|
Although this thread does not function under the same strict guidelines as the USPMT, it is still a general practice on TL to provide a source with an explanation on why it is relevant and what purpose it adds to the discussion. Failure to do so will result in a mod action. |
Guys, there is simply no 100% certainty unless your data sample is so big, that the unmeasured data cannot turn the result around.
I have no clue how the pollsters' tests work, but it is mathematically not possible with reasonable assumptions. The whole point of these tests is that you cannot ask everyone, so you try to find a reasonable representation, you try to find reasonable distribution function and then, under these assumptions you predict the values or draw the conclusions that are best at managing the uncertainty.
You may round up, you may take a decision based on the test result, but you will never have a 100% certainty.
|
United Kingdom13775 Posts
In general most predictions just weigh the information available in a way that yields the desired results. Many fools predicted 99.9% Clinton as well.
|
On April 26 2017 02:12 Big J wrote:
Guys, there is simply no 100% certainty unless your data sample is so big, that the unmeasured data cannot turn the result around.
I have no clue how the pollsters' tests work, but it is mathematically not possible with reasonable assumptions. The whole point of these tests is that you cannot ask everyone, so you try to find a reasonable representation, you try to find reasonable distribution function and then, under these assumptions you predict the values or draw the conclusions that are best at managing the uncertainty.
You may round up, you may take a decision based on the test result, but you will never have a 100% certainty.
depends on how you use stats. If you're simply saying who wins now given the current polling than 100 percent is a possibility, say if it's like a 40 point lead with a 3 point margin of error. obviously it's not literally 100 percent but it's close enough that 100 percent makes more sense than 99 percent
|
On April 25 2017 11:53 cLutZ wrote:Show nested quote +On April 25 2017 10:42 warding wrote:On April 25 2017 10:19 cLutZ wrote:
Can someone from France explain how Le Pen is considered so threatening to the status quo? Her economic illiteracy is the same as all the other candidates (except the one that finished 3rd), and my understanding is that if she tries to unilaterally do something on EU you can no-confidence vote and get a new President?
Unless immigration is really the #1 issue of the time in France, but idk how that's possible given that they don't get to vote. Why do you consider Macron economically illiterate? In the realm of ideas, Le Pen comes from the same place fascism came from. We tried it out a whole ago on this side of the pond and it left us a really bad taste in our mouths. Edit: to add a little bit of seriousness to the snark, she doesn't have to unilaterally try to leave the EU - nor could she - but I'm guessing she could call s referendum which would run the risk of going the same way the UK did. To everyone in Europe, that is real economic and social upheaval if it happens. Its my understanding he is marketed as an "economic centrist/realist" but doesn't have anything about the early retirement age, 35 hr work week, or pension cuts in his platform. Also I haven't seen a plan for reducing the youth unemployment rate. He's fine elsewhere, and certainly better than Le Pen, but I dont see why she's cataclysmic. The immigration thing, IMO seems like a predictable response to a country with a large welfare state. Open borders or a welfare state, pick one, is the old Milton Friedman saying.
Given that Le Pen had a large immovable chunk of the right, Fillon another one of party loyalists, there didn't seem to be any more space in the electorate for people with the stomach to hear someone say they're going to cut pensions, increase working hours and delay retirements.
I don't think it's possible for someone who's worked in investment banking and as minister for the economy to not have at least a decent understanding about the sustainability of the pension system and insights over labor law rigidity. What I think is at play is an understanding of what fights can be fought in a highly corporativist country.
I'm not french and my understanding of French politics is not a lot more than superficial so I may be wrong. A lot of it seems to be very similar to Portugal though.
About Le Pen, this isn't about immigration policy and the left-right spectrum. On one hand, it's civilizational and a matter of national identity. Like the French here have mentioned, being French means something beyond ethnicity and what she represents is a different vision of what France is to that of the majority of French people. On the other hand, leaving the EU isn't simply like signing an executive order poo-pooing NAFTA. It actually means launching an environment of uncertainty and change that would disrupt industries, close down companies and put even more people out of work. All the while you're being led by an economically illiterate racist.
Most elections in most western countries mean going from center right to center left and I can't think of any case where a new government effectively caused an economic downturn right off the bat. Electing someone like Melenchon or Marine Le Pen, assuming they could be effective on their promises, would really test the waters on that, so the stakes really are higher than they usually are. It's fashionable to hate the boring ping-pong of the center parties but that is the type of political systems that have allowed for the standard of living those in developed countries enjoy.
What's really baffling to me is the aspirational side of what each candidate represents and the preferences of (some of) the French. If you look at which countries are ahead of France in both the Human Development Index and GDP per capita this is the list you get: Luxembourg, Ireland, Norway, Switzerland, United States, Netherlands, Sweden, Australia, Germany, Iceland, Denmark, Canada and the United Kingdom. It seems to me that jobs and the economy is the single most important issue in the election. What those countries have in common are much more liberal economic and labour market policies - as per the doing business ranking or the economic freedom ranking. Macron's ideas for France aspire to that model. Meanwhile, Melanchon's aspiration is Hugo Chavez's Venezuela and Fidel's Cuba. Le Pen's aspirations belong in the first half of the XX century. I don't understand how someone can enjoy the high standard of living the French do and think "you know what's missing? Toilet paper shortages and rationed soap".
I hate sounding like the Cato Institute but I had to vent.
|
On April 26 2017 02:18 Karis Vas Ryaar wrote:Show nested quote +On April 26 2017 02:12 Big J wrote:
Guys, there is simply no 100% certainty unless your data sample is so big, that the unmeasured data cannot turn the result around.
I have no clue how the pollsters' tests work, but it is mathematically not possible with reasonable assumptions. The whole point of these tests is that you cannot ask everyone, so you try to find a reasonable representation, you try to find reasonable distribution function and then, under these assumptions you predict the values or draw the conclusions that are best at managing the uncertainty.
You may round up, you may take a decision based on the test result, but you will never have a 100% certainty. depends on how you use stats. If you're simply saying who wins now given the current polling than 100 percent is a possibility, say if it's like a 40 point lead with a 3 point margin of error. obviously it's not literally 100 percent but it's close enough that 100 percent makes more sense than 99 percent
100% means there is only one possible outcome, which is why you should never round up to 100%, but rather use something like >99%.
|
United Kingdom13775 Posts
With sufficiently low variance in your model no deviation from the desired result is possible.
|
On April 26 2017 02:27 warding wrote:Show nested quote +On April 25 2017 11:53 cLutZ wrote:On April 25 2017 10:42 warding wrote:On April 25 2017 10:19 cLutZ wrote:
Can someone from France explain how Le Pen is considered so threatening to the status quo? Her economic illiteracy is the same as all the other candidates (except the one that finished 3rd), and my understanding is that if she tries to unilaterally do something on EU you can no-confidence vote and get a new President?
Unless immigration is really the #1 issue of the time in France, but idk how that's possible given that they don't get to vote. Why do you consider Macron economically illiterate? In the realm of ideas, Le Pen comes from the same place fascism came from. We tried it out a whole ago on this side of the pond and it left us a really bad taste in our mouths. Edit: to add a little bit of seriousness to the snark, she doesn't have to unilaterally try to leave the EU - nor could she - but I'm guessing she could call s referendum which would run the risk of going the same way the UK did. To everyone in Europe, that is real economic and social upheaval if it happens. Its my understanding he is marketed as an "economic centrist/realist" but doesn't have anything about the early retirement age, 35 hr work week, or pension cuts in his platform. Also I haven't seen a plan for reducing the youth unemployment rate. He's fine elsewhere, and certainly better than Le Pen, but I dont see why she's cataclysmic. The immigration thing, IMO seems like a predictable response to a country with a large welfare state. Open borders or a welfare state, pick one, is the old Milton Friedman saying. (...) What's really baffling to me is the aspirational side of what each candidate represents and the preferences of (some of) the French. If you look at which countries are ahead of France in both the Human Development Index and GDP per capita this is the list you get: Luxembourg, Ireland, Norway, Switzerland, United States, Netherlands, Sweden, Australia, Germany, Iceland, Denmark, Canada and the United Kingdom. It seems to me that jobs and the economy is the single most important issue in the election. What those countries have in common are much more liberal economic and labour market policies - as per the doing business ranking or the economic freedom ranking. Macron's ideas for France aspire to that model. Meanwhile, Melanchon's aspiration is Hugo Chavez's Venezuela and Fidel's Cuba. Le Pen's aspirations belong in the first half of the XX century. I don't understand how someone can enjoy the high standard of living the French do and think "you know what's missing? Toilet paper shortages and rationed soap".
Repeating stupid caricatures from the neoliberal press does not make them any more true, stick to your TINA worshipping instead of being smug and pretending like you understand anything to the socialist left
|
France12886 Posts
I am not sure what problem you have with the 100%: is it because it's not 100%-ε?
In which case it's just cleaner to just write 100%.
Or is it that you think that writing 100% means we predict the future for sure and this is what will happen?
In which case it's not predicting the future, it is giving a picture of the outcome at time T according to a model which can give different results every day since it's based on data that is renewed every day (but they are very likely not to change that much). With this model it's possible that its output is 100% / rounded to 100%.
For example, if I were to play a bo7 against ByuN, aligulac would predict that ByuN wins with around 100%, maybe they will say 99.999% if they are cheeky and don't like esthetic ways to present results, but it's the same.
|
|
|
|
|
On April 26 2017 03:12 Poopi wrote:
I am not sure what problem you have with the 100%: is it because it's not 100%-ε?
In which case it's just cleaner to just write 100%.
Or is it that you think that writing 100% means we predict the future for sure and this is what will happen?
In which case it's not predicting the future, it is giving a picture of the outcome at time T according to a model which can give different results every day since it's based on data that is renewed every day (but they are very likely not to change that much). With this model it's possible that its output is 100% / rounded to 100%.
For example, if I were to play a bo7 against ByuN, aligulac would predict that ByuN wins with around 100%, maybe they will say 99.999% if they are cheeky and don't like esthetic ways to present results, but it's the same.
Yes, the first one. As I said, and as Danglar said a few pages ago: You shouldn't mention 100% certainty with a prediction. It's hard enough alreay to have people understand, that these are exact mathematical procedures, based on data that might be not be as well-researched as a scientist would like.
It's not the math that's wrong, it's people's interpretations of it and working with rounded 100% certainty for these purposes is missleading them further.
|
That whole article is full of straw men. It uses Naomi Klein as some sort of authority (she's not even an economist). They see Hayek and his road to serfdom as bad and see Keynes as an alternative yet don't even know that Keynes praised the book (He had his critique on it of course).
It maintains that “the market” delivers benefits that could never be achieved by planning.
This is especially a terrible part. Apparently Keynesian is an alternative but they don't realise that Keynesian uses the market too? It has more government intervention yes but the economic system is still based on the market.
The article also neglects the fact that we have a whole economic school called New Keynesian which is still very influential and had great contributions to economics like sticky pricing etc. Some well known economists like Stiglitz and Krugman belong to this school.
I can give a more comprehensive critique of the article if you want but it's pretty terrible imo.
|
France12886 Posts
On April 26 2017 03:28 Big J wrote:Show nested quote +On April 26 2017 03:12 Poopi wrote:
I am not sure what problem you have with the 100%: is it because it's not 100%-ε?
In which case it's just cleaner to just write 100%.
Or is it that you think that writing 100% means we predict the future for sure and this is what will happen?
In which case it's not predicting the future, it is giving a picture of the outcome at time T according to a model which can give different results every day since it's based on data that is renewed every day (but they are very likely not to change that much). With this model it's possible that its output is 100% / rounded to 100%.
For example, if I were to play a bo7 against ByuN, aligulac would predict that ByuN wins with around 100%, maybe they will say 99.999% if they are cheeky and don't like esthetic ways to present results, but it's the same. Yes, the first one. As I said, and as Danglar said a few pages ago: You shouldn't mention 100% certainty with a prediction. It's hard enough alreay to have people understand, that these are exact mathematical procedures, based on data that might be not be as well-researched as a scientist would like. It's not the math that's wrong, it's people's interpretations of it and working with rounded 100% certainty for these purposes is missleading them further.
Yet you accept that the probability density of a single point in continuous probabilities is 0? Okay
|
On April 26 2017 03:44 Poopi wrote:Show nested quote +On April 26 2017 03:28 Big J wrote:On April 26 2017 03:12 Poopi wrote:
I am not sure what problem you have with the 100%: is it because it's not 100%-ε?
In which case it's just cleaner to just write 100%.
Or is it that you think that writing 100% means we predict the future for sure and this is what will happen?
In which case it's not predicting the future, it is giving a picture of the outcome at time T according to a model which can give different results every day since it's based on data that is renewed every day (but they are very likely not to change that much). With this model it's possible that its output is 100% / rounded to 100%.
For example, if I were to play a bo7 against ByuN, aligulac would predict that ByuN wins with around 100%, maybe they will say 99.999% if they are cheeky and don't like esthetic ways to present results, but it's the same. Yes, the first one. As I said, and as Danglar said a few pages ago: You shouldn't mention 100% certainty with a prediction. It's hard enough alreay to have people understand, that these are exact mathematical procedures, based on data that might be not be as well-researched as a scientist would like. It's not the math that's wrong, it's people's interpretations of it and working with rounded 100% certainty for these purposes is missleading them further. Yet you accept that the probability density of a single point in continuous probabilities is 0? Okay
Yeah, of course I accept that. It is not rounded to zero, it is actually zero.
|
France12886 Posts
On April 26 2017 03:53 Big J wrote:Show nested quote +On April 26 2017 03:44 Poopi wrote:On April 26 2017 03:28 Big J wrote:On April 26 2017 03:12 Poopi wrote:
I am not sure what problem you have with the 100%: is it because it's not 100%-ε?
In which case it's just cleaner to just write 100%.
Or is it that you think that writing 100% means we predict the future for sure and this is what will happen?
In which case it's not predicting the future, it is giving a picture of the outcome at time T according to a model which can give different results every day since it's based on data that is renewed every day (but they are very likely not to change that much). With this model it's possible that its output is 100% / rounded to 100%.
For example, if I were to play a bo7 against ByuN, aligulac would predict that ByuN wins with around 100%, maybe they will say 99.999% if they are cheeky and don't like esthetic ways to present results, but it's the same. Yes, the first one. As I said, and as Danglar said a few pages ago: You shouldn't mention 100% certainty with a prediction. It's hard enough alreay to have people understand, that these are exact mathematical procedures, based on data that might be not be as well-researched as a scientist would like. It's not the math that's wrong, it's people's interpretations of it and working with rounded 100% certainty for these purposes is missleading them further. Yet you accept that the probability density of a single point in continuous probabilities is 0? Okay Yeah, of course I accept that. It is not rounded to zero, it is actually zero.
Yes but do we call such an event impossible or almost impossible? 
|
On April 26 2017 04:24 Poopi wrote:Show nested quote +On April 26 2017 03:53 Big J wrote:On April 26 2017 03:44 Poopi wrote:On April 26 2017 03:28 Big J wrote:On April 26 2017 03:12 Poopi wrote:
I am not sure what problem you have with the 100%: is it because it's not 100%-ε?
In which case it's just cleaner to just write 100%.
Or is it that you think that writing 100% means we predict the future for sure and this is what will happen?
In which case it's not predicting the future, it is giving a picture of the outcome at time T according to a model which can give different results every day since it's based on data that is renewed every day (but they are very likely not to change that much). With this model it's possible that its output is 100% / rounded to 100%.
For example, if I were to play a bo7 against ByuN, aligulac would predict that ByuN wins with around 100%, maybe they will say 99.999% if they are cheeky and don't like esthetic ways to present results, but it's the same. Yes, the first one. As I said, and as Danglar said a few pages ago: You shouldn't mention 100% certainty with a prediction. It's hard enough alreay to have people understand, that these are exact mathematical procedures, based on data that might be not be as well-researched as a scientist would like. It's not the math that's wrong, it's people's interpretations of it and working with rounded 100% certainty for these purposes is missleading them further. Yet you accept that the probability density of a single point in continuous probabilities is 0? Okay Yeah, of course I accept that. It is not rounded to zero, it is actually zero. Yes but do we call such an event impossible or almost impossible?
Obviously impossible. 
|
France12886 Posts
On April 26 2017 04:26 Big J wrote:Show nested quote +On April 26 2017 04:24 Poopi wrote:On April 26 2017 03:53 Big J wrote:On April 26 2017 03:44 Poopi wrote:On April 26 2017 03:28 Big J wrote:On April 26 2017 03:12 Poopi wrote:
I am not sure what problem you have with the 100%: is it because it's not 100%-ε?
In which case it's just cleaner to just write 100%.
Or is it that you think that writing 100% means we predict the future for sure and this is what will happen?
In which case it's not predicting the future, it is giving a picture of the outcome at time T according to a model which can give different results every day since it's based on data that is renewed every day (but they are very likely not to change that much). With this model it's possible that its output is 100% / rounded to 100%.
For example, if I were to play a bo7 against ByuN, aligulac would predict that ByuN wins with around 100%, maybe they will say 99.999% if they are cheeky and don't like esthetic ways to present results, but it's the same. Yes, the first one. As I said, and as Danglar said a few pages ago: You shouldn't mention 100% certainty with a prediction. It's hard enough alreay to have people understand, that these are exact mathematical procedures, based on data that might be not be as well-researched as a scientist would like. It's not the math that's wrong, it's people's interpretations of it and working with rounded 100% certainty for these purposes is missleading them further. Yet you accept that the probability density of a single point in continuous probabilities is 0? Okay Yeah, of course I accept that. It is not rounded to zero, it is actually zero. Yes but do we call such an event impossible or almost impossible? Obviously impossible.
Then I'd like the source, because for an event of probability 1 I'm pretty sure that we call that "almost certain" or "almost sure" so I guess it's the other way for 0.
|
On April 26 2017 04:27 Poopi wrote:Show nested quote +On April 26 2017 04:26 Big J wrote:On April 26 2017 04:24 Poopi wrote:On April 26 2017 03:53 Big J wrote:On April 26 2017 03:44 Poopi wrote:On April 26 2017 03:28 Big J wrote:On April 26 2017 03:12 Poopi wrote:
I am not sure what problem you have with the 100%: is it because it's not 100%-ε?
In which case it's just cleaner to just write 100%.
Or is it that you think that writing 100% means we predict the future for sure and this is what will happen?
In which case it's not predicting the future, it is giving a picture of the outcome at time T according to a model which can give different results every day since it's based on data that is renewed every day (but they are very likely not to change that much). With this model it's possible that its output is 100% / rounded to 100%.
For example, if I were to play a bo7 against ByuN, aligulac would predict that ByuN wins with around 100%, maybe they will say 99.999% if they are cheeky and don't like esthetic ways to present results, but it's the same. Yes, the first one. As I said, and as Danglar said a few pages ago: You shouldn't mention 100% certainty with a prediction. It's hard enough alreay to have people understand, that these are exact mathematical procedures, based on data that might be not be as well-researched as a scientist would like. It's not the math that's wrong, it's people's interpretations of it and working with rounded 100% certainty for these purposes is missleading them further. Yet you accept that the probability density of a single point in continuous probabilities is 0? Okay Yeah, of course I accept that. It is not rounded to zero, it is actually zero. Yes but do we call such an event impossible or almost impossible? Obviously impossible. Then I'd like the source, because for an event of probability 1 I'm pretty sure that we call that "almost certain" or "almost sure" so I guess it's the other way for 0.
Oh sorry, you are right. It is possible. The probability is still actually zero, unlike what we were taking about before with your rounded up 100%, that is not actually 100%.
|
On April 26 2017 04:27 Poopi wrote:Show nested quote +On April 26 2017 04:26 Big J wrote:On April 26 2017 04:24 Poopi wrote:On April 26 2017 03:53 Big J wrote:On April 26 2017 03:44 Poopi wrote:On April 26 2017 03:28 Big J wrote:On April 26 2017 03:12 Poopi wrote:
I am not sure what problem you have with the 100%: is it because it's not 100%-ε?
In which case it's just cleaner to just write 100%.
Or is it that you think that writing 100% means we predict the future for sure and this is what will happen?
In which case it's not predicting the future, it is giving a picture of the outcome at time T according to a model which can give different results every day since it's based on data that is renewed every day (but they are very likely not to change that much). With this model it's possible that its output is 100% / rounded to 100%.
For example, if I were to play a bo7 against ByuN, aligulac would predict that ByuN wins with around 100%, maybe they will say 99.999% if they are cheeky and don't like esthetic ways to present results, but it's the same. Yes, the first one. As I said, and as Danglar said a few pages ago: You shouldn't mention 100% certainty with a prediction. It's hard enough alreay to have people understand, that these are exact mathematical procedures, based on data that might be not be as well-researched as a scientist would like. It's not the math that's wrong, it's people's interpretations of it and working with rounded 100% certainty for these purposes is missleading them further. Yet you accept that the probability density of a single point in continuous probabilities is 0? Okay Yeah, of course I accept that. It is not rounded to zero, it is actually zero. Yes but do we call such an event impossible or almost impossible? Obviously impossible. Then I'd like the source, because for an event of probability 1 I'm pretty sure that we call that "almost certain" or "almost sure" so I guess it's the other way for 0.
"Almost never" is used to describe zero-probability events, you are correct. I had some fun researching that when Cenk and Sam Harris disagreed on the probability of Jesus coming back specifically in Missouri vs Jesus coming back anywhere in the world.
edit: in summary, it is incorrect to assume that an event is impossible just because it has a zero probability of happening, cause zero probability events aren't necessarily impossible. It's not really an intuitive thing.
|
France12886 Posts
On April 26 2017 04:34 Big J wrote:Show nested quote +On April 26 2017 04:27 Poopi wrote:On April 26 2017 04:26 Big J wrote:On April 26 2017 04:24 Poopi wrote:On April 26 2017 03:53 Big J wrote:On April 26 2017 03:44 Poopi wrote:On April 26 2017 03:28 Big J wrote:On April 26 2017 03:12 Poopi wrote:
I am not sure what problem you have with the 100%: is it because it's not 100%-ε?
In which case it's just cleaner to just write 100%.
Or is it that you think that writing 100% means we predict the future for sure and this is what will happen?
In which case it's not predicting the future, it is giving a picture of the outcome at time T according to a model which can give different results every day since it's based on data that is renewed every day (but they are very likely not to change that much). With this model it's possible that its output is 100% / rounded to 100%.
For example, if I were to play a bo7 against ByuN, aligulac would predict that ByuN wins with around 100%, maybe they will say 99.999% if they are cheeky and don't like esthetic ways to present results, but it's the same. Yes, the first one. As I said, and as Danglar said a few pages ago: You shouldn't mention 100% certainty with a prediction. It's hard enough alreay to have people understand, that these are exact mathematical procedures, based on data that might be not be as well-researched as a scientist would like. It's not the math that's wrong, it's people's interpretations of it and working with rounded 100% certainty for these purposes is missleading them further. Yet you accept that the probability density of a single point in continuous probabilities is 0? Okay Yeah, of course I accept that. It is not rounded to zero, it is actually zero. Yes but do we call such an event impossible or almost impossible? Obviously impossible. Then I'd like the source, because for an event of probability 1 I'm pretty sure that we call that "almost certain" or "almost sure" so I guess it's the other way for 0. Oh sorry, you are right. It is possible. The probability is still actually zero, unlike what we were taking about before with your rounded up 100%, that is not actually 100%.
Yes I know that it is not exactly 100% but if we can agree on calling a 0 probability event "almost never" instead of "never" we could agree on calling 99.99x% 100%  . Anyways seeing the other probabilities they are giving, they only show one decimal and seem to round to the superior, so their prediction is most likely between 99.9 % and 100%.
About the fact that it's hard to make people understand the intricacies of such processes, well if everyone understood it easily, you wouldn't work in the same field would you? :o
|
|
|
|
|
|
EPT
