Is the Carter(Doomsday) hypothesis valid?

[Vanamonde]

This thing irritates the hell out of me because it is obviously wrong, but I can't pinpoint why it's wrong. 

Perhaps this? The hypothesis attempts to draw a conclusion from a sample, but the sample is not finalized in the sense that more items are still being added to the set and to the sample. 

[Superfluous J]

1 hour ago, YNM said:

But what's the flaw ? Is it a flaw within the finiteness of anything ?

The flaw is that it makes YOU (not YOU you, but the YOU in the thought experiment) more important than anybody else in history, even while it says you're not more important.

From what I can tell, the hypothesis makes 3 assumptions*:

1. There are a finite number of humans in all of history: Fine. Not guaranteed because who knows what smart people will think up in 15 billion years but whatever. They probably won't even be "Human" then.
2. If you pick a set of random people from that entire pool, then half of them will roughly be before the midpoint of all humans and half will be after it, and on average one of them will be halfway between the first and last person in history: Fine. That's statistics and probability and even makes common sense.
3. YOU are eligible to be pick for this: Not fine. You specifically picked yourself. You may as well have picked Socrates. Or Jean Luc Picard. Or Hari Seldon**. Picking a specific person is not picking a person at random. Picking ANYBODY ALIVE TODAY is not picking a random person. Picking anybody AT ALL when you know who they are and didn't pick randomly from the pool of everybody alive in all of time and space is NOT random. Until the last human is alive and somehow has a huge list of all humans who had lived prior, it is impossible to make such a pick because the list literally does not exist until then. But even when it does exist, picking a specific person for any reason other than the random number generator spat out their name is NOT picking at random.

*It makes a lot more than 3 assumptions. For example, it assumes that a doomsday scenario is all that can end humanity. But it makes 3 that together make it invalid in my mind.

**He lives several dozen thousand years from now in Asimov's future history. He was the biggest name I could think of that lived as far in the future that I could think of off the top of my head.

Edited May 23, 2018 by 5thHorseman
Clarified my language, mostly in the 2nd point but also added the final line of point 3.

[YNM]

4 hours ago, 5thHorseman said:

The flaw is that it makes YOU (not YOU you, but the YOU in the thought experiment) more important than anybody else in history, even while it says you're not more important.

Why ? It's always the now.

Your death is your demise too.

[Superfluous J]

59 minutes ago, YNM said:

Why ? It's always the now.

Your death is your demise too.

I can't continue the discussion. You either don't understand me, don't agree with me, or are trolling. I can't explain myself better to get you to understand, won't convince you with any other points if you don't agree, and won't bother if you're trolling.

[YNM]

3 hours ago, 5thHorseman said:

You either don't understand me, don't agree with me, or are trolling.

Sure ! Someone with a track record of 13 infraction points on the forum and 18 days of monitoring, is definitely the worst troll on the forum possible.

If the possibility ever existed, it's probably a constant.

And why should I be special ? What about the first guy that existed ? Ain't they more definitely special ?

The thing is, everything always looks normal at it's time. In the ice age it was probably normal to hunt down hairy elephants. Try do that today, and the zookeeper probably got hold of you first. We adjust to our surroundings, and inevitably we never fail to adjust (or at least the younger ones tend to be able to catch up).

The chance that befel the last guy alive won't be that unnormal (special) either. At least compared to the first one ever. All he knew is that when he knew it, it's too late.

And this is why I said there's never a way to tell whether such things will occur in any given time. Everything will look normal until it happens. Then it'll just "happen". Bam. Die.

 

The same goes for personal deaths too. This is probably why many people's account with their recently deceased relative/colleague goes with "well I just conversated with him last night !" or something. You never know until it happens.

Edited May 23, 2018 by YNM

[Vanamonde]

Don't let this turn into an argument, please. There's no need to be angry with each other. 

[sevenperforce]

20 hours ago, Vanamonde said:

This thing irritates the hell out of me because it is obviously wrong, but I can't pinpoint why it's wrong. 

It's hard to pinpoint why it's wrong, but once you pinpoint it, it makes perfect sense.

The answer?

It's not wrong; it's imprecise.

Given the available sample and the question you're asking, then yes -- the expected value of the total human population across all history is 1.2 trillion. But the precision of that value is the problem. The prediction depends on the conjecture that choosing a human being alive today is a random selection, but we do not know how random that selection is. This introduces a variance on the order of the uncertainty in the randomness of the selection.

There are 7 billion people alive today out of 107 billion who have ever lived. What are the odds that picking someone at random out of today's population is, in fact, a random selection relative to the entire historical human population? Well, 6.5%. So your actual answer is "the expected value of the total human population across all history is 1.2 trillion, plus or minus 1.12 trillion. Since the prediction (1.2 trillion) is smaller than the variance (2.24 trillion), the whole exercise is hogwash.

[YNM]

8 hours ago, sevenperforce said:

the whole exercise is hogwash.

But still presents a possibility. (I'd prefer without the numbers, you can't barely tell anyway.)

[DAL59]

On 5/22/2018 at 10:52 PM, 5thHorseman said:

Picking a specific person is not picking a person at random. Picking ANYBODY ALIVE TODAY is not picking a random person.

But you can't pick someone from the future, as you don't know if they'll exist.  

[Superfluous J]

59 minutes ago, DAL59 said:

But you can't pick someone from the future, as you don't know if they'll exist.  

Exactly

[sevenperforce]

5 hours ago, YNM said:

14 hours ago, sevenperforce said:

It's not wrong; it's imprecise.

There are 7 billion people alive today out of 107 billion who have ever lived. What are the odds that picking someone at random out of today's population is, in fact, a random selection relative to the entire historical human population? Well, 6.5%. So your actual answer is "the expected value of the total human population across all history is 1.2 trillion, plus or minus 1.12 trillion. Since the prediction (1.2 trillion) is smaller than the variance (2.24 trillion), the whole exercise is hogwash.

But still presents a possibility. (I'd prefer without the numbers, you can't barely tell anyway.)

The point is that if a statistically-generated answer has error bars on the order of the answer* it generates, then the statistically-generated answer is not useful for anything.

It would be like the German tank problem, but instead of predicting a production run of 270 tanks, they said "German has made 300 tanks, plus or minus 290 tanks". If a statistical estimation produces this sort of variance then that's a clue that statistical estimation will not provide a meaningful answer.

And that's what we already knew: intuitively, estimating the timing of a doomsday event using the number of humans alive today is clearly not meaningful. Examining the error bars gives us the mathematical reason why.

_{*Technically, the breaking point is where the error bars are on the order of the range of the dependent variable in the estimation, but when the output range is a series of natural numbers, like a number of people or a number of tanks, you can approximate the range as the answer generated.}

[wumpus]

I still think the problem assumes the wrong shape of a random variable.  The assumption appears that new species come and go, while old species stay around forever.  I've heard that while this sounds likely, in practice the odds of a species go extinct has little to do with how long they have existed.  A more likely issue is how many niches/environments they can survive in.

Quick question, how old do you think the various species are on this list: https://en.wikipedia.org/wiki/List_of_threatened_sharks

Humans are bad at probability.  Unless *all* the models are explained, and the probabilities worked out, there is little reason to believe some random internet hypothesis (this even include more credited cranks who do well fooling experts in stuff outside of their field).

[YNM]

36 minutes ago, wumpus said:

Humans are bad at probability.

Engineering are forced to rely on rather inexact data.

This theory could put numbers but IMO it's just a thinking. Just because it's not accurate doesn't quite make it invalid.

[magnemoe]

6 hours ago, wumpus said:

I still think the problem assumes the wrong shape of a random variable.  The assumption appears that new species come and go, while old species stay around forever.  I've heard that while this sounds likely, in practice the odds of a species go extinct has little to do with how long they have existed.  A more likely issue is how many niches/environments they can survive in.

Quick question, how old do you think the various species are on this list: https://en.wikipedia.org/wiki/List_of_threatened_sharks

Humans are bad at probability.  Unless *all* the models are explained, and the probabilities worked out, there is little reason to believe some random internet hypothesis (this even include more credited cranks who do well fooling experts in stuff outside of their field).

The shark list is also bad, many shark species live in the deep ocean and we don't have good numbers. 
Now if an species is rare its better to put it as endangered just to be sure. 
Other who is local only and hit hard by fishing is an obvious endangered species. 

[Xeorm]

The two bits that make it seem wrong to me are the inaccuracy of the method and our knowledge of how non-linear human population has been. The first has already been shown so I don't want to go over it. But the second should be said.

The hypothesis runs only off of limited data - how many humans have existed up to now. But, there's more information than that. Take the tank problem. If you knew only the number of tanks that were produced and happened to run the numbers very shortly after the tanks went into production, you'd have very different results than if you ran the numbers later in the process. Just knowing the time frame of production would give you better data. Same with humans. The data over the entire span indicates there will be so and so number of humans, but the current population growth and number of humans right now gives the impression that we're only starting to come off of the production line. 

[kerbiloid]

Spoiler

According to wiki, Earth population was 2 bln in 1927, 4 bln in 1974, and will be 8 bln in 2024.
I.e. it grows twice per 50 year.
Human mass = 70 kg.
Earth mass = 6*10²⁴ kg.
50 * ln(6*10²⁴ / (70 * 8*10⁹)) / ln 2 = 2164 years.

So, ~2000 years later the humanity will absorb all terrestrial matter and stop growing.
This result competes with Doomsday Argument estimation and makes that theory highly speculative.

 

[sevenperforce]

6 hours ago, Xeorm said:

The hypothesis runs only off of limited data - how many humans have existed up to now. But, there's more information than that. Take the tank problem. If you knew only the number of tanks that were produced and happened to run the numbers very shortly after the tanks went into production, you'd have very different results than if you ran the numbers later in the process. Just knowing the time frame of production would give you better data. Same with humans. The data over the entire span indicates there will be so and so number of humans, but the current population growth and number of humans right now gives the impression that we're only starting to come off of the production line. 

You cannot exactly use a method to derive the total timeframe of production if one of your independent variables is the timeframe of production.

[wumpus]

17 hours ago, magnemoe said:

The shark list is also bad, many shark species live in the deep ocean and we don't have good numbers. 
Now if an species is rare its better to put it as endangered just to be sure. 
Other who is local only and hit hard by fishing is an obvious endangered species. 

You'll notice that none of your objections have anything to do with how old the species is, and the same issues ("don't have good numbers") pop up in the Carter "methodology".

[magnemoe]

7 hours ago, wumpus said:

You'll notice that none of your objections have anything to do with how old the species is, and the same issues ("don't have good numbers") pop up in the Carter "methodology".

Age of species is irrelevant, their issue is human fishing with an secondary on pollution, as we don't have good numbers its best to play safe. 
Without industrialization sharks would either run into others players as in fish on mammals getting into their habitat and win over them or some extinction level event, this is likely to be deep time or gradually 

[wumpus]

20 hours ago, magnemoe said:

Age of species is irrelevant, their issue is human fishing with an secondary on pollution, as we don't have good numbers its best to play safe. 
Without industrialization sharks would either run into others players as in fish on mammals getting into their habitat and win over them or some extinction level event, this is likely to be deep time or gradually 

The problem is that the Carter hypothesis claims the opposite.  That looking at the numbers of humans (generally limited by age of species), we can determine humanities lifespan.

It wasn't obvious that age of species was irrelevant to extinction until the anthrocene mass extinction gave us a ton of details on species going extinct.  And now we know it is irrelevant.  So the Carter hypothesis is invalid.

[kerbiloid]

Humble heroes, standing between us and Carter.
 

Spoiler

Every time when it's going to carterize U all, they reset it.

[Laie]

On 5/23/2018 at 3:29 AM, Vanamonde said:

This thing irritates the hell out of me because it is obviously wrong, but I can't pinpoint why it's wrong.

In a nutshell, "insufficient data for a meaningful answer".

All the bayesian claptrap is about how you can make reasonable assumptions from insufficient data. I think the irritation comes from the contrast between the unassailable maths and the single, almost-but-not-quite random data point that underlies it all.

[wumpus]

1 hour ago, Laie said:

In a nutshell, "insufficient data for a meaningful answer".

All the bayesian claptrap is about how you can make reasonable assumptions from insufficient data. I think the irritation comes from the contrast between the unassailable maths and the single, almost-but-not-quite random data point that underlies it all.

While the math might be unassailable, expect to find tons of assumptions about the shapes of random variables there.  Even worse, there exist plenty of controls (all other species) that are summarily ignored out of some "human exceptionalism" refusal to check the data.  There is a reason that science typically demands experimental proof (regardless of how far the disciplines of theory and experiment may have split).  Doing math and expecting it to match the physical universe without understanding the actual mechanisms leads to situations like this.  Bayes theory gives you a big step up in developing inferences if you know a good deal about a non-independent random variable.  If you are dealing with "beyesian" solutions that aren't concerned with accurately calculating said random variables (as well as the degree of independence: something often quite difficult), you can safely ignore their claptrap.

There simply wasn't attempt to check this type of thing, which is why it is such a crank paper.  You could check it against other species and find out your methods are useless (species become extinct because humans (or changes caused by humans) destroy their habitat.  Is has nothing to do with how many individuals have lived since the species evolved.  Or you could calculate rigorous error calculation on the known accuracy of your assumptions.  Since all of them are purely assumptions, watch your error hit +/- 100%.

[kerbiloid]

And also why think that population size is a scalar?
Maybe it's just a scalar convolution of a fractal value?

[Laie]

15 minutes ago, wumpus said:

While the math might be unassailable, expect to find tons of assumptions about the shapes of random variables there. 

That's what I was trying to get at, really.

Then again, the choice of data nicely follows from the approach. If you agree to the bayesian premise, you also agree to what it's based on -- then shake yourself like a wet poodle and wonder what the hell just happened.