The Statistical Worldview

[Carl-Richard]

When you're in the statistical worldview,

- you are acutely aware that many things can influence one thing, and their relationship is statistical (quantitative). Some things can have a strong influence, other things less of a strong influence, and some things only a weak influence (e.g. the butterfly effect). In reality, there is a huge web of influences, where each influence is a particular node or string on the web, and each node is weighted with a certain strength of influence or statistical value.

For example, ADHD can be influenced by beliefs, experiences, genetics, etc. Even if you think one of these things have a stronger influence, it doesn't mean it can only be reduced to that thing, and talking as if it can be reduced to that thing can lead to problems with accurately talking about and perceiving reality. Words like "partially", "mostly", "some of", "many", are often used.

- you often say things are "probably so", "most likely", "less likely", "probably not". It does not preclude you from making firm and exclusive analytical statements (e.g. "given x and y, z is true or false, coherent or inconsistent"). But you are very acutely aware when something is statistical and probabilistic so you don't overstep or overgeneralize or oversimplify.

- you realize a thing can be many things at the same time. There is often not just one way to do things, or one thing you can do at any one time. "Should I meditate every day or should I do retreats where I meditate more deeply?" Why not both? "That's the placebo effect". Why can't it be a real effect and placebo at the same time? "Trans is social contagion". Why can't some of it be real trans and some of it be social contagion (both within and across individuals)? "Yes — both" is very often realized to be the answer.

 

The statistical worldview is a way to conceptualize nuance and holism, as opposed to black-and-white thinking and naive reductionism. It's also related to the modern scientific framework of putting numbers and quantities to these relationships. Modern science, especially human-oriented science (e.g. medicine, psychology), primes this kind of statistical thinking where everything is viewed through statistical associations (mediation, moderation, correlation) and ways of quantifying them (effect sizes, correlation coefficients, measures of statistical significance). If you do enough scientific thinking, in the right fields of science, you will eventually end up viewing a large chunk of the world this way.

Edited Saturday at 10:26 PM by Carl-Richard

[Davino]

Isn't this systems thinking with emphasis on the axis of statistics in the web of relationships? 

Or are you proposing a nuance to the bayesian worldview?

Edited Friday at 10:57 PM by Davino

[Carl-Richard]

4 hours ago, Davino said:

Isn't this systems thinking with emphasis on the axis of statistics in the web of relationships?

Systems thinking is thinking with emphasis on relationships and challenging simple analytical thinking (naive reductionism: "a -> b"), often dealing with notions like complexity, circularity, context. I'm not necessarily invoking systems thinking as much as multiplistic thinking, simply acknowledging there are multiple things, and these things are related in degrees. "Multiple" is statistical, "degrees" also. Probabilistic thinking acknowledges degrees of probability of multiple outcomes; that's also statistical.

 

4 hours ago, Davino said:

Or are you proposing a nuance to the bayesian worldview?

Bayesian thinking is a very specific framework. The probabilistic aspect I'm talking about is simply about acknowledging probabilities. It's very simple.

Everybody should be familar with the concepts I'm talking about. It's just some are maybe less deeply practiced in it or less able to spot the common errors our mind makes. Especially the last paragraph about spotting how things can be two or more things at the same time. For example about the Placebo effect. People often seem to have an idea that once "placebo" is invoked, every other effect is somehow irrelevant. But it doesn't have to be that way, and it most probably isn't that way in most cases.

Edited Saturday at 03:14 AM by Carl-Richard

[zurew]

I like this approach, and I mostly live by this -  but when it comes to applying it to meta frames and different meta-philosophy frames , it can get so complicated so fast that almost no one practice it at that level (or they might, but given that they are nowhere near informed about other alternative views and frames, their probability assignment will be meaningless and it will only represent a very small fraction of things that they know about).

This is not a criticism of this approach though, because the problem I proposed is not even related to this, it is related to our cognitive limitations.

 

Like it would be very cool if we would be able to do something like this - " Okay, currently here is 10000 different possible positions that can be taken on metaphyiscs , here is a 100 different theories in phil of language, here is 10000 different views on epistemology ,  here is 10000 different views on ethics  - I understand each and every one of these theories in depth and let me now create a 100 million different configurations (worldviews) using  these views and then let me think through all the possible problems under each and every one of these configurations and then let me come up with a synthesis that has the least amount of problems  with it and let me also assign a probability to each and every one of these views "

Edited Saturday at 07:14 AM by zurew

[LastThursday]

It would be interesting to come up with a model of non-standard thinking. I'd say the default mode for most folks is social thinking: who did or said what to whom and why, how one person relates to another, what you feel about things. Here's a random bullet list pulled out of nowhere of different types of thinking:

• Social - you keep a ledger of interactions between things (people), you apply a value function (feelings) to those interactions
• Relational - you accept that nothing happens in isolation, everything affects everything else with varying strengths
• Systems - everything works like a machine with distinct interacting parts, the whole is greater than the sum of the parts
• Construct aware - you realise that every type of thing is an artificially constructed entity made of other things
• Causal - you know that if A and B happen, there are different scenarios for their synchronicity: A caused B, A and B were caused by C etc.
• Statistical - see @Carl-Richard
• Ambiguous - you accept that it is not possible to know the detail of causes for A and B and therefore have to conclude that A and B are equally likely even if contradictory. You know that knowledge and information are always lacking.
• Big picture - you zoom out or bring more of the world into the scenario in order to explain things, i.e. there a nearly always outside influences, outside of your knowledge.
• Meta - you constantly try and see problems from different angles, and through different paradigms, or by thinking laterally.
• Abstract - you use ideas not rooted in every day things to explain things: mathematics, language, symbols, geometry, numbers.

I also wonder if there can be a distinct progression in different styles of thinking. 

Edited Saturday at 07:42 AM by LastThursday

[caspex]

8 hours ago, LastThursday said:

It would be interesting to come up with a model of non-standard thinking. I'd say the default mode for most folks is social thinking: who did or said what to whom and why, how one person relates to another, what you feel about things. Here's a random bullet list pulled out of nowhere of different types of thinking:

• Social - you keep a ledger of interactions between things (people), you apply a value function (feelings) to those interactions
• Relational - you accept that nothing happens in isolation, everything affects everything else with varying strengths
• Systems - everything works like a machine with distinct interacting parts, the whole is greater than the sum of the parts
• Construct aware - you realise that every type of thing is an artificially constructed entity made of other things
• Causal - you know that if A and B happen, there are different scenarios for their synchronicity: A caused B, A and B were caused by C etc.
• Statistical - see @Carl-Richard
• Ambiguous - you accept that it is not possible to know the detail of causes for A and B and therefore have to conclude that A and B are equally likely even if contradictory. You know that knowledge and information are always lacking.
• Big picture - you zoom out or bring more of the world into the scenario in order to explain things, i.e. there a nearly always outside influences, outside of your knowledge.
• Meta - you constantly try and see problems from different angles, and through different paradigms, or by thinking laterally.
• Abstract - you use ideas not rooted in every day things to explain things: mathematics, language, symbols, geometry, numbers.

I also wonder if there can be a distinct progression in different styles of thinking. 

I like this idea.

[Joseph Maynor]

This can be combined with the analogical worldview which is what I have from working in law.  This is where you can spot the metaphor in almost every statement and know how those analogies are working.  And you can criticize the strength of that this is like that frame.

[Carl-Richard]

Visualizations of the statistical worldview (mediation models; one variable affecting another through one or more variables):

For the uninitiated, C' is the effect that remains when you account for / control for all the other effects in the model, C is the total effect (all the effects combined). Bigger number means bigger effect, negative number means negative effect (but how to interpret that depends on how the variables are conceptualized/measured/operationalized).

The more stars the more significant the effect (the more we trust the results to be accurate); complete lack of stars means the effect is not significant (considered not worthy of consideration in this data set). The numbers in the parentheses are related to the stars (if it's too high relative to the effect, it means lack of stars).

Edited yesterday at 09:34 AM by Carl-Richard

[zurew]

A good chunk of hypothesis picking  will come down to abductive reasoning - So Carl, do you have any thoughts on IBE (abductive, inference to the best explanation arguments) - more specifically how to cash out different epistemic virtues when you compare two or more hypotheses?

More specifically would be curious about how you approach issues like miracle claims or supernatural claims and you how update your view and whether you have some kind of sophisticated principled approach by which you can compare supernatural claims to natural claims and would be also curious by what epistemic virtues you compare hypotheses.

 

I will give you a specific example , and if you have an answer to this, then please walk me through your reasoning and approach using IBE or your statistical worldview and show me how you assign probability and or how you decide which hypothesis is more likely to be true and why.

Example: There is an event that breaks the known laws of physics and you need to pick which hypothesis is correct or more likely to be true.

• H1) A powerful enough agent broke the actual laws of physics 
• H2) The actual laws of physics was never broken, we just had a wrong understanding of the actual laws of physics , so the event that we observed was consistent with the actual laws of nature.
• H3) The laws of physics was actually broken, but no agent caused it, and this event was brute in principle, meaning nothing caused/triggered the event and it doesnt have any further explanation in principle.

(I dont think these three hypotheses exhaust the possibility space, but for the sake of not making this issue even more complex than it already is, lets just only entertain and compare these three hypotheses).

 

 

Side note: There is so much more that I would want to talk about related to worldview forming and hypotheses picking and issues surrounding these things (like your views on the PSR, your potential approach about how you navigate disagreements about assigned priors to hypotheses specifically in the context of apriori arguments and more) but I dont want to overwhelm you and the earlier given example is already complex and time-consuming enough.

Edited 17 hours ago by zurew

[Joseph Maynor]

On 5/9/2026 at 0:40 AM, LastThursday said:

It would be interesting to come up with a model of non-standard thinking. I'd say the default mode for most folks is social thinking: who did or said what to whom and why, how one person relates to another, what you feel about things. Here's a random bullet list pulled out of nowhere of different types of thinking:

• Social - you keep a ledger of interactions between things (people), you apply a value function (feelings) to those interactions
• Relational - you accept that nothing happens in isolation, everything affects everything else with varying strengths
• Systems - everything works like a machine with distinct interacting parts, the whole is greater than the sum of the parts
• Construct aware - you realise that every type of thing is an artificially constructed entity made of other things
• Causal - you know that if A and B happen, there are different scenarios for their synchronicity: A caused B, A and B were caused by C etc.
• Statistical - see @Carl-Richard
• Ambiguous - you accept that it is not possible to know the detail of causes for A and B and therefore have to conclude that A and B are equally likely even if contradictory. You know that knowledge and information are always lacking.
• Big picture - you zoom out or bring more of the world into the scenario in order to explain things, i.e. there a nearly always outside influences, outside of your knowledge.
• Meta - you constantly try and see problems from different angles, and through different paradigms, or by thinking laterally.
• Abstract - you use ideas not rooted in every day things to explain things: mathematics, language, symbols, geometry, numbers.

I also wonder if there can be a distinct progression in different styles of thinking. 

I would say this would be the wrong approach.  This is way too theoretical and nobody will integrate this.

[LastThursday]

@Joseph Maynor you've hit the nail on the head. Nobody thinks "statistically" for example, because it's hard to integrate. And, thinking "theoretically" isn't something that comes at all naturally to most. Nevertheless, it is possible to think differently, but as you say it isn't easy to integrate. My list was just a toy model, nothing more.

[Carl-Richard]

5 hours ago, zurew said:

Example: There is an event that breaks the known laws of physics and you need to pick which hypothesis is correct or more likely to be true.

• H1) A powerful enough agent broke the actual laws of physics 
• H2) The actual laws of physics was never broken, we just had a wrong understanding of the actual laws of physics , so the event that we observed was consistent with the actual laws of nature.
• H3) The laws of physics was actually broken, but no agent caused it, and this event was brute in principle, meaning nothing caused/triggered the event and it doesnt have any further explanation in principle.

I actually have a hard time imagining an example of something breaking the known laws of physics in a way that relates to the listed hypotheses. Can you give me a concrete example?

Edited 12 hours ago by Carl-Richard

[zurew]

55 minutes ago, Carl-Richard said:

Can you give me a concrete example?

A macroscopic object suddenly spawns in your room out of thin air.

For instance a chair or 10 million dollars or anything else.

 

or maybe a better example that would be more applicable would be something like : there being 0 gravity (but just exclusively inside your room) for a duration of 10 minutes.

 

But the difficulty to come up with an example that can tease out the relevant epistemic differences between the listed hypotheses is sort of part of the problem I want to highlight. It seems to be the case, that even in the context of a random event that violates the laws of physics - even in that context -  we still have almost 0 epistemic tools at our disposal how to figure out which of the listed hypotheses is more likely to be true (especially if we cant assume anything specific about the desires or nature of the powerful agent).

Edited 11 hours ago by zurew

[integral]

The vast majority of people that have this world view are hyper detail oriented, where they're epistemology is just what statistical scientists told them. 

I want to see how a statistical worldview functions in at least 5 different epistemic derivations.

Edited 11 hours ago by integral

[Carl-Richard]

1 hour ago, zurew said:

A macroscopic object suddenly spawns in your room out of thin air.

For instance a chair or 10 million dollars or anything else.

 

or maybe a better example that would be more applicable would be something like : there being 0 gravity (but just exclusively inside your room) for a duration of 10 minutes.

 

But the difficulty to come up with an example that can tease out the relevant epistemic differences between the listed hypotheses is sort of part of the problem I want to highlight. It seems to be the case, that even in the context of a random event that violates the laws of physics - even in that context -  we still have almost 0 epistemic tools at our disposal how to figure out which of the listed hypotheses is more likely to be true (especially if we cant assume anything specific about the desires or nature of the powerful agent).

I mean the known laws of physics is just the known laws of physics. If something appears to break those, that doesn't give much of a sign that it broke "the" laws of physics. And if the laws of physics is just reality and not some conceptual model or constraint we have defined, then I would say it was in line with the laws of physics. Everything would be in line with the laws of physics.

But this is a categorical issue, not about probabilities. You would have to define more what you mean about the laws of physics if I were to start going into probabilities.

[zurew]

7 hours ago, Carl-Richard said:

I mean the known laws of physics is just the known laws of physics. If something appears to break those, that doesn't give much of a sign that it broke "the" laws of physics. And if the laws of physics is just reality and not some conceptual model or constraint we have defined, then I would say it was in line with the laws of physics. Everything would be in line with the laws of physics.

But this is a categorical issue, not about probabilities. You would have to define more what you mean about the laws of physics if I were to start going into probabilities.

Okay I grant that under most interpretations "actually breaking the laws of physics" is a gibberish statement, but I think I can give a sense  under  which it is intelligible.

Lets say that by nature we just mean the Universe, and the Universe has certain behavioral patterns (behavior that repeats and that can be observed and replicated given the necessary conditions universally everywhere) and that would be basically the "actual" laws of physics. "Breaking" it would mean changing those behavioral patterns, meaning - even if you replicate the exact same set of conditions, the behavior that applied before do not apply anymore. 

 

But we can do fine-tuning or miracle hypotheticals if they are easier to make sense (where lets say multiple limbs are fully grown back under 1 second).  

Im just curious how you run through miracle examples, and supernatural examples , and how you update in principle towards supernaturalism being a more probable explanation for any given event. Because, I have seen your criticism of Bernardo, and I genuinely struggle to see it in principle how you can update towards a more robust supernatural view using abductive or any kind of reasoning. And this isnt about closed mindedness for me, this is genuinely lacking the epistemic ability to deal with underdetermination issues (meaning any given observable event or state of affair will be compatible with both naturalism and supernaturalism). 

For instance - I dont understand how and why given NDE facts you update towards God having metacognition, values , desires rather than staying with God not having any of the listed things. Like why think that NDE facts are more expected under an agential God than under a non-agential one? And if they are not more expected, then what reasons can you appeal to that should motivate Bernardo to update towards an agential God given NDE facts?

Because if the agential God is merely  a just so story (its merely constructed to explain the set of facts on the table and it doesnt make any novel predictions) and it commits you to make trade-offs on other epistemic virtues ,then why would you ever update towards it being the case?

Edited 3 hours ago by zurew

[AtmanIsBrahman]

My concern is that the statistical worldview precludes making meaningful statements. Sure, it’s good to be aware that (almost) everything is context-dependent and not 100% x or y, but it’s important to make clear statements too.

If someone is asking for practical meditation advice, it’s probably best to give an answer one way or another based on your experience or say you don’t know rather than making a vacuous statement like “why not both” or “it depends.”

I think the way of thinking you’re describing @Carl-Richard is a form of rationalism, which comes with many limitations. This is what Leo’s deconstructing rationalism series was aimed at. If you’re so much of a rationalist that you’re qualifying everything with phrases like “probably” or “most likely,” you’ve definitely gone wrong somewhere.

[zurew]

10 hours ago, Carl-Richard said:

But this is a categorical issue, not about probabilities.

I think what you call categorical issue that is basically where one of the main problem of hypothesis  creation and picking lies and its highly related to probability issues as well imo.

There is a tradeoff between specificity and vagueness in hypothesis creation. The more vague a hypothesis is, the more weight evidence for that given hypothesis will carry, because it will confirm a larger set of things (for instance, if you want to argue for God, then finding evidence that is expected under God simpliciter - meaning a God with any set of desires and nature) that evidence will confirm God's existence in general and given this evidence , it doesnt matter whether you are wrong about the specific attributes of God. However, if the hypothesis is too vague, then it wont make any expectations and hence you wont be able to find any evidence for it. (I personally dont think God simpliciter makes any expectations whatsoever)

On the other hand, the good thing with more specified hypotheses is that its easier to find evidence for them, because they make more traceable expectations. The bad thing about specified hypotheses is that they are much more to be rendered false, because their specificity gives them more attack surface to be wrong about. (basically they affirm more propositions, and affirming more propositions makes you more prone to be wrong all else equal, because if just one of those affirmed propositions turn out to be false, then your hypothesis will be immediately rendered false). 

 

And this is how this problem relates to probabilities: Vague hypotheses are more likely to be true (because they are compatible with a large number of specific hypotheses), but harder to be confirmed, because harder to find evidence for them. Specific hypotheses are less likely to be true and its easier to find evidence for them.

 

Edited 49 minutes ago by zurew

[Carl-Richard]

3 hours ago, AtmanIsBrahman said:

My concern is that the statistical worldview precludes making meaningful statements. Sure, it’s good to be aware that (almost) everything is context-dependent and not 100% x or y, but it’s important to make clear statements too.

The statistical worldview is largely confined to observations or empiricism (what science is primarily concerned with; how reality behaves dynamically in the world, moment to moment; factual things, things you can measure; essentially gathering and making conclusions about data). That's why I said it does not preclude you from making analytical statements that have clear-cut Yes- and No-answers (which is more general, more philosophical, less tied to on-the-ground facts, more tied to logical contigencies, pure structure). Theorizing, hypothesizing, the thing you do before measuring, before putting a number to things.

 

3 hours ago, AtmanIsBrahman said:

If someone is asking for practical meditation advice, it’s probably best to give an answer one way or another based on your experience or say you don’t know rather than making a vacuous statement like “why not both” or “it depends.”

"Try meditation", or "try this in your meditation". Suggesting something does not require absolutistic statements of fact.

 

3 hours ago, AtmanIsBrahman said:

I think the way of thinking you’re describing @Carl-Richard is a form of rationalism, which comes with many limitations. This is what Leo’s deconstructing rationalism series was aimed at.

Then I will say "why not both?". Deconstructing rationality does not preclude you from engaging in rationality. I did not describe the statistical worldview as an exclusive one (and perhaps that would be antithetical to the idea). I also thought about calling it "statistical thinking" instead but I feel it encapsulates a larger way of viewing the world than simply a way of thinking, it (ironically) encapsulates "many" ways of thinking (about structure in space, time, of how to use language and approach problems). But of course it's not an all-encapsulating thing or worldview, at least for me. It's just seemingly a large part of how my thinking mind works.

 

3 hours ago, AtmanIsBrahman said:

If you’re so much of a rationalist that you’re qualifying everything with phrases like “probably” or “most likely,” you’ve definitely gone wrong somewhere.

Not everything, only some things.

Edited 2 hours ago by Carl-Richard