SoylentNews is people
SoylentNews
from the walking-will-be-mandatory-soon dept.
It's official: France bans short haul domestic flights in favour of train travel:
France has been given the green light to ban short haul domestic flights.
The European Commission has approved the move which will abolish flights between cities that are linked by a train journey of less than 2.5 hours.
[...] France is also cracking down on the use of private jets for short journeys in a bid to make transport greener and fairer for the population.
Transport minister Clément Beaune said the country could no longer tolerate the super rich using private planes while the public are making cutbacks to deal with the energy crisis and climate change.
[...] The ban on short-haul flights will be valid for three years, after which it must be reassessed by the Commission.
"[This] is a major step forward in the policy of reducing greenhouse gas emissions," transport minister Beaune said in a press release.
[...] Sarah Fayolle, Greenpeace France transport campaign manager, told Euronews that there were both "negative and positive aspects" to the European Commission's decision given that only three routes are affected.
"It's going in the right direction, but the initial measure is one that's (not very) ambitious. We must go even further," she said.
(Score: 2) by JoeMerchant on Friday December 23 2022, @08:37PM (46 children)
Regarding normal distribution as a model for meteor strike frequency, if that is all your source has to offer, then their data set is seriously lacking in real-universe data for the distribution.
Meteors impacting Earth come in showers (clusters), not normal distributions.
I particularly liked the "closest approach watch" for DA 2014 in February 2013, when a 500 megaton bolide came out of our rather large blind spot and said "Surprise MFer!" to Chelyabinsk on the same day. Those were supposed to be unrelated, not part of a shower type event, but they still fit very poorly into any normal distribution model for space rocks.
🌻🌻 [google.com]
(Score: 1) by khallow on Friday December 23 2022, @11:28PM (45 children)
Normal distribution as a model for averaging processes. It's backed by math [wikipedia.org] which is a lot more solid than JoeMerchant hand waving.
(Score: 2) by JoeMerchant on Saturday December 24 2022, @12:05AM (44 children)
Normal distribution is the model you use when you don't know what the actual distribution looks like. It isn't a bad distribution, it works well for many natural processes, and is the best choice for a lot of things, particularly things that aren't very important and you just want a reasonable distribution that won't draw a lot of criticism when you choose it.
And, I'm so impressed that it's backed by "math"!!! I just can't tell you how impressed, really!!!!!
🌻🌻 [google.com]
(Score: 1) by khallow on Saturday December 24 2022, @12:09AM (43 children)
It's also the model you use when you know the actual distribution is normal. We can keep playing this game, but what's the point?
(Score: 2) by JoeMerchant on Saturday December 24 2022, @04:39PM (42 children)
Asteroids are not normally distributed. They are in belts, bands, clusters and groups.
🌻🌻 [google.com]
(Score: 1) by khallow on Saturday December 24 2022, @05:13PM (41 children)
If you had read my posts rather than go on this fantasy kick, you would have seen that I didn't claim otherwise. For example:
Recall that I was speaking of the distribution of an average. That is normally distributed by math.
(Score: 2) by JoeMerchant on Saturday December 24 2022, @05:38PM (15 children)
>I was speaking of the distribution of an average. That is normally distributed by math.
I think you should apply as a technical consultant for the Associated Press, you're playing on the next level above which most wire stories are written.
However, when meteors of a certain size strike the earth at an "average of every 100-200 years":
First, an average is a single number, not a range.
Second, I might agree that the "best data available" is currently modeled as a normal distribution, but that's telling about the quality of that data which is anything but normally distributed in reality.
Third, four major strikes every 600 years may average to a strike every 150 years but if the strikes come in clusters of 3 to 5 within a few days, that's not a normal distribution and use of things like standard deviations don't apply, at least not appropriately like they do on a dataset that is more or less normally distributed.
Fourth, maybe you should stay away from the AP after all since the vast majority of their readers are challenged when trying to comprehend a simple average, much more so a normal distribution.
Our process engineers are proud when someone technically oriented "gets" the ramifications of a presentation with a bimodal distribution, and these are 2%er stats nerds. Still, I would trust about 1
1 in 50 of those process engineers to do a summary risk analysis for something as complex and important as a nuclear power plant, and based on my limited exposure to investors controlling $10M or more, I would trust less than 1 in 500 of them to do the right thing with that risk analysis.
🌻🌻 [google.com]
(Score: 1) by khallow on Saturday December 24 2022, @11:38PM (14 children)
An average is a distribution not a single number.
(Score: 2) by JoeMerchant on Sunday December 25 2022, @03:10AM (13 children)
Bzzzt. Webster:
a single value (such as a mean, mode, or median) that summarizes or represents the general significance of a set of unequal values
https://www.merriam-webster.com/dictionary/average [merriam-webster.com]
A distribution is a better way of evaluating probability.
A normal distribution is the common default choice when the actual distribution is unknown.
🌻🌻 [google.com]
(Score: 1) by khallow on Sunday December 25 2022, @07:47AM (12 children)
And also when the distribution is known to be normal - as I repeatedly noted before. The average is such a distribution as I repeatedly noted before. And there's no point to your argument as I repeatedly noted before.
(Score: 2) by JoeMerchant on Sunday December 25 2022, @04:17PM (11 children)
Webster defines the word you used: average, perfectly.
A distribution is almost always a better characterization of a statistical process, but again, you used that incorrectly as well, giving two numbers and failing to define what distribution is being used and how, probably because your source did as well.
A normal distribution is the most commonly used, and abused, to the point that a common subspecies of layman treats it as the only distribution.
When using a range of the normal distribution, the most commonly abused range is 95% confidence, or very roughly +/-2SD, but assuming this is both prone to misinterpretation and extremely inappropriate for high risk applications like potentially spreading radioactive isotopes across large highly populated regions and or farmland.
If you are using a distribution, both the shape of the distribution and the confidence limits should be specified along with the bounds numbers.
But, if you are ok with not really knowing the risks of these things you are pontificating about, that's fine for you, you have a better idea of what you are talking about than the average bear, still no real grasp, but better than average.
🌻🌻 [google.com]
(Score: 1) by khallow on Sunday December 25 2022, @06:24PM (10 children)
And wrong. Please don't forget that this time.
Should any of that ever be in error, please be sure to tell me. In the meantime, I'm just not interested in your criticism.
And completely irrelevant to our discussion since that isn't a problem here.
Remember we were speaking of something that was more than two orders of magnitude off. It's not going to be a problem with a slight increase in the number of standard deviations we choose to take.
You don't have a clue what you're talking about.
(Score: 2) by JoeMerchant on Sunday December 25 2022, @06:59PM (9 children)
>a problem with a slight increase in the number of standard deviations we choose to take.
What's your idea of "slight" differences from 2 standard deviations?
🌻🌻 [google.com]
(Score: 1) by khallow on Sunday December 25 2022, @07:07PM (8 children)
20 standard deviations.
(Score: 2) by JoeMerchant on Sunday December 25 2022, @08:51PM (7 children)
As accurate as you need to be, I see.
🌻🌻 [google.com]
(Score: 1) by khallow on Sunday December 25 2022, @10:12PM (6 children)
(Score: 2) by JoeMerchant on Monday December 26 2022, @12:09AM (5 children)
20 standard deviations is outrageous, in any field.
🌻🌻 [google.com]
(Score: 1) by khallow on Monday December 26 2022, @04:20AM (4 children)
And yet, quite adequate for this task. We don't need to know the precision of the risk when the order of magnitude is more than sufficient.
(Score: 2) by JoeMerchant on Monday December 26 2022, @02:02PM (3 children)
Learn something today:
https://www.scribbr.com/statistics/standard-deviation/#:~:text=The%20standard%20deviation%20is%20the,clustered%20close%20to%20the%20mean. [scribbr.com]
🌻🌻 [google.com]
(Score: 1) by khallow on Monday December 26 2022, @04:20PM (2 children)
Similarly, I can evaluate orders of magnitude risks without requiring significant precision. That's why I'm not concerned about a few standard deviations when I find that a risk is more than two orders of magnitude short of being a real problem. Conversely, imagine if I told you that a risk was negligible because it was 9.7% instead of a back of envelope calculation of 10%? That precision make you feel better?
(Score: 2) by JoeMerchant on Tuesday December 27 2022, @12:39AM (1 child)
When speaking of 20SD, that's - generously - over 10 sigma from the mean:
http://www.aleph.se/andart/archives/2009/09/ten_sigma_numerics_and_finance.html [aleph.se]
1.529245*10^-23
Ain't nobody got that kind of data on much of anything in the real universe.
🌻🌻 [google.com]
(Score: 1) by khallow on Tuesday December 27 2022, @03:21AM
And as I have already observed, 10 sigma from the mean is not significant for the parameters I looked at because they were two orders of magnitude shy of being relevant.
(Score: 2) by JoeMerchant on Saturday December 24 2022, @06:15PM (24 children)
Note the absence of tidy sound bite 100-200 year averages:
https://www.sciencedirect.com/science/article/pii/S0019103518304676 [sciencedirect.com]
🌻🌻 [google.com]
(Score: 1) by khallow on Saturday December 24 2022, @11:43PM (23 children)
Indeed. And that's because the study is about estimating the damage from a Tunguska-class asteroid, not its frequency of occurrence. If you had read the abstract:
(Score: 2) by JoeMerchant on Sunday December 25 2022, @03:20AM (22 children)
Because what we have data on is evidence of damage from impacts, like they are attempting to link back to the size, composition and relative speed of the impactor objects.
The actual rate of impacts is still out in your hand waving land, as demonstrated recently in Chelyabinsk. How often does that happen? Well, there was that one in Sodom and Gomorrah, then we keep finding more and more evidence of pre-historic massive impact craters... Then this happened:
https://www.google.com/amp/s/www.nytimes.com/2021/10/14/world/canada/meteorite-bed.amp.html [google.com]
🌻🌻 [google.com]
(Score: 1) by khallow on Sunday December 25 2022, @07:57AM (21 children)
So what? If we were to instead look at data on frequency of impacts by energy released, we would find a lot more relevant data. For example, nuclear detection systems routinely spot impacts up to about two orders of magnitude less energy released than Tunguska. We also have a well-established power distribution from small meteorites on up to large asteroids.
A theory of impact frequency isn't disproved by a single incident - especially one that was expected to occur!
Again, we're not finding such "more and more" evidence out of line with expectations.
Really, what is the point of this argument, Joe? When my crude model predicts that a Tunguska-type event happens every 100-200 years, does that mean that such impacts never would happen through tens of thousands of years of history and prehistory? Is this really an argument you want to waste my time with?
(Score: 2) by JoeMerchant on Sunday December 25 2022, @04:25PM (20 children)
>nuclear detection systems routinely spot impacts up to about two orders of magnitude less energy released than Tunguska. We also have a well-established power distribution from small meteorites on up to large asteroids.
For the last what? 50 years? It's good data, but against a 4 billion years local system that is also occasionally visited by interstellar and intergalactic phenomenon, about 27% of which we estimate we have limited if any ability to detect, much less characterize.... The unknown unknowns still outweigh what we do know for impactors larger than Tunguska.
Really, what is the point of any argument, Khallow?
🌻🌻 [google.com]
(Score: 1) by khallow on Sunday December 25 2022, @06:34PM (10 children)
Sounds really good to me actually. This is a typical god of the gaps argument - that the behavior you claim could be there is squeezed into smaller and smaller gaps. This is actually a very extensive set of data you describe above. Most studies would consider 27% coverage of a population to be both remarkable and excessive, for example. And nobody, particularly you, has explained what's supposed to have changed in the last 50 years that such extensive data collection can't be applied to periods before present.
(Score: 2) by JoeMerchant on Sunday December 25 2022, @07:04PM (9 children)
>Sounds really good to me actually.
Good for you.
Crappy risk estimation based on limited data and experience was behind the failure at Fukushima.
🌻🌻 [google.com]
(Score: 1) by khallow on Sunday December 25 2022, @07:09PM (8 children)
"Was". We have better risk estimation now. Honestly, I think they should just build new reactors on the site now. The accident is a sunk cost. To just let everything go to seed is a waste.
(Score: 2) by JoeMerchant on Sunday December 25 2022, @08:59PM (7 children)
>We have better risk estimation now.
One would hope so.
>I think they should just build new reactors on the site now.
What you think, and I think, and the best risk estimates available today say, are far less important in the permitting process for new reactor sites than public opinion as expressed to the politicians who have ultimate veto power over any new construction.
Would be cool if "Science" could actually be trusted to get their predictions right and effectively communicated to the public often enough for the public at large to trust Science over reality TV star and other charismatic figures. Cool, but improbable today and apparently getting less likely for the near term future.
🌻🌻 [google.com]
(Score: 1) by khallow on Monday December 26 2022, @04:23AM (6 children)
Which will be fine until the places that don't give public hysteria the same weight have a growing economic advantage over those that do. Constrained democracy isn't mob rule.
Sounds like we've moved to a different set of movable goalposts here. If we can just do stuff without regard to reality, then actual risk doesn't actually matter and your risk/reality based arguments on such can be conveniently ignored.
(Score: 2) by JoeMerchant on Monday December 26 2022, @02:29PM (5 children)
>the places that don't give public hysteria the same weight have a growing economic advantage
The nature of global economics, tragedy of the commons in our atmosphere and oceans, and preponderance of economic history all detract from the credibility of your idealistic statement.
>If we can just do stuff without regard to reality
We have been doing stuff in near total ignorance of reality since forever, case in point maybe since the very beginning of the industrial revolution.
As usual you miss my meaning and pontificate out your own tangent while complaining about goal posts that you think have moved but really haven't outside your personal perception.
What I said (more directly stated for your clearer understanding) was: political hysteria has veto power over the construction of large projects.
Always implicit, but also for clarity: Reality has ultimate say in projects' success and unintentional consequences.
Our limited knowledge of reality is improving, particularly when large unintentional consequences like Fukushima happen. However, it's not a steady forward march, particularly when "Science" is funded with the aim of influencing the powerful political hysteria rather than establishing better knowledge of reality.
🌻🌻 [google.com]
(Score: 1) by khallow on Monday December 26 2022, @03:59PM (4 children)
How? I say the opposite routinely happens even in your examples. For example, offshoring. Global economics shows that if you can't do X in country Y, then you often can move production to country Z and skip the public hysteria. Even with nuclear power, France allows plenty of it even if Germany happens to think it's scary-dangerous. And there's plenty of other examples such as rare earths mining, textiles, or earning revenue free from excessive taxation. That checks off both global economics and preponderance of economic history BTW.
And there's a fair number of entertainers and protesters who operate by generating shock (Satanists, fruit mashers, flag burners). Or merely belong to an unpopular minority (Jews, Blacks, rich people). Or use recreational drugs, keep and bear firearms, prostitutes, hobby chemistry kits, or any number of unpopular or scary-dangerous activities.
Going back to nuclear power, public hysteria drives some really bad decisions like not making safer nuclear power plants because the existing ones are dangerous. You noted at one point that existing nuclear plants were operating past their expected life span. Well, the big reason why is that in many countries, new reactor construction has been effectively halted by public hysteria - for example, US, Japan, and Germany. So when you can't build new plants, but your country desperately needs the power from existing plants, then you get the situation where existing plants are run well beyond their design lifespan.
(Score: 2) by JoeMerchant on Monday December 26 2022, @08:35PM (3 children)
And France sure is kicking Germany's ass, economically...
>public hysteria drives some really bad decisions
100% agree, but... That doesn't change the real world power it wields.
>new reactor construction has been effectively halted
Since the 1-2 punch of Three Mile Island and Chernobyl, the growth rate of nuclear power plants world wide basically flat-lined.
🌻🌻 [google.com]
(Score: 1) by khallow on Monday December 26 2022, @09:19PM (2 children)
In March, 2022 Germany had household electricity prices [statista.com] of $0.46 per kWh and France $0.19 per kWh. There's much more to economics than that, but it is a huge advantage.
(Score: 2) by JoeMerchant on Monday December 26 2022, @09:42PM (1 child)
And Tennessee has super low prices due to installed hydro... Cheap electricity is nice, it makes it cheaper to run trains for one thing....
But, overall... Neither Tennessee nor France nor Venezuela are impressing me with how they are using their cheap energy to kick economic ass on neighbors with higher energy prices.
🌻🌻 [google.com]
(Score: 1) by khallow on Tuesday December 27 2022, @04:12AM
You ever going to make a point with that?
So what? I didn't say that price of electricity was the only relevant economic data point. But the take home is that Germany would be doing considerably better with a more sensible and cheaper electricity policy. Say like Texas is.
(Score: 1) by khallow on Sunday December 25 2022, @06:57PM (8 children)
Also, we're not building a nuclear plant at some random point in the last 4 billion years, but within the next few decades. 50 years of current data trumps 4 billion years of behavior that doesn't happen any more. Unless you're Velikovsky (the "Worlds in Collision" guy), you don't expect routinely planetary collisions, breakups, and such in the near future. That's the driver for the huge impact frequencies of the first billion years of the Solar System, for example.
(Score: 2) by JoeMerchant on Sunday December 25 2022, @08:49PM (7 children)
Next few decades, seems to be stretching up to 10 or more the way things are going.
Just one decade ago we had never consciously witnessed an interstellar object in this solar system, now we have confirmed two...
Point is, there is a systematic overconfidence in "The Science" which hurts the credibility that science deserves.
When a colleague read me his "mathematical proof" that the 8 bit checksum on the company's flagship product communication protocol would "protect against errors with a rate of one erroneous communication passing unblocked per so many million years" I, on my first day at the new company, expressed my scepticism but left it alone. Less than a year later we were gathered into a room to "solve the problem" which was causing erroneous (and painful) programming to pass with dozens of reports from the field in the prior 90 days.
Shit like that should not be proffered as proof to make the investors happy, but it is every day in all walks of life. That same mathematical proof wiz also regularly serves as a paid expert witness in court cases. My disgust with that system knows no bounds.
🌻🌻 [google.com]
(Score: 1) by khallow on Sunday December 25 2022, @10:10PM (6 children)
Which is irrelevant here. We have copious evidence of the actual incidence of asteroid impacts over a relevant range.
With math, you have to look at the initial conditions. Here, the likely problem was two-fold: underestimating error rates in the first place, and ignoring correlation errors - anomalies can wipe out large runs of bytes, not just one.
Here, we have no significant analogy. An asteroid or comet only brings a fixed amount of mass. A data anomaly can generate a lot of errors with no real upper bound. And as I noted before, we have a good idea of how much mass is brought to an impact and they usually come in fast enough (20+ km/s) that tidal forces don't have a significant effect.
(Score: 2) by JoeMerchant on Sunday December 25 2022, @10:40PM (5 children)
>We have copious evidence of the actual incidence of asteroid impacts over a relevant range.
If you haven't gathered: I disagree, for all the aforementioned reasons.
>Here, the likely problem was two-fold
In my estimation, after working with the expert witness for over two years, the actual problem was a desire to demonstrate what his employer wanted to hear: this won't be a problem until after you have sold all your shares. Like so many other similar demonstrations, the flaws were rooted in willful ignorance of relevant information, and he was proven painfully (the patients did experience highly painful direct neurostimulation, 3x the amplitude approved for use in humans) incorrect in a short time.
🌻🌻 [google.com]
(Score: 1) by khallow on Monday December 26 2022, @04:06PM (4 children)
Then that doesn't sound like a relevant example. The various pieces of evidence I've mentioned in this thread weren't driven by wishful thinking/willful ignorance.
(Score: 2) by JoeMerchant on Monday December 26 2022, @08:44PM (3 children)
>weren't driven by wishful thinking/willful ignorance.
Do you actually know this, or just assume that academians tell the truth because... because?
The more I learn, the lower the ratio of bonafide trustable information becomes. There is still good information out there, but it is far more common to learn a trusted source was not worthy than vice versa.
🌻🌻 [google.com]
(Score: 1) by khallow on Monday December 26 2022, @09:22PM (2 children)
Know this. And why wouldn't academicians tell the truth here? When there are problems with truth-telling, it's invariably due to conflict of interest. Here, you present no such conflict and just assert some radical doubt narrative without a reason for it.
(Score: 2) by JoeMerchant on Tuesday December 27 2022, @12:36AM (1 child)
>When there are problems with truth-telling, it's invariably due to conflict of interest.
We can start with publish or perish. Follow that up with "my grant proposal is sexier than your grant proposal" and on and on, academia is anything but perfect and the conflicts of interest can be as shallow as: my ego won't allow any other result.
🌻🌻 [google.com]
(Score: 1) by khallow on Tuesday December 27 2022, @04:14AM
So?
So no actual evidence, of course. The problem with the narrative is that you still have an allegedly too low prediction rate for asteroid impacts by two orders of magnitude and somehow that's supposed to be sexy? Maybe we'll get a better narrative if we pull your other finger?