fela

I think the best term would have been statistically surprising, because it strongly hint at the fact that the result would be surprising under the null hypothesis, witch really is all that "statistically significant" really means. Sometimes surprising results happen, but all other things being equal they might hint at the null hypothesis being false. I could also live with "statistically interesting". "Detectable", suggested in another comment, seems to have some of the same issues as significant, it is too strong and seems to imply that now we know something is really there.

I don't think this is true unless you have a ridiculously high electricity bill. When I checked, one intercontinental retour flight was roughly equivalent to my yearly electricity bill, in terms of CO2 emissions. I have to admit I'm not sure how to reconcile this with CO2 credit prices, but I'm quite sure it's not really possible to offset the CO2 emissions of a intercontinental flight with 20 euro, that would mean that a minor tax on flights would make them practically carbon neutral, this is definitely not the case.

I assume you meant: If an airplane is as safe as average then it has PUT_NUMBER chance of having 2 incidents after 150k flights. 0.01% is actually the number I'm getting, assuming parent estimates are correct and making naive assumptions. In other words only 1 every 10 000 airplane models will have 2 incidents that early on if they are of average safety.

That is different then stating the probability of it being as safe as the average airplane, which you can't do as easily without additional modelling/priors and bayesian statistics.

I like to split the theorem in the following way:

P(Hypothesis|Data) = P(Hypothesis) * evidence_factor

P(Hypothesis) is the prior probability of the Hypothesis being true, in other words the probability we gave to the Hypothesis before seeing any of the data we are using in the theorem. When new data is observed, we use Bayes' theorem to update our believe in the hypothesis, which in practice means multiplying our prior probability by a number that depends on how well the new data fits our hypothesis. More precisely:

evidence_factor = P(Data|Hypothesis)/P(Data)

So it is the ratio of how likely our data is if our hypothesis is true, compared to (divided by) how likely it is in general. If it is more likely to occur in our Hypothesis, our probability of it being true increases, if it is more likely in general (and thus also more likely in case our hypothesis is not true, you can prove mathematically that those two statements are the same), then our believe in the hypothesis decreases.

TLDR: Prob(Hypothesis after I have seen new data) = Prob(Hypothesis before I saw the new data) * (how likely I am to see the data if my hypothesis is true, compared to in general)

Don't you need time to even define movement? Movement is a change of the position with respect to time. So without time you can't have movement. But obviously you are right that the two concepts are strictly related, that doesn't mean time doesn't "exist" (however you define "exist").

But that is very interesting, that means that when saying the numbers they were basically using something similar to Arabic numbers!

The current stock price (and thus market cap) already assumes future growth. The market cap would increase further only if growth exceeds the current expectations of investors. (Or due to other factors unrelated to growth).

I stopped using bookmarks after I realized I wasn't using them, thanks to a combination of:

1. Autocompletion: for any website I use regularly I just write a substring of the url or Title (Firefox does this especially well). This covers probably 70% of my browsing.

2. Google. This might take slightly longer in case I want to find a specific article I had read some time ago, but it still seems less effort that having to bother with bookmarks, in my experience: either you have a very long list of unsorted bookmarks, in witch it's hard to search, or you have to spend time sorting them into sub-folders.

Now that I think of it, the following would be a very useful Google feature: +1 an url so that it becomes much more likely to bubble to the top in future searches.

Looks a bit like an over engineered coffee machine to me. On the other, if customers enjoy the experience, why not.

The definition of p-value is the same independent of method, as far as I can tell the only real difference is that by Neyman–Pearson you just look at whether the p-value is below a threshold, and Fisher looks at p-value as "strength of evidence" valuable in itself. It's still not the probability that your result was due to chance, it's the probability that under the null hypothesis (and you will definitely need one) you would get that value (or more extreme) by chance.

"it’s telling you that there’s at most an alpha chance that the difference arose from random chance. In 95 out of 100 parallel universes, your paper found a difference that actually exists. I’d take that bet."

This is wrong. It’s telling you that there’s at most an alpha chance that a difference like that (or more) would have arisen from random chance if the quantities are actually equal. And if the quantities are equal 95 out of 100 parallel universes would not be able to reject the null hypothesis.

Is he saying that he would take the xkcd bet[0] on the frequentist side?

[0] https://xkcd.com/1132/

"what are the odds that the results you observed could have arisen by chance?"

If you say it like this it will very easily be misinterpreted. Once your results are in there are two cases: (1) either the null hypothesis is true and you got those results due to chance, or (2) the null hypothesis is false and there was some actual effect outside of the null hypothesis that helped you get the results.

Due to this it is very easy to interpret you statement as referring to the probability of (1).

Two two following definitions of p-values sound similar but are not:

[Correct] The probability of getting the results by chance if the null hypothesis is true P(Results|H0)

[Wrong] The probability that you got the results by chance and thus the null hypothesis was actually true P(H0|Results)

I'm not saying you didn't get it, but somebody reading what you wrote can very easily be fooled. And there are a lot of dead wrong definitions on the web[0][1][2][3].

[0] https://www.americannursetoday.com/the-p-value-what-it-reall...

[1] https://practice.sph.umich.edu/micphp/epicentral/p_value.php

[2] http://natajournals.org/doi/full/10.4085/1062-6050-51.1.04

[3] http://www.cdc.gov/des/consumers/research/understanding_scie...

Unfortunately no. Very much no, even though it's widely believed that that is a good definition/intuition (and used in many places).

It's the odds of having that results due to chance, if the null hypothesis is true[0]. That latter part might sound pedantic, but the whole point is that we don't know how likely the null hypothesis is. If I test wheather the sun has just died[1] and get a p-value of 0.01 it's still very likely that this result is due to change (surely more than 1%)! We need a prior probability (i.e. bayesian statistics) to calculate the probability that the result was due to chance, that is why that partial definition is incomplete and actually very misleading. This point is subtle, but very important to really understand p-values.

Another way to look at it is: if we knew the probability that the result was due to chance we could also just take 1-p and have to probability of there actually being some effect, a probability that hypothesis testing cannot give us.

There is one nice property that hypothesis testing does have (and why presumably it's so widely used): if the idea you are testing is wrong (which actually means "null hypothesis true") you will most likely (1-p) not find any positive results. This is good, this means that if the sun in fact did not die, and use 0.01 as your threshold, 99% of the experiments will conclude that there is no reason to believe the sun has died. So hypothesis testing does limit the number of false positive findings. The xkcd comic is a bit misleading it this regard, yes it does highlight the limitations of frequentist hypothesis testing, but the scenario depicted is a very unlikely one, in 99% of the cases there would have been a boring and reasonable "No, the sun hasn't died".

For an incredibly interesting article about the difficulty of concluding anything definitive from scientific results I highly recommend "The Control Group is out of Control" at slatestarcodex[2].

[0] To be even more pedantic you would have to add "equal or more extreme", and "under a given model", but "if the null hypothesis is true" is by far the most important piece often missing.

[1] https://xkcd.com/1132/

[2] http://slatestarcodex.com/2014/04/28/the-control-group-is-ou...

While that sounds easy what if swerving off course definitely saves one live, but might cause one death with 60% probability? What if it definitely will save 3 lives, and might cause one death with 10% probability? Do you see the problem with absolute rules? Human morality is quite complex and not so easy to model with simple rules.

On a single participant yes, but their real goal, I presume, is to aggregate the data and then they will be able to reach more concrete conclusions.

We do make moral choices, and there are rules and heuristics we use, they might be quite complicated, and they might not be what we think they are, but I think nonetheless that it should be possible to come close to predicting a human moral decision making quite well by using an accurate enough model.

And as autonomous vehicles will have to make decisions that have moral implications, they better do so in a way that humans will be happy with. I think this is an important area of research. This won't mean a machines will have morals of his own, whatever that means, but that they should do what (most?) humans would consider morally right. And what do humans consider morally right? Well that is exactly what we should try to find out.

Your results might seem spurious because of the small sample size, but when aggregating the results of all the participants they will have enough data to be able to conclude how many people did act like you did with apparent preferences due to chance, and how many actually where "biased" in some way.

I don't think this is about creating realistic scenarios, but about finding out what people take into consideration when making moral judgements. The experiment seems to be designed to gather as many such preferences as possible.

The hope must be that if people consistently prefer saving the life of young people in this made up scenario they will have similar preferences in a more realistic scenario. Of course weather such a generalization holds will have to be confirmed by further studies. But this seems like a good first step to explore moral decisions more.

The reporting on The Verge seems slightly better

http://www.theverge.com/2016/8/22/12592938/roborace-self-dri...

You can change a password, and you can calculate how hard it is to for an attacker to obtain a randomly generated password.

It is much harder to formalise how hard it is for an attacker to find out what algorithm you use, so it is risky relying too much on him not being able to do so.

-) Other users paying for the premium version

Or rephrasing it a little, in a way that explain why it makes sense to let you use the free version

-) The x% probability that a free user will create a paid user (either my becoming one or by referring other free users that become one)

The word pirate has a new and different meaning in the 21st century, I think it's too late to try to fight that. And the title presumably helps with conveying what the website is about, compare it with my submission where I used the original title (and which got one upvote in total): https://news.ycombinator.com/item?id=11093454

That is true, but you could argue that the same is true with elections. What if everybody that wants to vote in a certain way is sick on election day? Not very likely, and of course in case of elections the number is larger and therefore the probability of (this specific type) of distortion is much smaller. But then one is again talking about probabilities. And in case of elections there are many other arbitrary factors that can influence the results.

Edit to say that your point is interesting and one I hadn't really thought about. This make elections a referendum on the most salient topics, which I guess makes sense. I still think that in elections there are a lot of other factors that influence decisions and given certain voting preferences there are a lot of arbitrary factors that influence the results, more so that in sortition, whose main drawback, as I see it is how little we know about how it would work in practice.

Original message: But even for salient issues there are a lot of random factors in elections. Suppose there is a issue so important that everybody cares and votes based only on that issue. Suppose there are only two candidate, and they have a clear and opposing position on this issue so that things are very simple for voters.

Suppose candidate A gets 50,999,897 votes Suppose candidate B gets 50,456,002 votes

B can still win, as happened with Bush, depending on the voting system. This is just an example and of course depends on the specific voting system. The real point being that for sortition you have simple statistical guarantees, always, independently from salience.

Although I very much like the idea of sortition I found the paper by Pluchino et al. very flawed, the simulation they made captures none of the effects of sortition, and the effects they measure have no equivalent in the real world.

Just to get an idea, in their model politicians make many laws that help the population a little bit, instead the randomly selected citizen make make laws that help the population a lot, but they make only a few laws. And things have been defined in such a way that the optimal solution happens when mixing the two. They do a pretty good job at analyzing this simulation, the problem is that the simulation has little to do with the real world.

(I read the paper a few years ago so I hope I'm remembering things correctly).

I feel like sortition is much closer to direct democracy because the distortion is limited by the law of large numbers and can be exactly calculated. Of course this excludes things like people feeling like participating, which happens in direct democracy but not in sortition.

Actually, one way to see sortition is as a way of scaling direct democracy to large populations.

I've long been interested in the use of sortition in political decision making, and it always surprises me how little it has been seriously studied and considered compared to the potential it seems to have.

Much of the information there is is of pretty low quality. It might of course just be such a bad idea that everybody smart enough to give high quality contributions on the topic does not want to waste their time with the idea. But if this is the case it is totally non obvious to me, and most criticism I've read seem to be from people that do not have a clear understanding of the potential advantages sortition might have.

Very briefly, for the uninitiated, the main potential advantage of sortition is that it would make political decision making a lot more democratic. People representative of the population at large would actually discuss to make the decision, instead of the citizens making their choice by casting one vote every few years among a set of very similar parties (I know, this simplifies the debate a lot, but it is the main idea). This is very interesting if you are of the opinion (as I am) that lack of democracy is a big problem of our political systems. I believe that most time politicians go agains the will of people they do so for the wrong reasons and with the wrong goals, and way too often.

The law of large number makes sure the randomness in sortition is limited and predictable. Whereas with elections there is a big number of arbitrary factors that can greatly influence the results.

Of course sortition in practice might have a number of problem often brought up, but none seems unsolvable to the point where it's not even worth exploring the idea further.

How do you separate expertise from decision power, while still being able to make proper use of the expertise? How to implement sortition in practice? Would they ever let us? Would people be able to handle the pressure? Would they accept the position? And all criticism to democracy in general applies even more to sortition.

I think however that if you talked about elections to somebody who never heard about it, you could come up with just as a big number of potential problems. I don't know if sortition really is a better idea, but maybe it's an idea worth thinking about.

I recently read this article on sortition that appeared on the Atlantic which I think is really good: http://www.theatlantic.com/education/archive/2014/05/the-cas...

Another good starting point for further exploration is the blog Equality by Lot: https://equalitybylot.wordpress.com/

It seems to be a case of leaky abstraction, where the way python caches small objects becomes visible, another example:

  >>> a = 'hn'
  >>> a is 'hn'
  True
  >>> a = ''.join(['h', 'n'])
  >>> a is 'hn'
  False
  >>> a
  'hn'
  >>> a = 'h' + 'n'
  >>> a is 'hn'
  True

Edit: found another interesting case

  >>> a = 1
  >>> b = 1
  >>> a is b
  True
  >>> a = 500
  >>> b = 500
  >>> a is b
  False

the behavior of is seems pretty hard to predict

  >>> 3.1 is 3.1
  True
  >>> a = 3.1
  >>> a is 3.1
  False