That's not what the article is saying. It's saying that atmospheres that contain a non-equilibrium mix of chemicals that would under normal conditions react and turn into something else, are thougt to be a strong indication of the presence of life, since some non-trivial process is needed to maintain that non-equilibrium. Basically one would say "Hey, isn't it really weird that this planet has both X and Y in its atmosphere at the same time, even though those should react and form Z on a time-scale of a few hunderd thousand years? Some sort of exotic process must be continually producing X and Y to maintain this situation, or we wouldn't observe it." But as it points out, you could have perfectly balanced atmospheres containing each of those chemicals separately, so if the planet has compound X in its atmosphere, and it has a moon with compund Y in its atmosphere, then due to our limited resolution, we will only observe a compound object which appears to have both X and Y in its atmosphere at the same time, making it seem to be out of equilibrium.
The alignment of the two objects does not requrie a huge coincidence if one is the moon of the other. Then they could always be close enough that they both occupy a single pixel in the image. No equivalent/opposite cancellation stuff is going on here.
Here is a non-paywalled version (any time you see a paywalled astronomy article, you can always find a free version on arxiv.org).
I've only skimmed the article, but I did not see any mention of doppler shifts in it. I would imagine that a planet+moon system would exhibit a time-dependent doppler shift of the moon atmosphere relative to the planet atmosphere, which would make it possible to disentangle the two. However, for an Earth-Moon situation, with a relative velocity of 1 km/s, the doppler shift will only be lambda/(delta lambda) of about 3e5, which is much smaller than the spectral resolution they are assuming (1600), so perhaps it doesn't work.
Humans have a known vulnerability to various exploits such as appeals to emotion, injection of associations, conditioning, suggestion, and so on, all of which are essential weapons in the advertisement/propaganda toolkit. Sadly, it's much harder to patch people than software, so these vulnerabilities are likely to stay. I therefore think a large part of the blame should lie with those exploiting these vulnerabilities rather than those who are exploited. And to a large extent, this "brain-hacking" is the domain of the Koch Brothers, Big Oil, etc.. So I think it's fair to say that they *are* a large part of the problem.
Well I can't speak for parent, but honestly this has been the case since political power overtook that whole tribal test of strength thing back in the days. Submit a single instance where those who held the highest concentration of resources (money, slaves, oil (crude or olive), land, etc...) didn't use them to get favorable status from those who represented the people and then we'll talk.
Yes, wealth leads to a democratically dispropotionate influence over politics. That's why it matters how skewed the wealth distribution is. The more skewed it is, the larger fraction of power will be in the hands of the few rich. Inequality in the USA is rising, and the problem did not use to be as bad as it is now. In the 70s, the United States had a significantly lower Gini coefficient (though still much higher than most European countries), but it has been rising since then: http://www.americanprogress.or... https://en.wikipedia.org/wiki/...
Inequality is also different from country to country. Again, due to the natural tendency for the rich to dominate, one expects that on average democracies with higher economic equality should be healthier. The USA does quite poorly on metrics of income inequality lately: https://en.wikipedia.org/wiki/...
So I agree that the huge influence of people like the Koch brothers is not that surpising in light of the huge income inequality in today's USA. But it's still scary, and should not be taken for granted. It can be fought, and the most obvious way of fighting it is by reducing the different between rich and poor. Saying that "this has always happened" ignores that the degree to which it has happened has changed and can be changed.
That was an interesting wall of text. I have also dreamed of a distributed replacement for HTTP in order to remove one of the most common reason for the internet being so advertisement-laden: that hosting costs rise in proportion with traffic. Are you or anybody working on implementing this? How much of a latency and disk space overhead would this have? Freenet has similar features, but has enormous overhead (but much of that is to achieve anonymity). How much would it be hampered by current "you're a client, not a server or peer" asymmetric internet connections?
Botulinium toxin is the most deadly poison known, with an LD-50 of about 1.4 ng/kg (though that's when injected - I'm not sure what it is when drunk. I think it's broken down pretty rapidly outdoors too). With that, you would need about 15 mg to make all 143 million liters deadly (for 68 kg persons drinking 1 liter each). It it also mass produced for use in cosmetics, so it's easier to get than Batrachotoxin. Thouh as I said I think it's broken down pretty quickly under normal conditions (was it by exposure to oxygen?), so it's not anything to worry about in this context. But it's a good illustration of how potent poisons exist.
It *might* be as you say, that the US is run according to the preferences of the upper middle class, with the very rich just agreeing with them. Or it might be the other way around.... They are using it as a proxy for the opinions of those much richer because their data set does not include the real economic elite.
No, it can't be the other way around. If a group of 30 million and a group of 30000 people agree on policy, you can't say that our policies are dominated by a small group of just 30000.
If you want to find out who out of two groups decides, you can't look at the cases where they agree. Those cases don't tell you anything about who calls the shots. You have to look at the ones where they disagree. If, for example a 30000 people group always wins when they disagree with a 30 milliion group, then either the 30000 group decides, or its interests align perfectly with a third group that does decide. In both cases, the 30 million group does not actually have any power. Just because a group is small doesn't mean it doesn't dominate over larger groups. For example, in a dictatorship, a group of 1 (or a handful) dominate over the whole rest of the population. So yes, it *can* be the other way around. That's why one needs to test these things. But the current article can't test this because they can't distinguish between the 90% quantile and the 99.9% quantile.
The political outrage is over the fact that it is a tiny group that supposedly dominates, the group isn't tiny if 30 million agree with it.
If 10% of the population dominates, that's also a problem, though less so.
But that's not the worst of it. The paper simply ignores the large number of people and issues on which there is agreement across income groups.
It doesn't ignore them. They enter into the the calculation, but they don't provide any information about the power structure. Only disagreement tells you about that. And just because one group dominates in terms of power does not mean that the other groups are being stomped into the ground. Just because you don't have the ability to influence decisions doesn't mean those decisions have to be something you disagree with. For example, children probably agree with a large fraction of the decisions made by their parents (such as "let's eat"), and even the ones where they don't agree, the parents' decisions are probably for the best of the child. That's the "benevolent dictator" situation. It's not necessarily bad to live in such a situation. But it's not democracy.
I dispute that. That isn't the norm among high earners.
Of course it's the norm, in particular among high income earners. Age alone means that that group has a high turnover, because people clearly don't earn those kinds of high incomes right out of college.
You speak as if the economic mobility in the USA were perfect. it isn't. If your parents belong in the top quintile, then you are likely to end up in the top two quintiles, and very unlikely to end up in the bottom quintile. And similarly the other way around.
Even if it were, the fact that the 90% quantile's preferences correlate more strongly with the richest and the median shows that either they do form a distinct, stable group, or their preferences change while they are in that quantile.
No, that's circular reasoning. The paper didn't show that these groups are groups with distinct interests that have consistent or stable preferences, it simply selected a subset of properties with those criteria under the assumption that these are groups with consistent and stable preferences. If they aren't, they ended up studying correlations among statistical artifacts.
My point wasn't about criticizing the degree of agreement they uncovered, it was about their characterization of "wealthy". The group they call "wealthy" and where they imply domination by an "economic elite" is simply older, educated upper-middle class folks.
I agree that you can't really call them the "economic elite". The poor wealth resolution is a weakness that should be addressed. Still, one would hope that not only the 10% richest would be able to influence political decisions.
They claim that those preferences correlate with those of the "truly wealthy", but even if that correlation were perfect, it wouldn't mean that the US is run according to the wishes of the "truly wealthy", it would mean that the US is run according to the preferences of the upper middle class and that the "truly wealthy" happen to share upper middle class preferences.
I think you're being too quick here. It *might* be as you say, that the US is run according to the preferences of the upper middle class, with the very rich just agreeing with them. Or it might be the other way around. One can test this by looking at what happens when they disagree. If the 90% quantile and the 99% quantile agree 70% of the time, but the 1% quantile has its way for those 30% when they disagree, then clearly the opinion of the 90% quantile does not matter. Conversely, if the 90% quantile has its way when they disagree, then they are the ones calling the shots. If something intermediate happens, then they share power. This needs to be measured.
First of all, high earners don't even exist as a stable group to express preferences, so they can't form an oligarchy. High earners each year are a temporary subset of older professionals and small business owners, people who have a few good years.
I dispute that. That isn't the norm among high earners. Even if it were, the fact that the 90% quantile's preferences correlate more strongly with the richest and the median shows that either they do form a distinct, stable group, or their preferences change while they are in that quantile. In any case, they empirically have different interests than those in lower quantiles, which is all that matters here.
Conversely, it is wrong to identify the preferences of people around the median income with "the majority". The group of people near the median income is no larger than the group of people near the 90th percentile. People around the median income aren't even a useful social group with common interests. For example, assume that the US consisted only of two groups, high income earners and low income earners, each with their own set of preferences, and each equally listened to by politicians, with the median income earner being equally likely to come from either group. If you did the same analysis as in the paper, you'd find the same result: their model would show domination by an economic elite, even though there is none.
The reason why they use the preferences of the median is because some of the theories they are trying to test in this article predict that political decisions should follow the opinion of the median citizen. That requires assumptions, and those assumptions can be wrong. It would be interesting to see the same calculation done for other percentiles, or for everybody below the 90% quantile as a whole. But there are good reasons for believing the 50% quantile to be more representative of the people as a whole than the 90% quantile (and especially the 99% quantile). The income distribution curve is much flatter at that point - i.e. the 49% quantile and the 51% quantile are much more similar to the 50% quantile than the 89% and 91% quantiles are to the 90% quantile. Here is an illustrative (if a bit slow) illustration of the distribution. So many more people have similar lives to the 50% quantile than the 90% quantile, though the latter still isn't as spe
The first a set of independent bivariate analyses, measuring the connection between the preferences of median americans vs. political decisions, the preferences of wealthy americans vs. political decisions and the preferences of lobby organizations vs. political decisions. Like others before them, they found that all of these had strong positive correlations, with wealthy americans having more influence than the preferences of median americans, but not by that much:
That's consistent with the picture you were painting. But as you know, correlation does not mean causation. So the authors went a step further and did a multivariate analysis of all these at the same time. This is the new thing about this study. What they found was that the preferences of median americans have no explanatory power for political decisions - they correlate with them simply because the median american often agrees with rich americans, who have the real influence.
Median americans: 0.03 +- 0.08 (i.e. zero within the error bars)
Wealthy americans: 0.76 +- 0.09
Interest groups: 0.56 +- 0.09
If you only know the preferences of the rich, you can predict political decisions moderately well. If you add the preferences of the most important interest groups, the predictions improve substantially. But if you add the preferences of median americans, the predictions do not improve at all.
As a simplified illustration of such a situation, consider a totalitarian regime with a dictator and a street-sweeper, with the dictator making every decision. If the dictator and street-sweeper happen to agree 75% of the time, it might look like the street-sweeper actually had a lot of influence if you look at him in isolation. But if you analyse him together with the dictator, you discover that the opinion of the street-sweeper does not actually influence the results at all - the dictator's opinion is all that matters.
The article used these findings to evaluate several leading models for how american democracy works:
1. Majoritarian electoral democracy (i..e how the system is supposed to work)
2. Economic elite domination (plutocrachy/oligarchy)
3. Majoritarian pluralism (lobby groups decide, but taken together these are representative of the will of the people)
4. Biased pluralism (lobby groups decide, and represent only a small subset of the population (oligarchy)) They found that hypothesis 1 and 3 were incompatible with the data, while 2 and 4 were supported. So the summary wasn't all that bad.
As a final note: The resolution of their data was quite poor when it comes to wealth, so they could not determine how much of the apparent influence of the 90% quantile individuals come simply from correlation with 99% quantile individuals. It would be interesting to see a follow-up study with higher resolution. It might well turn out that the 90%-99% quantiles have as little influence as the 50% quantile. But at the moment there isn't enough data available to investigate that.
No, grandparent is right. The huge masses you describe are only needed if you assume that the background metric of the universe is unchanged. But if you are in a reference frame co-rotating with the world, then in that frame the whole universe is rotating at extreme speeds. This produces an extreme frame-dragging effect which makes it impossible not to rotate at or nearly at the same velocity as the other objects at that distance - no huge masses needed. This is an example of the more general Mach's principle.
For an example, look at the "rotating polar" example here. That's what you would have in the case where the earth is massless (the massive case is "rotating shwartzschild" on the next page (not to be confused with the Kerr metric), but the mass of the earth is so small compared to the effects of the rotating frame that it's not worth worrying about it). Here you can check that r=const, theta=pi/2 and dphi/dt = omega and dt/dtau = 1 (i.e. an object orbiting with angular velocity omega at distance r) fulfills the geodesic equation. This could be Proxima Centauri, for example. And that's for zero solar masses.
Thanks for the book reference (though the missing pages were very annoying). You have convinced me. I guess it's good my job isn't designing telescope optics.
To answer your question about the telescope: It's f-number is 2.5 at Gregorian focus, and we observe at 148 GHz. The optical transfer function plots I showed had multipole number on the horizontal axis.
The thing that's the self-convolution of the pupil function is the point spread function (g(theta) in my example from a few posts back). For the case of an ideal, top-hat shaped pupil function, the point spread function will fall to zero, and stay zero, at theta = 2*theta_pupil. But the optical transfer function (G(k) in my post) is the fourier transform of the point spread function. And the fourier transform of a function with limited support has unlimited support in fourier space. Hence, while the optical transfer function will never fall to zero in this case (or any other case with a sharp edge to the pupil function), except for occasional zero crossings.
I think I found the figure you you're referring to. It's this one, right? I don't think that figure lets you distinguish small from zero due to its very poor dynamic range. A logarithic second axis would be much more informative.
Here is an example of an MTF from an experiment I've worked on. It looks quite similar to the figure on the Wikipedia page, and by eye one might think that's it's reached zero by 18000 or so. But consider the logarithmic version of the same graph. As you can see, the graph had only fallen by about 20 dB by that point, and even at the end of the figure it's only down by 45 dB or so. So I don't think the Wikipedia figure supports your point.
Then that particular frequency cannot be recovered. But this usually only happens at zero crossings, which make up a vanishingly small fraction of the frequencies involved. Of course when noise is also present, then it's enough for G to be very small rather than exactly zero, and that would kill all the higher frequencies.
I didn't make this explicit, but nowhere does one need to assume that one is looking at a point source (though that helps). I use this for looking at the cosmic microwave background, which is as wide-field as it gets.
The effect of a telescope's point spread function is to convolve the image. A raw image f(x) is turned into f'(x) = int dy f(x-dy) g(y), where g(dx) is the point spread function. By the convolution theorem, a convolution is simply a product in fourier space, so F'(k) = F(k)*G(k), where uppercase functions are fourier-transpformed versions of lowercase ones, and k is the wavenumber. From this you can see that recovering the raw, unblurred image (i.e. overcoming the diffraction limit), is just a question of computing F(k) = F'(k)/G(k), i.e. dividing by by the point spread function in fourier space, or deconvoluting it in normal coordinates.
What limits our ability to do so is the presence of noise in the final image. So a more realistic data model is F'(k) = F(k)*G(k) + N(k), where N is the noise, and when we now try to remove the point spread function, we get F_estimate(k) = F'(k)/G(k) + N(k)/G(k) = F(k) + N(k)/G(k). So we get back our unblurred original image, but with an extra noise term. And since G(k) usually falls exponentially with k (meaning that high wavenumbers = small scales in the image are progressively blurred), N(k)/G(k) will grow exponentially with k, making the noise ever greater as one tries to recover smaller and smaller scales. This puts a limit on how small scales one can recover. But as you can see, it is not given just by G(k). I.e. not just by the diffraction limit. It is given by a combination of the telescope's noise level and the point spread function. In the absence of noise we can fully recover all the small scales. And even with noise one can push down a bit below the diffraction limit with enough data. But due to that exponential rise of the noise with wavenumber, it's an extreme case of diminishing returns. It is much cheaper to make the telescope bigger.
Sadly, I don't think that's the case. This is all voluntary agreements between Google and various coorporations that kick in *before* any DMCA stuff. I think what happens is that Google runs video/audio matching programs on behalf of other companies, and when something matches they take it down, notifying the user. The user can then assert that they do in fact have the right to upload the video. Once they do, the video is put back up, and the company is notified. They then file a real DMCA claim. The video is then taken down again, and the user is notified. They can then file a DMCA counterclaim, which would bring the video back but expose them to a lawsuit, or back down, in which case they get a "copyright strike", which leads to the loss of the ability to upload long videos, and eventually being banned from youtube.
I think these voluntary agreements are a perversion of an already pretty nasty law. I've had one of my own videos affected in a somewhat milder fashion: They put advertisements on the video instead of taking it down. That makes it seems like Iæm selling out my viewers to advertisers, but though the video was quite clearly fair use (a video comparing the current and previous world rectord speedruns of a computer game), I would have to consult a lawyer before contesting it, which would take days, and be expensive. The power asymmetry means that Sony etc. can accuse you as much as they want with no worry, while defending oneself is a costly and risky endeavor to normal users.
If this thing is falling from the parachute, that would make it a small, nearby, slow object rather than a larger, faster object further away. Assuming a 7 m tall parachute (including lines, ) + 1 m more of skydiver legs below that, we find that the rock traveled from around the top of the parachute to definitely below the skydiver in 15 frames of video, which gives us a velocity of about 13 m/s relative to the skydiver. Assuming that the parachute has slowed the skydiver down to a landing velocity of 7 m/s, this brings it to a total velocity of 20 m/s.
The terminal velocity of a rock depends on its size. I couldn't find numbers for rocks, but for hailstones (which should fall more slowly due to their lower density (2.5 times lower density gives sqrt(2.5) = 1.6 times lower terminal speed)), I find that one 8 cm in diameter falls at 48 m/s (so we would expect 75 m/s for rock). In general we have rho r^3 = Ar^2v^2 giving v proportional to sqrt(r). So to get a terminal velocity of 20 m/s we need a 0.6 cm diameter pebble.
This doesn't mean that the object necessarily is a nearby small object, but it is plasuible so far.
Another thing that points in the direction of a small object is the way it suddenly appears in the front-facing camera's field of view, quite far from the edge (indicating a very large angular velocity), but then only moves a few pixels the next frame. This camera's field of view seems to be about 60 degrees or so, which would give the object an angular velocity of more than 20 degrees per frame, or 500 degrees per second for the first frame when it appears in the forwards view. For 13 m/s relative to the skydiver, this would put the object at a distance of less than 1.5 m, which is consistent. The object's size is about 3 pixels in diameter at this point (out of a image width of 1280 pixels), which would give it a size of 0.35 cm, which is a bit too small.
According to the NRK article, the object was moving at 300 km/h, or 80 m/s. That would put it at a distance of 9.5 m (still really close!), and a size of 2.2 cm in diameter. Such an object should have a terminal velocity of 40 m/s, which is about half the 80 m/s we assumed. But the uncertainty of these measurements is big enough that that's probably well inside the margin of error. (To get 80 m/s we want a 10 cm diameter object, which might work if we take into acocount that the object is further away on the first frame when it appears in the forward camera than it would have been on average between that and the previous frame).
So based on these back-of-the-envelope non-expert calculations, I'd tentatively say that your pebble-in-the-parachute explanation fits the video just as well as the meteorite explanation. But I'd trust the experts who looked at this video more than me.
Just to clarify: The problem wasn't that you didn't have the right exponent. Exponential is not a collective name for x^2, x^3, x^4, etc. These are called power laws, not exponentials. Exponentials are when your variable is in the exponent, e.g. 2^x, 3^x, 4^x, etc. These start off slowly, but eventually grow extremely quickly. What you probably wanted to say was "increases as some power higher than one of the velocity". Though that doesn't roll off the tounge.
Surely you mean that drag increases quadratically with spead, not exponentially. (And at lower speeds (before the air flow becomes turbulent), it's linear.) If it really were exponential, planes and rockets wouldn't be going anywhere fast.
To illustrate: With quadratic drag, drag is four times greater at 100 km/h than at 50 km/h, and 16 times greater at 200 km/h, and so on, to 400 times greater at 1000 km/h. But if it were exponential, then the progression that matches the first two points would be one which quadruples for every 50 km/h increase in speed. So one would have: 100 km/h: 4 times, 150 km/h: 16 times, 200 km/h: 32 times, 1000 km/h: 1,099,511,627,776 times greater!
Thanks for all the details. I'd just like to point out a few points that won't change the overall point of what you're saying (though surely the overall energy cost per distance of normal and electric cars has been studied in great detail before - it would be nice to see the non-back-of-the-envelope version of this calculation):
Firstly, note that coal power plants (or any other plant that burns fuel) can be made much more efficient than 45% in cold areas where homes need heating, in the form of the combined heat and power plant, which have efficiencies of about 80%. Also, it's not a given that power plants would use coal. "Only" 27% of the worl'd energy use is in the form of coal. The remaining 73% come from cleaner sources of energy, with about 20% being much cleaner. And these numbers change over time too.
The purpose of patents is noble and laudible, but sometimes things don't end up working as intended. For a long while, doctors thought bleeding patiens was the best cure to all ills, but we now know that bleeding doesn't work, and only harms the patient. So despite the purpose of bleeding being ever so noble, what actually came of it was harmful.
Evaluating the overall effect of patents is a lot of work. The book I cited attempts to do so, but a bigger, more scientific investigation is needed. But I think the burden of proof should be on those advocating for patents. Monopolies should be avoided as much as possible (especially monopolies on abstract things like how to solve a problem), so one really should prove empirically that patents both achive their purpose, and that the harm does not outweigh the benefits, before creating any more than necessary.
At least patents have a time limit, though (unlike copyright (when's the next extension scheduled for again?)), so in that aspect they are better than copyright. But a striking example of the problems of patents (software patents in particular) is that what is widely considered to be the best h.264 encoder, the free software program x264, is illegal in many countries due to not paying license fees to the MPEG licensing authority (and this wouldn't work with the free software development model anyway). The same applies to other ffmpeg projects, and stuff in mplayer and many other very popular programs. These do not break copyright, and were not plagiarized from anybody - they just implement public standards.
Slashdot Mirror
That's not what the article is saying. It's saying that atmospheres that contain a non-equilibrium mix of chemicals that would under normal conditions react and turn into something else, are thougt to be a strong indication of the presence of life, since some non-trivial process is needed to maintain that non-equilibrium. Basically one would say "Hey, isn't it really weird that this planet has both X and Y in its atmosphere at the same time, even though those should react and form Z on a time-scale of a few hunderd thousand years? Some sort of exotic process must be continually producing X and Y to maintain this situation, or we wouldn't observe it." But as it points out, you could have perfectly balanced atmospheres containing each of those chemicals separately, so if the planet has compound X in its atmosphere, and it has a moon with compund Y in its atmosphere, then due to our limited resolution, we will only observe a compound object which appears to have both X and Y in its atmosphere at the same time, making it seem to be out of equilibrium.
The alignment of the two objects does not requrie a huge coincidence if one is the moon of the other. Then they could always be close enough that they both occupy a single pixel in the image. No equivalent/opposite cancellation stuff is going on here.
Here is a non-paywalled version (any time you see a paywalled astronomy article, you can always find a free version on arxiv.org).
I've only skimmed the article, but I did not see any mention of doppler shifts in it. I would imagine that a planet+moon system would exhibit a time-dependent doppler shift of the moon atmosphere relative to the planet atmosphere, which would make it possible to disentangle the two. However, for an Earth-Moon situation, with a relative velocity of 1 km/s, the doppler shift will only be lambda/(delta lambda) of about 3e5, which is much smaller than the spectral resolution they are assuming (1600), so perhaps it doesn't work.
All evens out in the end.
[dubious][citation needed]
Humans have a known vulnerability to various exploits such as appeals to emotion, injection of associations, conditioning, suggestion, and so on, all of which are essential weapons in the advertisement/propaganda toolkit. Sadly, it's much harder to patch people than software, so these vulnerabilities are likely to stay. I therefore think a large part of the blame should lie with those exploiting these vulnerabilities rather than those who are exploited. And to a large extent, this "brain-hacking" is the domain of the Koch Brothers, Big Oil, etc.. So I think it's fair to say that they *are* a large part of the problem.
Well I can't speak for parent, but honestly this has been the case since political power overtook that whole tribal test of strength thing back in the days. Submit a single instance where those who held the highest concentration of resources (money, slaves, oil (crude or olive), land, etc...) didn't use them to get favorable status from those who represented the people and then we'll talk.
Yes, wealth leads to a democratically dispropotionate influence over politics. That's why it matters how skewed the wealth distribution is. The more skewed it is, the larger fraction of power will be in the hands of the few rich. Inequality in the USA is rising, and the problem did not use to be as bad as it is now. In the 70s, the United States had a significantly lower Gini coefficient (though still much higher than most European countries), but it has been rising since then:
http://www.americanprogress.or...
https://en.wikipedia.org/wiki/...
Inequality is also different from country to country. Again, due to the natural tendency for the rich to dominate, one expects that on average democracies with higher economic equality should be healthier. The USA does quite poorly on metrics of income inequality lately:
https://en.wikipedia.org/wiki/...
So I agree that the huge influence of people like the Koch brothers is not that surpising in light of the huge income inequality in today's USA. But it's still scary, and should not be taken for granted. It can be fought, and the most obvious way of fighting it is by reducing the different between rich and poor. Saying that "this has always happened" ignores that the degree to which it has happened has changed and can be changed.
Commenting to undu mistaken moderation.
That was an interesting wall of text. I have also dreamed of a distributed replacement for HTTP in order to remove one of the most common reason for the internet being so advertisement-laden: that hosting costs rise in proportion with traffic. Are you or anybody working on implementing this? How much of a latency and disk space overhead would this have? Freenet has similar features, but has enormous overhead (but much of that is to achieve anonymity). How much would it be hampered by current "you're a client, not a server or peer" asymmetric internet connections?
Botulinium toxin is the most deadly poison known, with an LD-50 of about 1.4 ng/kg (though that's when injected - I'm not sure what it is when drunk. I think it's broken down pretty rapidly outdoors too). With that, you would need about 15 mg to make all 143 million liters deadly (for 68 kg persons drinking 1 liter each). It it also mass produced for use in cosmetics, so it's easier to get than Batrachotoxin. Thouh as I said I think it's broken down pretty quickly under normal conditions (was it by exposure to oxygen?), so it's not anything to worry about in this context. But it's a good illustration of how potent poisons exist.
It *might* be as you say, that the US is run according to the preferences of the upper middle class, with the very rich just agreeing with them. Or it might be the other way around. ... They are using it as a proxy for the opinions of those much richer because their data set does not include the real economic elite.
No, it can't be the other way around. If a group of 30 million and a group of 30000 people agree on policy, you can't say that our policies are dominated by a small group of just 30000.
If you want to find out who out of two groups decides, you can't look at the cases where they agree. Those cases don't tell you anything about who calls the shots. You have to look at the ones where they disagree. If, for example a 30000 people group always wins when they disagree with a 30 milliion group, then either the 30000 group decides, or its interests align perfectly with a third group that does decide. In both cases, the 30 million group does not actually have any power. Just because a group is small doesn't mean it doesn't dominate over larger groups. For example, in a dictatorship, a group of 1 (or a handful) dominate over the whole rest of the population. So yes, it *can* be the other way around. That's why one needs to test these things. But the current article can't test this because they can't distinguish between the 90% quantile and the 99.9% quantile.
The political outrage is over the fact that it is a tiny group that supposedly dominates, the group isn't tiny if 30 million agree with it.
If 10% of the population dominates, that's also a problem, though less so.
But that's not the worst of it. The paper simply ignores the large number of people and issues on which there is agreement across income groups.
It doesn't ignore them. They enter into the the calculation, but they don't provide any information about the power structure. Only disagreement tells you about that. And just because one group dominates in terms of power does not mean that the other groups are being stomped into the ground. Just because you don't have the ability to influence decisions doesn't mean those decisions have to be something you disagree with. For example, children probably agree with a large fraction of the decisions made by their parents (such as "let's eat"), and even the ones where they don't agree, the parents' decisions are probably for the best of the child. That's the "benevolent dictator" situation. It's not necessarily bad to live in such a situation. But it's not democracy.
I dispute that. That isn't the norm among high earners.
Of course it's the norm, in particular among high income earners. Age alone means that that group has a high turnover, because people clearly don't earn those kinds of high incomes right out of college.
You speak as if the economic mobility in the USA were perfect. it isn't. If your parents belong in the top quintile, then you are likely to end up in the top two quintiles, and very unlikely to end up in the bottom quintile. And similarly the other way around.
Even if it were, the fact that the 90% quantile's preferences correlate more strongly with the richest and the median shows that either they do form a distinct, stable group, or their preferences change while they are in that quantile.
No, that's circular reasoning. The paper didn't show that these groups are groups with distinct interests that have consistent or stable preferences, it simply selected a subset of properties with those criteria under the assumption that these are groups with consistent and stable preferences. If they aren't, they ended up studying correlations among statistical artifacts.
I don't understand your reasoning here.
My point wasn't about criticizing the degree of agreement they uncovered, it was about their characterization of "wealthy". The group they call "wealthy" and where they imply domination by an "economic elite" is simply older, educated upper-middle class folks.
I agree that you can't really call them the "economic elite". The poor wealth resolution is a weakness that should be addressed. Still, one would hope that not only the 10% richest would be able to influence political decisions.
They claim that those preferences correlate with those of the "truly wealthy", but even if that correlation were perfect, it wouldn't mean that the US is run according to the wishes of the "truly wealthy", it would mean that the US is run according to the preferences of the upper middle class and that the "truly wealthy" happen to share upper middle class preferences.
I think you're being too quick here. It *might* be as you say, that the US is run according to the preferences of the upper middle class, with the very rich just agreeing with them. Or it might be the other way around. One can test this by looking at what happens when they disagree. If the 90% quantile and the 99% quantile agree 70% of the time, but the 1% quantile has its way for those 30% when they disagree, then clearly the opinion of the 90% quantile does not matter. Conversely, if the 90% quantile has its way when they disagree, then they are the ones calling the shots. If something intermediate happens, then they share power. This needs to be measured.
First of all, high earners don't even exist as a stable group to express preferences, so they can't form an oligarchy. High earners each year are a temporary subset of older professionals and small business owners, people who have a few good years.
I dispute that. That isn't the norm among high earners. Even if it were, the fact that the 90% quantile's preferences correlate more strongly with the richest and the median shows that either they do form a distinct, stable group, or their preferences change while they are in that quantile. In any case, they empirically have different interests than those in lower quantiles, which is all that matters here.
Conversely, it is wrong to identify the preferences of people around the median income with "the majority". The group of people near the median income is no larger than the group of people near the 90th percentile. People around the median income aren't even a useful social group with common interests. For example, assume that the US consisted only of two groups, high income earners and low income earners, each with their own set of preferences, and each equally listened to by politicians, with the median income earner being equally likely to come from either group. If you did the same analysis as in the paper, you'd find the same result: their model would show domination by an economic elite, even though there is none.
The reason why they use the preferences of the median is because some of the theories they are trying to test in this article predict that political decisions should follow the opinion of the median citizen. That requires assumptions, and those assumptions can be wrong. It would be interesting to see the same calculation done for other percentiles, or for everybody below the 90% quantile as a whole. But there are good reasons for believing the 50% quantile to be more representative of the people as a whole than the 90% quantile (and especially the 99% quantile). The income distribution curve is much flatter at that point - i.e. the 49% quantile and the 51% quantile are much more similar to the 50% quantile than the 89% and 91% quantiles are to the 90% quantile. Here is an illustrative (if a bit slow) illustration of the distribution. So many more people have similar lives to the 50% quantile than the 90% quantile, though the latter still isn't as spe
The study did two things.
The first a set of independent bivariate analyses, measuring the connection between the preferences of median americans vs. political decisions, the preferences of wealthy americans vs. political decisions and the preferences of lobby organizations vs. political decisions. Like others before them, they found that all of these had strong positive correlations, with wealthy americans having more influence than the preferences of median americans, but not by that much:
Median americans: 0.64 +- 0.08
Wealthy americans: 0.81 +- 0.08
Interest groups: 0.59 +- 0.09
That's consistent with the picture you were painting. But as you know, correlation does not mean causation. So the authors went a step further and did a multivariate analysis of all these at the same time. This is the new thing about this study. What they found was that the preferences of median americans have no explanatory power for political decisions - they correlate with them simply because the median american often agrees with rich americans, who have the real influence.
Median americans: 0.03 +- 0.08 (i.e. zero within the error bars)
Wealthy americans: 0.76 +- 0.09
Interest groups: 0.56 +- 0.09
If you only know the preferences of the rich, you can predict political decisions moderately well. If you add the preferences of the most important interest groups, the predictions improve substantially. But if you add the preferences of median americans, the predictions do not improve at all.
As a simplified illustration of such a situation, consider a totalitarian regime with a dictator and a street-sweeper, with the dictator making every decision. If the dictator and street-sweeper happen to agree 75% of the time, it might look like the street-sweeper actually had a lot of influence if you look at him in isolation. But if you analyse him together with the dictator, you discover that the opinion of the street-sweeper does not actually influence the results at all - the dictator's opinion is all that matters.
The article used these findings to evaluate several leading models for how american democracy works:
1. Majoritarian electoral democracy (i..e how the system is supposed to work)
2. Economic elite domination (plutocrachy/oligarchy)
3. Majoritarian pluralism (lobby groups decide, but taken together these are representative of the will of the people)
4. Biased pluralism (lobby groups decide, and represent only a small subset of the population (oligarchy))
They found that hypothesis 1 and 3 were incompatible with the data, while 2 and 4 were supported. So the summary wasn't all that bad.
As a final note: The resolution of their data was quite poor when it comes to wealth, so they could not determine how much of the apparent influence of the 90% quantile individuals come simply from correlation with 99% quantile individuals. It would be interesting to see a follow-up study with higher resolution. It might well turn out that the 90%-99% quantiles have as little influence as the 50% quantile. But at the moment there isn't enough data available to investigate that.
No, grandparent is right. The huge masses you describe are only needed if you assume that the background metric of the universe is unchanged. But if you are in a reference frame co-rotating with the world, then in that frame the whole universe is rotating at extreme speeds. This produces an extreme frame-dragging effect which makes it impossible not to rotate at or nearly at the same velocity as the other objects at that distance - no huge masses needed. This is an example of the more general Mach's principle.
For an example, look at the "rotating polar" example here. That's what you would have in the case where the earth is massless (the massive case is "rotating shwartzschild" on the next page (not to be confused with the Kerr metric), but the mass of the earth is so small compared to the effects of the rotating frame that it's not worth worrying about it). Here you can check that r=const, theta=pi/2 and dphi/dt = omega and dt/dtau = 1 (i.e. an object orbiting with angular velocity omega at distance r) fulfills the geodesic equation. This could be Proxima Centauri, for example. And that's for zero solar masses.
Thanks for the book reference (though the missing pages were very annoying). You have convinced me. I guess it's good my job isn't designing telescope optics.
To answer your question about the telescope: It's f-number is 2.5 at Gregorian focus, and we observe at 148 GHz. The optical transfer function plots I showed had multipole number on the horizontal axis.
The thing that's the self-convolution of the pupil function is the point spread function (g(theta) in my example from a few posts back). For the case of an ideal, top-hat shaped pupil function, the point spread function will fall to zero, and stay zero, at theta = 2*theta_pupil. But the optical transfer function (G(k) in my post) is the fourier transform of the point spread function. And the fourier transform of a function with limited support has unlimited support in fourier space. Hence, while the optical transfer function will never fall to zero in this case (or any other case with a sharp edge to the pupil function), except for occasional zero crossings.
I think I found the figure you you're referring to. It's this one, right? I don't think that figure lets you distinguish small from zero due to its very poor dynamic range. A logarithic second axis would be much more informative.
Here is an example of an MTF from an experiment I've worked on. It looks quite similar to the figure on the Wikipedia page, and by eye one might think that's it's reached zero by 18000 or so. But consider the logarithmic version of the same graph. As you can see, the graph had only fallen by about 20 dB by that point, and even at the end of the figure it's only down by 45 dB or so. So I don't think the Wikipedia figure supports your point.
Then that particular frequency cannot be recovered. But this usually only happens at zero crossings, which make up a vanishingly small fraction of the frequencies involved. Of course when noise is also present, then it's enough for G to be very small rather than exactly zero, and that would kill all the higher frequencies.
I didn't make this explicit, but nowhere does one need to assume that one is looking at a point source (though that helps). I use this for looking at the cosmic microwave background, which is as wide-field as it gets.
The effect of a telescope's point spread function is to convolve the image. A raw image f(x) is turned into f'(x) = int dy f(x-dy) g(y), where g(dx) is the point spread function. By the convolution theorem, a convolution is simply a product in fourier space, so F'(k) = F(k)*G(k), where uppercase functions are fourier-transpformed versions of lowercase ones, and k is the wavenumber. From this you can see that recovering the raw, unblurred image (i.e. overcoming the diffraction limit), is just a question of computing F(k) = F'(k)/G(k), i.e. dividing by by the point spread function in fourier space, or deconvoluting it in normal coordinates.
What limits our ability to do so is the presence of noise in the final image. So a more realistic data model is F'(k) = F(k)*G(k) + N(k), where N is the noise, and when we now try to remove the point spread function, we get F_estimate(k) = F'(k)/G(k) + N(k)/G(k) = F(k) + N(k)/G(k). So we get back our unblurred original image, but with an extra noise term. And since G(k) usually falls exponentially with k (meaning that high wavenumbers = small scales in the image are progressively blurred), N(k)/G(k) will grow exponentially with k, making the noise ever greater as one tries to recover smaller and smaller scales. This puts a limit on how small scales one can recover. But as you can see, it is not given just by G(k). I.e. not just by the diffraction limit. It is given by a combination of the telescope's noise level and the point spread function. In the absence of noise we can fully recover all the small scales. And even with noise one can push down a bit below the diffraction limit with enough data. But due to that exponential rise of the noise with wavenumber, it's an extreme case of diminishing returns. It is much cheaper to make the telescope bigger.
Sadly, I don't think that's the case. This is all voluntary agreements between Google and various coorporations that kick in *before* any DMCA stuff. I think what happens is that Google runs video/audio matching programs on behalf of other companies, and when something matches they take it down, notifying the user. The user can then assert that they do in fact have the right to upload the video. Once they do, the video is put back up, and the company is notified. They then file a real DMCA claim. The video is then taken down again, and the user is notified. They can then file a DMCA counterclaim, which would bring the video back but expose them to a lawsuit, or back down, in which case they get a "copyright strike", which leads to the loss of the ability to upload long videos, and eventually being banned from youtube.
I think these voluntary agreements are a perversion of an already pretty nasty law. I've had one of my own videos affected in a somewhat milder fashion: They put advertisements on the video instead of taking it down. That makes it seems like Iæm selling out my viewers to advertisers, but though the video was quite clearly fair use (a video comparing the current and previous world rectord speedruns of a computer game), I would have to consult a lawyer before contesting it, which would take days, and be expensive. The power asymmetry means that Sony etc. can accuse you as much as they want with no worry, while defending oneself is a costly and risky endeavor to normal users.
If this thing is falling from the parachute, that would make it a small, nearby, slow object rather than a larger, faster object further away. Assuming a 7 m tall parachute (including lines, ) + 1 m more of skydiver legs below that, we find that the rock traveled from around the top of the parachute to definitely below the skydiver in 15 frames of video, which gives us a velocity of about 13 m/s relative to the skydiver. Assuming that the parachute has slowed the skydiver down to a landing velocity of 7 m/s, this brings it to a total velocity of 20 m/s.
The terminal velocity of a rock depends on its size. I couldn't find numbers for rocks, but for hailstones (which should fall more slowly due to their lower density (2.5 times lower density gives sqrt(2.5) = 1.6 times lower terminal speed)), I find that one 8 cm in diameter falls at 48 m/s (so we would expect 75 m/s for rock). In general we have rho r^3 = Ar^2v^2 giving v proportional to sqrt(r). So to get a terminal velocity of 20 m/s we need a 0.6 cm diameter pebble.
This doesn't mean that the object necessarily is a nearby small object, but it is plasuible so far.
Another thing that points in the direction of a small object is the way it suddenly appears in the front-facing camera's field of view, quite far from the edge (indicating a very large angular velocity), but then only moves a few pixels the next frame. This camera's field of view seems to be about 60 degrees or so, which would give the object an angular velocity of more than 20 degrees per frame, or 500 degrees per second for the first frame when it appears in the forwards view. For 13 m/s relative to the skydiver, this would put the object at a distance of less than 1.5 m, which is consistent. The object's size is about 3 pixels in diameter at this point (out of a image width of 1280 pixels), which would give it a size of 0.35 cm, which is a bit too small.
According to the NRK article, the object was moving at 300 km/h, or 80 m/s. That would put it at a distance of 9.5 m (still really close!), and a size of 2.2 cm in diameter. Such an object should have a terminal velocity of 40 m/s, which is about half the 80 m/s we assumed. But the uncertainty of these measurements is big enough that that's probably well inside the margin of error. (To get 80 m/s we want a 10 cm diameter object, which might work if we take into acocount that the object is further away on the first frame when it appears in the forward camera than it would have been on average between that and the previous frame).
So based on these back-of-the-envelope non-expert calculations, I'd tentatively say that your pebble-in-the-parachute explanation fits the video just as well as the meteorite explanation. But I'd trust the experts who looked at this video more than me.
Just to clarify: The problem wasn't that you didn't have the right exponent. Exponential is not a collective name for x^2, x^3, x^4, etc. These are called power laws, not exponentials. Exponentials are when your variable is in the exponent, e.g. 2^x, 3^x, 4^x, etc. These start off slowly, but eventually grow extremely quickly. What you probably wanted to say was "increases as some power higher than one of the velocity". Though that doesn't roll off the tounge.
Why? What happened? Isn't carbon fiber the magical Space Elevator material of the glorious 3D printed future?
Carbon fiber or carbon fiber reinfoced polymer are not the same thing as carbon nanotubes. For example, carbon nanotubes have 15 times higher tensile strength than carbon fibre. I couldn't find toughness numbers for carbon nanotubes, though. And we can't really make anything large with nanotubes yet (not without losing most of their strength).
Surely you mean that drag increases quadratically with spead, not exponentially. (And at lower speeds (before the air flow becomes turbulent), it's linear.) If it really were exponential, planes and rockets wouldn't be going anywhere fast.
To illustrate: With quadratic drag, drag is four times greater at 100 km/h than at 50 km/h, and 16 times greater at 200 km/h, and so on, to 400 times greater at 1000 km/h. But if it were exponential, then the progression that matches the first two points would be one which quadruples for every 50 km/h increase in speed. So one would have: 100 km/h: 4 times, 150 km/h: 16 times, 200 km/h: 32 times, 1000 km/h: 1,099,511,627,776 times greater!
Thanks for all the details. I'd just like to point out a few points that won't change the overall point of what you're saying (though surely the overall energy cost per distance of normal and electric cars has been studied in great detail before - it would be nice to see the non-back-of-the-envelope version of this calculation):
Firstly, note that coal power plants (or any other plant that burns fuel) can be made much more efficient than 45% in cold areas where homes need heating, in the form of the combined heat and power plant, which have efficiencies of about 80%. Also, it's not a given that power plants would use coal. "Only" 27% of the worl'd energy use is in the form of coal. The remaining 73% come from cleaner sources of energy, with about 20% being much cleaner. And these numbers change over time too.
The purpose of patents is noble and laudible, but sometimes things don't end up working as intended. For a long while, doctors thought bleeding patiens was the best cure to all ills, but we now know that bleeding doesn't work, and only harms the patient. So despite the purpose of bleeding being ever so noble, what actually came of it was harmful.
Evaluating the overall effect of patents is a lot of work. The book I cited attempts to do so, but a bigger, more scientific investigation is needed. But I think the burden of proof should be on those advocating for patents. Monopolies should be avoided as much as possible (especially monopolies on abstract things like how to solve a problem), so one really should prove empirically that patents both achive their purpose, and that the harm does not outweigh the benefits, before creating any more than necessary.
At least patents have a time limit, though (unlike copyright (when's the next extension scheduled for again?)), so in that aspect they are better than copyright. But a striking example of the problems of patents (software patents in particular) is that what is widely considered to be the best h.264 encoder, the free software program x264, is illegal in many countries due to not paying license fees to the MPEG licensing authority (and this wouldn't work with the free software development model anyway). The same applies to other ffmpeg projects, and stuff in mplayer and many other very popular programs. These do not break copyright, and were not plagiarized from anybody - they just implement public standards.