Amazing and critical read. Critical in this times were machines "Thinking" is evolving exponentially.
Like you said, we build patterns over time based on our experiences and data. Machines do that too, but we do it on a dimension that's not understandable for a machine. The emotional dimension, a computer will never like a tacos restaurant over another, but they can integrate millions of tacos reviews in their system (we cant do that).
Machines capabilities are horizontal, human capabilities are vertical.
I belive it's important to understand this, as we integrate technology into our life and work.
Exceptional explanations with great humor. Graphics helpful and appreciate the book recommendations. Will now be looking for Bayes projects everywhere!
What do you recommend for people without access to paper and pen, phone, app, when they need to make a decision based on priors using Bayesian updating?
Data "in formation" becomes knowledge, and through a recursive, ever collapsing and expanding series of wave functions becomes what we might call "Schrödinger’s Wisdom"?
Only when data is arranged in a particular formation does it become knowledge. A best guess of where we think the particle might be, a collapsing of the wave function. But we are always adding new data, forcing us to reconfigure the formation and update or knowledge. We change our best guess of the particles location, expanding the wave function. The problem is recursive, and ever ongoing. Therefore "...no matter how much we know about something, we never know everything and there’s always room for more information". We can never know if our knowledge leads to wisdom, me might find new data in the box.
"It is the mark of an educated mind to be able to entertain a thought without accepting it"
It was a play on words that I felt summed up what you saying, expressed as a question. Feel free to disregard.
I found this written well and clear but I have a doubt: Normally when we face these situations in life we go by gut feeling/intuition, we don’t really sit and calculate the Bayesian probability.
1) Would it be considered excessive if we start judging situations by doing Bayesian calculations? I suppose it’s not needed to do this every time yeah? Is it okay to make decisions based solely on this math?
2) By doing these calculations enough number of times, can we get better at it and not require sitting down every time to calculate? Have you been able to do that in your experience? I ask because it’s a little cumbersome to do this.
Thanks for a great article once again. Looking forward to more articles on Bayesian thinking.
Thanks, Abhiraj! Yes, it would be excessive to expect calculations every time, but it is important to understand how reasoning with data works. We do this intuitively anyway, and clarifying the structure behind the process is enlightening.
I already know there’s bad weather which means there’s 60% chance of train delay. Isn’t that the answer? Or should I take into consideration the probability of a rush hour overlap? In which case:
Given that,
Probability of rush hour happening irrespective of bad weather is 30%
Probability of train delay given that rush hour occurs is 50%
Combining these two by multiplication gets me 0.15 or 15% chance of train delay if rush hour happens (in 30% of the cases).
That first assumption is quite unfounded, as thought doesn’t follow chronological linearity in experience, nor does one precedence deny simultaneous patterns & lateral or circular cohesion, also the “we” form is quite invasive identification, no offence
Don’t you mean ‘Bosnian Critical Thinking’?. . But Bayesian thinking is what a bookie at the racetrack does all the time altering the odds on his board to reflect the direction of the crowd’s money not the actual odds of a particular nag coming in first, second or third
Amazing and critical read. Critical in this times were machines "Thinking" is evolving exponentially.
Like you said, we build patterns over time based on our experiences and data. Machines do that too, but we do it on a dimension that's not understandable for a machine. The emotional dimension, a computer will never like a tacos restaurant over another, but they can integrate millions of tacos reviews in their system (we cant do that).
Machines capabilities are horizontal, human capabilities are vertical.
I belive it's important to understand this, as we integrate technology into our life and work.
Thanks, Boris, I appreciate it. I love how you put it: machine capabilities are horizontal, human capabilities are vertical.
Exceptional explanations with great humor. Graphics helpful and appreciate the book recommendations. Will now be looking for Bayes projects everywhere!
Thanks, Cathie, I’m glad you like it!
Clear and educational, thanks.
What do you recommend for people without access to paper and pen, phone, app, when they need to make a decision based on priors using Bayesian updating?
That’s a really good question. Right now I’d say just try to weigh the priors and the data and try to combine them as suggested by the theorem.
I’ll think more about other possible ways of processing that in our mind, thanks for prompting me!
Data "in formation" becomes knowledge, and through a recursive, ever collapsing and expanding series of wave functions becomes what we might call "Schrödinger’s Wisdom"?
Interesting way to put it. Care about expanding and explaining more?
Only when data is arranged in a particular formation does it become knowledge. A best guess of where we think the particle might be, a collapsing of the wave function. But we are always adding new data, forcing us to reconfigure the formation and update or knowledge. We change our best guess of the particles location, expanding the wave function. The problem is recursive, and ever ongoing. Therefore "...no matter how much we know about something, we never know everything and there’s always room for more information". We can never know if our knowledge leads to wisdom, me might find new data in the box.
"It is the mark of an educated mind to be able to entertain a thought without accepting it"
It was a play on words that I felt summed up what you saying, expressed as a question. Feel free to disregard.
Thanks, I love the analogy! I think you’re onto something here, seems worthy of more thought.
It's yours now.
I haven’t enjoyed a post this much in quite some time. You are an excellent teacher and your writing is humorous and well paced.
Thank you
Thanks, I appreciate it!
I found this written well and clear but I have a doubt: Normally when we face these situations in life we go by gut feeling/intuition, we don’t really sit and calculate the Bayesian probability.
1) Would it be considered excessive if we start judging situations by doing Bayesian calculations? I suppose it’s not needed to do this every time yeah? Is it okay to make decisions based solely on this math?
2) By doing these calculations enough number of times, can we get better at it and not require sitting down every time to calculate? Have you been able to do that in your experience? I ask because it’s a little cumbersome to do this.
Thanks for a great article once again. Looking forward to more articles on Bayesian thinking.
Thanks, Abhiraj! Yes, it would be excessive to expect calculations every time, but it is important to understand how reasoning with data works. We do this intuitively anyway, and clarifying the structure behind the process is enlightening.
How do I approach exercise 3?
I already know there’s bad weather which means there’s 60% chance of train delay. Isn’t that the answer? Or should I take into consideration the probability of a rush hour overlap? In which case:
Given that,
Probability of rush hour happening irrespective of bad weather is 30%
Probability of train delay given that rush hour occurs is 50%
Combining these two by multiplication gets me 0.15 or 15% chance of train delay if rush hour happens (in 30% of the cases).
How do I combine these two (the 60% and 15%)?
Have I understood the problem correctly?
Thanks in advance!
I think you did understand it correctly. We always have to combine prior information and data (that's the whole point of Bayesian updates).
Oh okay got it! Thanks!
But that's not Kant, is it?
That first assumption is quite unfounded, as thought doesn’t follow chronological linearity in experience, nor does one precedence deny simultaneous patterns & lateral or circular cohesion, also the “we” form is quite invasive identification, no offence
None taken because I have no idea what are you talking about 😀
Precisely why writing in a "we"format is an invasive identification, there is no we to assume, nor is thought linear ...
Don’t you mean ‘Bosnian Critical Thinking’?. . But Bayesian thinking is what a bookie at the racetrack does all the time altering the odds on his board to reflect the direction of the crowd’s money not the actual odds of a particular nag coming in first, second or third
Yes, but I’m sure he starts with some priors and then updates them. Unfortunately, there’s no such thing as Bosnian Critical Thinking 😀
Awesome example, thanks! I’ll use that next time!