shas3

I think you forget that you have to completely set aside human assumptions.

How will the aliens even know they are looking at signals encoding video and audio? Do aliens have technology and knowhow in the RF regime? Will they even be able to receive, store, or analyze the TV waveforms? Do they have the notion of amplitude and frequency modulation? Also, how will they know that the signal encodes video? Absent ground truth, low entropy (or equivalent measures like bigrams, Markov models, etc.) is the only way one can determine that a sequence is a language. Even if they are able to convert the RF signal to some low-entropy sequence of symbols, there is a very real possibility of them going off on a tangent: mapping the information to a completely different language.

A real world example is the symbols of Indus valley civilization. Several decades into the information age, and about a century since the seals were discovered, researchers haven't even conclusively established that the symbols encode a language.

The story sorts this out. Absent an interpreter, you necessarily need to be able to talk to someone who knows the other language so that you can query them, etc. Realistically, there's no way to intercept or interpret communications as the protocols are almost impossible to figure out (that is another layer of 'language'). Let say an alien intercepts a TV signal, can they convert it back to video? It is almost impossible that they will be able to infer the entire set of protocols: signal -> video, aligning video, etc. Especially given that aliens may be so different from us that their technologies may have developed in an orthogonal direction to our own.

[spoiler] Also, the story clearly has a theme of a person being a helpless observer of their own life (prescience or not (see Gwern's post [0]). The movie has a time-travel twist that's very, to use an indian term, "masala".

[0] https://www.gwern.net/Story%20Of%20Your%20Life

A counterpoint to this is Derek Abbott's "The Reasomable Ineffectiveness of Mathematics." Personally, I think domains with low-N (dimensions and sample sizes) and high complexity remain beyond the optimistic expectations of Eugene Wigner. https://pdfs.semanticscholar.org/462d/7b6b1ee8243b6aa8897be3...

Yeah! And they use "feature engineering" where "feature selection" would suffice. The hard part of feature engineering is usually not selection, but creatively generating features based on expert knowledge, etc. Also, something like a wrapper would be an overkill if you had roughly, #data/class > #features and used a relatively overfitting-resistant classifier like SVM, regularized regression, etc. (more rigorous statement of this would involve VC-dimension).

Richard P. Feynman (who else!) had a scathing critique of this: http://calteches.library.caltech.edu/2362/1/feynman.pdf

To quote him, in reference to new mathematics of the post-Sputnik era,

"In regard to this question of words, there is also in the new mathematics books a great deal of talk about the value of precise language - such things as that one must be very careful to distinguish a number from a numeral and, in general, a symbol from the object that it represents. The real problem in speech is not precise language. The problem is clear language. The desire is to have the idea clearly communicated to the other person. It is only necessary to be precise when there is some doubt as to the meaning of a phrase, and then the precision should be put in the place where the doubt exists."

I think we should look at the correct operation T on the set of all computer scientists. If T is an operation that restricts the set to computer scientists in the field of theoretical computer science (TCS), then the opposite seems to hold as mentioned in this post, which talks about TCS math being more rigorous than applied math: https://windowsontheory.org/2014/10/12/applied-mathematician...

If, loosely speaking, T maps you to 'average computer scientist', then the situation is different. And so on.

So depending on the type of T, subset or groups of computer scientists or metrics you are looking at, Lamport's observations hold water.

This is not implausible. Maritime Southeast Asia had trading relationships [1] with aboriginal Australians in the northern coast of Australia. It is not too surprising (still very interesting) if a ship every now and then made it to New Zealand or further. The main difference between them and the European Colonial Age is that most ships did not return. Note that there was a long Indian 'colonial age' in Southeast Asia, though it was more cultural and trade exchange with naval influence rather than remote hegemony. Likewise the Western Coast of India had maritime activity with the civilizations of Middle East, Europe, and Africa, from as long ago as the Indus Civilization. The point is that coastal Indians were reasonably adept sailors and maritime trade played a major role in South Indian and Western Indian civilization(s) and culture(s). Over several thousand years, it is not too unlikely that a few ships made their way to Australia, etc.

[1] http://blogs.bl.uk/untoldlives/2012/05/australian-aborigines...

Very cool! However, I think the author should have spent a a few more words and figures to distinguish support vector machines from standard perceptrons. Maximum margin classification and the definition of 'support vectors,' in my experience, helps demystify the algorithm.

Yes and communities and networks of merchants (e.g. Sogdians) ensured trust and fidelity in delivery of goods and money. A remnant of the Silk Road is the continuing practice of 'hawala': https://en.wikipedia.org/wiki/Hawala

There are only four South Indian states. I am aware of caste-based quotas, but not gender-based ones. Can you provide a source for this?

It was very difficult to do anything other than engineering, medicine, finance, law, and science in India pre-2000s. Then, with the software and internet and mobile revolutions, in the late 90s and 2000s, software and electronics were substantially more lucrative than pure science or other branches of engineering. Things are becoming more like the US nowadays- with liberal arts being a viable option as the middle class is becoming richer and as children of the first wave of software nouveau riche come of age and their wealth providing some cushion for the liberal-arts and creative types.

There were not enough well-paying jobs (copywriter, creative, real estate agent, retail manager, etc. like in the US) like in the US for people with generic qualifications like English literature, etc. One way out for people with those qualifications was to write banking or governmental entrance exams for jobs. But these too leaned quantitative and had math components making them harder for liberal arts graduates.

From what I've heard from the few people I've met from USSR and China, the situation was similar there too.

If you are interested in Indian women scientists and engineers, there is a nice compilation (a bit tiresome to read, but worth it, IMO) of biographical essays called 'Lilavati's Daughters' https://ia800402.us.archive.org/33/items/LilavatisDaughters-...

My perception of it, growing up plugged in to the Indian science and engineering community is this: Indian Government organizations (essentially, the major employers till the 1990s) were quite progressive when it came to women in the workplace than you'd think based on population statistics of women-at-work, etc. The reason for this imbalance is inequality: millions of women suffering terrible inequality in the lower socio-economic and lower-caste segments of the population offset gains on the higher end.

Yeah it looks more like the math component of an engineering exam. If people look at this and think "oh cool, I can do well in this, I will become a mathematician," will be in for a rude shock when they take Grad School Real Analysis 1.

Great stuff!

What's the Wifi range of each of these "Dabbas"? Little tuck shops, bakeries, and tea stalls in India don't always have places to sit or hangout. Would people be able to sit a little distance away at a bus stop or something be able to use Wifi? Would also be cool if you could expand into small restaurants.

This article is from 2016.

Your comment is premised on the notion that immigration vs. locals is a zero-sum game. It is not uniformly so.

Restrictionists claim: "If fewer Mexicans, Indians, and Chinese came to the US, then American citizens would be better off."

There frankly isn't compelling evidence that this will happen. The number of immigrants using up social services is minuscule and the amount of tax coming from 100s of thousands of H1Bs (and illegals/undocumenteds) is all going into social services used by veterans, American seniors, etc.

On top of this, because immigrants are usually the cream of their own societies and/or have taken big risks coming to the US, themselves and their children contribute via starting new businesses, etc.

This notion that immigration is part of a zero-sum game is likely flawed.

> I guess you missed the part of the article where it explained that the largest group of poor people are children. Do you think they made poor financial choices? And you also missed "Rates of intergenerational income mobility are, in fact, higher in France, Spain, Germany, Canada, Japan, New Zealand and other countries in the world than they are here in the United States."

Are you suggesting that I am not acknowledging bad-luck (the sole reason for childhood poverty) as a factor? Because, I do.

As to Germany, Japan, etc. there are many reasons why those countries are better off. "Germany does this, their poverty is lower, so America should do this," isn't a valid argument for anything.

> Blaming them isn't one of them.

Blaming them (i.e. chiding them) doesn't serve any purpose. But helping them modify behavior, educating them, etc. does. The Government should acknowledge this.

You wouldn't if you had compassion. Holding compassion and objectivity in your head simultaneously would enable you to both help the person and acknowledge that the person may have made poor choices and decisions to get them to where they are. If you ignore that people make poor choices, you are doing a disservice to the poor, since you won't then be able to address this cause/aspect of poverty.

Going against the grain, I don't think Chaffetz meant it literally.

While he may or may not have been taking an extremist Right ideological view of poverty, the OP article takes another extreme ideological view, that poverty is inevitable and individuals can do little to avoid it.

While it sounds heartless, I think there is some truth to the assertion that sound financial choices can lift you up from poverty. Duflo, Banerjee, etc. from the randomista economists crowd have shown that poor people are sometimes likely to spend extra cash in suboptimal ways (TV ownership when they need extra calories, etc.). http://economics.mit.edu/files/530

This includes buying health insurance. That health insurance is too expensive is unfortunate, and I think GOP earnestly believes that their plan makes it cheaper for people to buy health insurance. This is not unlike how Obama and Democrats implemented their healthcare overhaul with the best of intentions, yet it lead to expensive premiums, etc.

While we are loathe to admit it, poverty is a result of both poor choices and unavoidable bad-luck. Like the common debate of 'nature vs. nurture', it is hard to delineate bad-luck and poor-choices as causes leading to poverty.

At the end of the day, we should have compassion and understand that many things that affect our situation are beyond our control, and yet not be afraid to consider honestly and objectively, the various causes of poverty.

MSM bias reflects primarily in opinions and reporting decisions. Most important events are reported by MSM in an unbiased fashion, as are (often) investigative reports. Vox does almost no original reporting and is just in the business of mixing up other news reports and opinionizing. In this, they are highly biased and to a large extent, they cherry pick news and views that conform with extreme liberal dogmas.

A good counterpoint to Vox is Real Clear Politics. Their front page features an equal number of direct links to opinion and news from both sides of the political spectrum. They also don't corrupt these links with rambling infantilzing '*splaining'.

Can someone clarify why the author talks specifically about Trump's America? This kind of excessive profiling at ports of entry happened under GWB and Obama too. I have heard many anecdotes in the last 15 years. Is there a provable and statistically significant increase in intensity or number of such incidents? In its absence, we should settle for a prior for likelihood-of-being-profiled that is defined by pre-Trump rates. We can't arbitrarily update this prior based on our political opinions.

I think this is already happening. In fact most AI pessimists don't consider that "man + machine >> machine." Consider chess. Kasparov's defeat happened in mid-1990s. Machine is considered > man. Yet, in freestyle chess, where players are allowed to consult machines, we have achieved ever greater heights of performance. In chess today, any sufficiently advanced chess engine can be incorporated into a (man+chess-engine) player to perform better. Similar things will happen in Go, Poker, etc. The handicap of not using a computer to inform your moves exists to make things more competitive, it doesn't mean that we are worse off!

Taking this further, most technology we use daily (word processors, email, the Internet, etc.) are enhancements of the man+machine organism and enhance our performance as well as the 'machine''s.

Who said anything about human form factor? I said non-forklift, I meant things like: last mile delivery of furniture and appliances, packing and moving, etc. where most of the lifting is manual. These are situations where things are more unstructured than a warehouse. The kind of problems for which Boston Dynamics seems to be building robots.

I agree. The numbers should be scaled to # of customers and/or population. To some extent, this is addressed in the text. ATMs reduced the cost of operating a branch, so banks opened more branches. It's easy to see that number of tellers has kept up with population growth. US population was ~200e6 in 1970 and is now ~300e6. From that AEI chart, tellers have grown almost as fast.

(Made some edits to this comment after I realized I mistook the tellers graph for the ATMs.)

I'm sure there is a good critique of that Frey and Osborne paper. Epistemologically, it sounds weird to assign 'probabilities of automation.' I don't have the time to read the paper in detail, but here are a few things:

1. Automation doesn't imply that jobs go away. Classic example, deposits and withdrawals were automated by ATMs. Yet, the number of cashiers increased.

2. Automation by itself is not sufficient. Will we reach a point where it is cheaper to automate a job than hire a person? Think of non-forklift lifting/moving of heavy stuff, we don't really need people doing this manually, robots should be able to do it. But we see with the Boston Dynamics robots that we are many years or decades away from commoditizing lifting robots.

3. 'useless' class already exists. Employment and usefulness aren't necessarily the same. There are many millions of people in jobs that are 'useless' by some metrics (gas station attendant, supermarket cashiers, etc).

> Just an aside, I think it's insane that women are pressured (or feel pressured) to wear heels. Is this common in California?

FTFY.

An earnest question I've heard from people trying to switch is: what's a good benchmark for testing where you stand in terms of coding skills? How do you calibrate and measure your production-quality coding skills in Scala, C, etc. when you have spent all your life in academia?

I feel Kaggle is sufficient for the exploratory part of data science. But Kaggle's relevance to testing and honing "production quality" code-writing skills is sometimes minimal.

In the larger context of programming and software engineering, scientific programming is fairly easy to code-up (though they are harder to conceptualize and understand mathematically). The coding part of non-CS/non-CE/non-some-parts-of-EE academia is pretty much mostly scientific programming. Yet, production quality code is seldom just scientific programming.

1. Undergrads who are exceptional in one field: Imagine applying this "undergrads aren't skilled workers" to Linus Trovalds in 1990.

2. Masters students with specialized skills and possible prior experience before their Masters.

Uh? SD is a super-generic title. H1B pool can include disproportionate number of fresh graduates (BS or MS) because OPT program gives them the flexibility to wait till April comes around to file for H1B. It is almost impossible (with exceptions for researchers and multinational managers/ O1/O2/L1) to hire experienced people from outside the US via H1B. Thus this skewing happens: more jobs for fresh graduates probably go to foreigners than do jobs for people experienced and established in their fields.