical as they may sound, and we believe them for what we consider good reasons, well attested by history, consistent with the most exacting standards of logic. We will profess them in this place of wrath and tears until the extraordinary event referenced above, for which men and women have hoped and prayed for nearly 2,000 years, comes to pass.

https://westhunt.wordpress.com/2018/02/05/bright-college-days-part-ii/
According to Caplan, employers are looking for conformity, conscientiousness, and intelligence. They use completion of high school, or completion of college as a sign of conformity and conscientiousness. College certainly looks as if it’s mostly signaling, and it’s hugely expensive signaling, in terms of college costs and foregone earnings.

But inserting conformity into the merit function is tricky: things become important signals… because they’re important signals. Otherwise useful actions are contraindicated because they’re “not done”. For example, test scores convey useful information. They could help show that an applicant is smart even though he attended a mediocre school – the same role they play in college admissions. But employers seldom request test scores, and although applicants may provide them, few do. Caplan says ” The word on the street… [more]
econotariat  pseudoE  broad-econ  economics  econometrics  growth-econ  education  human-capital  labor  correlation  null-result  world  developing-world  commentary  spearhead  garett-jones  twitter  social  pic  discussion  econ-metrics  rindermann-thompson  causation  endo-exo  biodet  data  chart  knowledge  article  wealth-of-nations  latin-america  study  path-dependence  divergence  🎩  curvature  microfoundations  multi  convexity-curvature  nonlinearity  hanushek  volo-avolo  endogenous-exogenous  backup  pdf  people  policy  monetary-fiscal  wonkish  cracker-econ  news  org:mag  local-global  higher-ed  impetus  signaling  rhetoric  contrarianism  domestication  propaganda  ratty  hanson  books  review  recommendations  distribution  externalities  cost-benefit  summary  natural-experiment  critique  rent-seeking  mobility  supply-demand  intervention  shift  social-choice  government  incentives  interests  q-n-a  street-fighting  objektbuch  X-not-about-Y  marginal-rev  c:***  qra  info-econ  info-dynamics  org:econlib  yvain  ssc  politics  medicine  stories
april 2017 by nhaliday
Malthus in the Bedroom: Birth Spacing as Birth Control in Pre-Transition England | SpringerLink
Randomness in the Bedroom: There Is No Evidence for Fertility Control in Pre-Industrial England: https://link.springer.com/article/10.1007/s13524-019-00786-2
- Gregory Clark et al.

https://archive.is/HUYPf
both cause and effect of England not being France , which lowered fertility significantly already in the 18th century, I believe largely through anal sex and coitus interruptus
--
Is there a source I can check on that? That's almost too French to be true. Lol.
study  anthropology  sociology  britain  history  early-modern  demographics  fertility  demographic-transition  sex  class  s-factor  spearhead  gregory-clark  malthus  broad-econ  multi  critique  methodology  gotchas  intricacy  stats  estimate  ratty  unaffiliated  twitter  social  commentary  backup  europe  gallic  idk  sexuality
april 2017 by nhaliday
Hoeffding’s Inequality
basic idea of standard pf: bound e^{tX} by line segment (convexity) then use Taylor expansion (in p = b/(b-a), the fraction of range to right of 0) of logarithm
pdf  lecture-notes  exposition  nibble  concentration-of-measure  estimate  proofs  ground-up  acm  probability  series  s:null
february 2017 by nhaliday
st.statistics - Lower bound for sum of binomial coefficients? - MathOverflow
- basically approximate w/ geometric sum (which scales as final term) and you can get it up to O(1) factor
- not good enough for many applications (want 1+o(1) approx.)
- Stirling can also give bound to constant factor precision w/ more calculation I believe
- tighter bound at Section 7.3 here: http://webbuild.knu.ac.kr/~trj/Combin/matousek-vondrak-prob-ln.pdf
q-n-a  overflow  nibble  math  math.CO  estimate  tidbits  magnitude  concentration-of-measure  stirling  binomial  metabuch  tricki  multi  tightness  pdf  lecture-notes  exposition  probability  probabilistic-method  yoga
february 2017 by nhaliday
Consider the following statements:
1. The shape with the largest volume enclosed by a given surface area is the n-dimensional sphere.
2. A marginal or sum of log-concave distributions is log-concave.
3. Any Lipschitz function of a standard n-dimensional Gaussian distribution concentrates around its mean.
What do these all have in common? Despite being fairly non-trivial and deep results, they all can be proved in less than half of a page using the Prékopa–Leindler inequality.

ie, Brunn-Minkowski
acmtariat  clever-rats  ratty  math  acm  geometry  measure  math.MG  estimate  distribution  concentration-of-measure  smoothness  regularity  org:bleg  nibble  brunn-minkowski  curvature  convexity-curvature
february 2017 by nhaliday
bounds - What is the variance of the maximum of a sample? - Cross Validated
- sum of variances is always a bound
- can't do better even for iid Bernoulli
- looks like nice argument from well-known probabilist (using E[(X-Y)^2] = 2Var X), but not clear to me how he gets to sum_i instead of sum_{i,j} in the union bound?
edit: argument is that, for j = argmax_k Y_k, we have r < X_i - Y_j <= X_i - Y_i for all i, including i = argmax_k X_k
- different proof here (later pages): http://www.ism.ac.jp/editsec/aism/pdf/047_1_0185.pdf
Var(X_n:n) <= sum Var(X_k:n) + 2 sum_{i < j} Cov(X_i:n, X_j:n) = Var(sum X_k:n) = Var(sum X_k) = nσ^2
why are the covariances nonnegative? (are they?). intuitively seems true.
- for that, see https://pinboard.in/u:nhaliday/b:ed4466204bb1
- note that this proof shows more generally that sum Var(X_k:n) <= sum Var(X_k)
- apparently that holds for dependent X_k too? http://mathoverflow.net/a/96943/20644
q-n-a  overflow  stats  acm  distribution  tails  bias-variance  moments  estimate  magnitude  probability  iidness  tidbits  concentration-of-measure  multi  orders  levers  extrema  nibble  bonferroni  coarse-fine  expert  symmetry  s:*  expert-experience  proofs
february 2017 by nhaliday
Energy of Seawater Desalination
0.66 kcal / liter is the minimum energy required to desalination of one liter of seawater, regardless of the technology applied to the process.
infrastructure  explanation  physics  thermo  objektbuch  data  lower-bounds  chemistry  the-world-is-just-atoms  geoengineering  phys-energy  nibble  oceans  h2o  applications  estimate  🔬  energy-resources  biophysical-econ  stylized-facts  ideas  fluid  volo-avolo
february 2017 by nhaliday
Embryo editing for intelligence - Gwern.net
My hunch is CRISPR/Cas9 will not play a big role in intelligence enhancement. You'd have to edit so many loci b/c of small effect sizes, increasing errors. Embryo selection is much more promising. Peoples with high avg genetic values, of course, have an in-built advantage there.
ratty  gwern  enhancement  scaling-up  genetics  genomics  iq  🌞  CRISPR  futurism  biodet  new-religion  nibble  intervention  🔬  behavioral-gen  faq  chart  ideas  article  multi  twitter  social  commentary  gnon  unaffiliated  prediction  accuracy  technology  QTL  biotech  selection  comparison  scale  magnitude  hard-tech  skunkworks  speedometer  abortion-contraception-embryo  estimate
february 2017 by nhaliday
The Brunn-Minkowski Inequality | The n-Category Café
For instance, this happens in the plane when A is a horizontal line segment and B is a vertical line segment. There’s obviously no hope of getting an equation for Vol(A+B) in terms of Vol(A) and Vol(B). But this example suggests that we might be able to get an inequality, stating that Vol(A+B) is at least as big as some function of Vol(A) and Vol(B).

The Brunn-Minkowski inequality does this, but it’s really about linearized volume, Vol^{1/n}, rather than volume itself. If length is measured in metres then so is Vol^{1/n}.

...

Nice post, Tom. To readers whose background isn’t in certain areas of geometry and analysis, it’s not obvious that the Brunn–Minkowski inequality is more than a curiosity, the proof of the isoperimetric inequality notwithstanding. So let me add that Brunn–Minkowski is an absolutely vital tool in many parts of geometry, analysis, and probability theory, with extremely diverse applications. Gardner’s survey is a great place to start, but by no means exhaustive.

I’ll also add a couple remarks about regularity issues. You point out that Brunn–Minkowski holds “in the vast generality of measurable sets”, but it may not be initially obvious that this needs to be interpreted as “when A, B, and A+B are all Lebesgue measurable”, since A+B need not be measurable when A and B are (although you can modify the definition of A+B to work for arbitrary measurable A and B; this is discussed by Gardner).
mathtariat  math  estimate  exposition  geometry  math.MG  measure  links  regularity  survey  papers  org:bleg  nibble  homogeneity  brunn-minkowski  curvature  convexity-curvature
february 2017 by nhaliday
per page:    204080120160

Copy this bookmark: