Options

Blog - Markov models of social change (part 1)

Hey Team,

I hear we are meant to put a forum topic up for when we are working on a blog post.

Well, here it is. The article in question is:

Currently in the very early stages of draft. I'll get back here once I've actually got something to show/critique.

Not exactly sure of the tone balance between casual/formalism, curently trying a few things out, we'll see what happens.

Cheers, Alastair.

Comments

  • 1.
    edited December 2013

    Hi! Sorry to take so long to respond... your post had slipped by me. Let me know when I should comment!

    For those not in the know, I'm hoping to work with Alastair, Vanessa Schweizer and Matteo Smerlak on the "cross impact balance" formalism, and Alistair offered to write a blog post explaining it.

    Comment Source:Hi! Sorry to take so long to respond... your post had slipped by me. Let me know when I should comment! For those not in the know, I'm hoping to work with Alastair, Vanessa Schweizer and Matteo Smerlak on the "cross impact balance" formalism, and Alistair offered to write a blog post explaining it.
  • 2.
    Hi John,

    Article is now pretty much finished. Just getting Feedback, both from collaborators from the original project, and also from anyone up on the forum who has a interest.
    Comment Source:Hi John, Article is now pretty much finished. Just getting Feedback, both from collaborators from the original project, and also from anyone up on the forum who has a interest.
  • 3.

    Great! I will start correcting typos and doing other polishing, aiming to be done by the time Vanessa finishes a first draft of her article, so people can read hers before they've forgotten yours.

    I hope other people have comments. One of the easiest ways people can contribute to Azimuth is to help make the articles we post really good.

    Comment Source:Great! I will start correcting typos and doing other polishing, aiming to be done by the time Vanessa finishes a first draft of her article, so people can read hers before they've forgotten yours. I hope other people have comments. One of the easiest ways people can contribute to Azimuth is to help make the articles we post really good.
  • 4.
    edited January 2014

    Nice work. Good style and delivery, and the subject matter is relevant and cool.

    I agree with the thought that you were considering, about splitting it into a multi-part blog, along the Seam that you indicated. Then you could end with a recap, and an indication that next time you will go into stochastic CIB analysis. Your first segment, leading up to this seam, is a good length, and contains a complete unit of thought. Then you could take a "writers breath" between the blog articles, and at the start of the next blog reorient yourself and the reader to what what you want to cover in the stochastic treatment. It's more complex, and deserves a planning of its own.

    Comment Source:Nice work. Good style and delivery, and the subject matter is relevant and cool. I agree with the thought that you were considering, about splitting it into a multi-part blog, along the Seam that you indicated. Then you could end with a recap, and an indication that next time you will go into _stochastic_ CIB analysis. Your first segment, leading up to this seam, is a good length, and contains a complete unit of thought. Then you could take a "writers breath" between the blog articles, and at the start of the next blog reorient yourself and the reader to what what you want to cover in the stochastic treatment. It's more complex, and deserves a planning of its own.
  • 5.
    nad
    edited January 2014

    Easy read. I agree with David.

    It would be nice to eventually get some references to applications, like games, if there are any you know of.

    I don't know, but one could eventually also mention that in a more realistic scenario the rules are probably generically not necessarily Boltzmann randomized according to individual total impact score. Like in a democracy a possible impact score for a descriptor called "public opinion" woud probably weigh in all other total impact scores. So in this case one would probably prefer to have at least something like weighted Boltzmann factors.

    The BBC climate change game seems to follow something similar like what you call cross impact balance CIB. Here the "experts" (i.e. the gamer) descriptor changes which - let's assume- were weighted based on certainty and expertise are then partially modified according to real world assumptions/rules/laws and for example eventually used to determine other descriptor changes. That is raising the descriptor CO_2 emissions may give a changed health descriptor state etc. Unfortunately the exact game description doesn't seem to be public, that is I couldn't find it. The public opinion seems to way heavily in this game. At least I usually end up in a state where environment, economy etc. is more or less fine, but I'm fired out of office because I am highly unpopular, because like there are less cars etc.

    Another thing is that by playing the game you would usually probably try to start to adapt your rules, according to long term outcomes and this may also be very different than the described Boltzmann randomization.

    Comment Source:Easy read. I agree with David. It would be nice to eventually get some references to applications, like games, if there are any you know of. I don't know, but one could eventually also mention that in a more realistic scenario the rules are probably generically not necessarily Boltzmann randomized according to individual total impact score. Like in a democracy a possible impact score for a descriptor called "public opinion" woud probably weigh in all other total impact scores. So in this case one would probably prefer to have at least something like weighted Boltzmann factors. The <a href="http://www.bbc.co.uk/sn/hottopics/climatechange/climate_challenge/aboutgame.shtml">BBC climate change game</a> seems to follow something similar like what you call cross impact balance CIB. Here the "experts" (i.e. the gamer) descriptor changes which - let's assume- were weighted based on certainty and expertise are then partially modified according to real world assumptions/rules/laws and for example eventually used to determine other descriptor changes. That is raising the descriptor CO_2 emissions may give a changed health descriptor state etc. Unfortunately the exact game description doesn't seem to be public, that is I couldn't find it. The public opinion seems to way heavily in this game. At least I usually end up in a state where environment, economy etc. is more or less fine, but I'm fired out of office because I am highly unpopular, because like there are less cars etc. Another thing is that by playing the game you would usually probably try to start to adapt your rules, according to long term outcomes and this may also be very different than the described Boltzmann randomization.
  • 6.

    David wrote:

    I agree with the thought that you were considering, about splitting it into a multi-part blog, along the Seam that you indicated.

    This "seam" was introduced by Alastair. Yet he also told me this article is approximately all he wants to say. So, I'm thinking it should be one single article. It's not very long and demanding, at least by Azimuth standards. Or am I being too optimistic here?

    Comment Source:David wrote: > I agree with the thought that you were considering, about splitting it into a multi-part blog, along the Seam that you indicated. This "seam" was introduced by Alastair. Yet he also told me this article is approximately all he wants to say. So, I'm thinking it should be one single article. It's not very long and demanding, at least by Azimuth standards. Or am I being too optimistic here?
  • 7.

    I did a round of editing Alastair's blog article, eliminating spelling and grammar mistakes, polishing some sentences, and getting the format right for the blog.

    It looks surprisingly close to being ready! A couple of things:

    1) Alastair said the Boltzmann model is "tunable", but did not define this concept. Let's say what it means. If this means something very broad like "it contains adjustable parameters", let's admit that: otherwise people (like me) may think some technical concept is being alluded to.

    2) Alastair asked Vanessa to contribute some more references.

    3) It may be good to include some pictures in the "The story", instead of making the reader click a dozen links to get the jokes. For example, under "wrath of Neptune" it would be good to have a photo of that typhoon.

    Comment Source:I did a round of editing Alastair's blog article, eliminating spelling and grammar mistakes, polishing some sentences, and getting the format right for the blog. It looks surprisingly close to being ready! A couple of things: 1) Alastair said the Boltzmann model is "tunable", but did not define this concept. Let's say what it means. If this means something very broad like "it contains adjustable parameters", let's admit that: otherwise people (like me) may think some technical concept is being alluded to. 2) Alastair asked Vanessa to contribute some more references. 3) It may be good to include some pictures in the "The story", instead of making the reader click a dozen links to get the jokes. For example, under "wrath of Neptune" it would be good to have a photo of that typhoon.
  • 8.

    It's looking really good! And it works as a complete article .

    I have a few comments, just on some of the fine points.

    • This is a matter of style, but I would suggest putting a hyphen, at least in the, to make it "stochastic cross-impact balance analysis." That cuts down on the ambiguity of the phrase. Notice also that the hyphen is used in the reference that you give, and in the main items that show up on a Google search.

    • Would be good if early on in the article you explain why CIB is called what it is. It's a study of the cross-impacts of various descriptors. Why "balance"? This isn't a reference to equilibrium, I assume. Is it because all the impacts are summed to get the total effect -- the balance?

    • The following sentence has some problems:

    Here $x$ is our current state, $x$ is some possible future state, and $I_x(y)$ is the total impact score of some possible future scenario $y$ from the perspective of $x$ (that is, it is the sum of relevant entries from the bottom summation row in the diagram above).

    Do you mean:

    Here $x$ is our current scenario, $y$ is some possible future scenario, and $I_x(y)$ is the total impact score of $y$ from the perspective of $x$ (that is, it is the sum of relevant entries from the bottom summation row in the diagram above).

    Also, I don't find the definition of $I_x(y)$ to be clear, and the parenthetical remark still doesn't quite do the job. At least for the forum, can you give an example of an $x$, a $y$, and $I_x(y)$, using the nice example table in the article.

    • Concerning this sentence:

    In the fully stochastic system we can ask "after a very long time, what is the probability that we will be in scenario $x$?

    Are there some implied caveats here, concerning the well-definition of these long term probabilities? Let $P(t,x)$ be probability that the state will be in state $x$ at time $t$. What if $P(t,x)$ does not converge to a limit, as $t$ goes to infinity? Are there theorems about such convergence, for different types of stochastic succession rules?

    If it is an issue, and you don't want to get into any subtleties here, I would still indicate the existence of the issue.

    • Towards the end of the section called "Effects of stochastic succession," you say:

    We can also notice states that while technically 'consistent' ... We might identify states that seem familiar in the real world...

    Using the terminology that you introduced earlier, don't you mean scenarios, in both of these sentences. It's nice terminology, so why not stick to it. I suggesting searching for all instances of "state" and "scenario" in the text.

    But details aside, it's a strong article, both in style and substance.

    Comment Source:It's looking really good! And it works as a complete article . I have a few comments, just on some of the fine points. * This is a matter of style, but I would suggest putting a hyphen, at least in the, to make it "stochastic cross-impact balance analysis." That cuts down on the ambiguity of the phrase. Notice also that the hyphen is used in the reference that you give, and in the main items that show up on a Google search. * Would be good if early on in the article you explain why CIB is called what it is. It's a study of the cross-impacts of various descriptors. Why "balance"? This isn't a reference to equilibrium, I assume. Is it because all the impacts are summed to get the total effect -- the balance? * The following sentence has some problems: > Here $x$ is our current state, $x$ is some possible future state, and $I_x(y)$ is the total <b>impact score</b> of some possible future scenario $y$ from the perspective of $x$ (that is, it is the sum of relevant entries from the bottom summation row in the diagram above). Do you mean: > Here $x$ is our current scenario, $y$ is some possible future scenario, and $I_x(y)$ is the total <b>impact score</b> of $y$ from the perspective of $x$ (that is, it is the sum of relevant entries from the bottom summation row in the diagram above). Also, I don't find the definition of $I_x(y)$ to be clear, and the parenthetical remark still doesn't quite do the job. At least for the forum, can you give an example of an $x$, a $y$, and $I_x(y)$, using the nice example table in the article. * Concerning this sentence: > In the fully stochastic system we can ask "after a very long time, what is the probability that we will be in scenario $x$? Are there some implied caveats here, concerning the well-definition of these long term probabilities? Let $P(t,x)$ be probability that the state will be in state $x$ at time $t$. What if $P(t,x)$ does not converge to a limit, as $t$ goes to infinity? Are there theorems about such convergence, for different types of stochastic succession rules? If it is an issue, and you don't want to get into any subtleties here, I would still indicate the existence of the issue. * Towards the end of the section called "Effects of stochastic succession," you say: > We can also notice states that while technically 'consistent' ... We might identify states that seem familiar in the real world... Using the terminology that you introduced earlier, don't you mean scenarios, in both of these sentences. It's nice terminology, so why not stick to it. I suggesting searching for all instances of "state" and "scenario" in the text. But details aside, it's a strong article, both in style and substance.
  • 9.
    edited January 2014

    Good comments, David. Clearly the second $x$ is just a typo here:

    Here $x$ is our current state, $x$ is some possible future state...

    I'll change it to a $y$. But your other remarks are also valid.

    1) I believe a "scenario" is being used as a synonym for "state" here. Flipping back and forth between different terms is always confusing to people who are trying to learn a subject and aren't sure if these terms are being used as synonyms. It's best to either stick to one term or explicitly declare that several terms are being used as synonyms.

    2) Also, Alastair should probably introduce some notation for the entries of the all-important matrix (e.g. the matrix in the ant farm example), and clearly explain how $I_y(x)$ is computed as a sum of matrix entries.

    About this:

    Are there theorems about such convergence, for different types of stochastic succession rules?

    Yes, but I don't think Alastair should discuss this in his article; it's more technical than the rest. This is the sort of thing you should ask about on the Azimuth Blog after the article is published! I'm dying to answer, but it's better if Alastair or I answer "in public".

    Comment Source:Good comments, David. Clearly the second $x$ is just a typo here: > Here $x$ is our current state, $x$ is some possible future state... I'll change it to a $y$. But your other remarks are also valid. 1) I believe a "scenario" is being used as a synonym for "state" here. Flipping back and forth between different terms is always confusing to people who are trying to learn a subject and aren't sure if these terms are being used as synonyms. It's best to either stick to one term or explicitly declare that several terms are being used as synonyms. 2) Also, Alastair should probably introduce some notation for the entries of the all-important matrix (e.g. the matrix in the ant farm example), and clearly explain how $I_y(x)$ is computed as a sum of matrix entries. About this: > Are there theorems about such convergence, for different types of stochastic succession rules? Yes, but I don't think Alastair should discuss this in his article; it's more technical than the rest. This is the sort of thing you should ask about on the Azimuth Blog after the article is published! I'm dying to answer, but it's better if Alastair or I answer "in public".
  • 10.
    edited January 2014

    Thanks you all kindly- very helpful comments!

    I've made most of the comments mentioned- aside from explaining the "Balance" part of "CIB"- I don't know that one myself, I could guess, but that would just be my guessing.

    Good call about being more careful with terminology David- much more precise now. They are in fact meant to be different words, hopefully any confusion regarding their meanings is tidied up now.

    Example and generic formula for $I_x(y)$ now given. Hopefully clear enough?

    Also- David- nice catching the "implied Caveat". There is one there, but it is satisfied by the "Fully stochastic" condition- will discuss more on the public blogs, as John suggests.

    Okay- so, I think all I need now is to hear back from Vanessa. I know she's got several fairly relevant examples close to hand.

    Comment Source:Thanks you all kindly- very helpful comments! I've made most of the comments mentioned- aside from explaining the "Balance" part of "CIB"- I don't know that one myself, I could guess, but that would just be my guessing. Good call about being more careful with terminology David- much more precise now. They are in fact meant to be different words, hopefully any confusion regarding their meanings is tidied up now. Example and generic formula for $I_x(y)$ now given. Hopefully clear enough? Also- David- nice catching the "implied Caveat". There is one there, but it is satisfied by the "Fully stochastic" condition- will discuss more on the public blogs, as John suggests. Okay- so, I think all I need now is to hear back from Vanessa. I know she's got several fairly relevant examples close to hand.
  • 11.
    edited January 2014

    I'll check out the changes—thanks, Alastair!

    By the way, simple LaTeX commands work here and on the wiki... but here you have to choose the "Markdown+ITeX" button under your comment as you're writing it. If you have cookies enabled, you just need to make that choice once.

    So, I did that in your comment above and changed I_x(y) to $I_x(y)$. No big deal... I hope this remark is useful for all the newish folks who read this!

    Comment Source:I'll check out the changes&mdash;thanks, Alastair! By the way, simple LaTeX commands work here and on the wiki... but here you have to choose the "Markdown+ITeX" button under your comment as you're writing it. If you have cookies enabled, you just need to make that choice once. So, I did that in your comment above and changed I_x(y) to $I_x(y)$. No big deal... I hope this remark is useful for all the newish folks who read this!
  • 12.
    edited January 2014

    Thanks for making the math stuff more precise, Alastair!

    I adjusted all the pictures so they will display when I move this article to the blog. The correct format for pictures in blog entries is something like this:

    <div align = "center"><a href="http://en.wikipedia.org/wiki/Intergovernmental_Panel_on_Climate_Change"><img src = "http://math.ucr.edu/home/baez/ecological/jamieson-lane/IPCC_report.jpg" alt = ""/></a>

    I also moved the pictures around a bit, and changed one. I thought that since your myth mentions "burning lumber", a picture of burning Amazon rainforest would be nicer than a picture of an oil well. I then changed the words to "burning the forests"... with a link to the Wikipedia article on greenhouse gas emissions.

    I think this article is great! It's ready except:

    1. It needs some more references (or else: delete the part where it says more references are coming).

    2. It needs Vanessa's companion article.

    Comment Source:Thanks for making the math stuff more precise, Alastair! I adjusted all the pictures so they will display when I move this article to the blog. The correct format for pictures in blog entries is something like this: `<div align = "center"><a href="http://en.wikipedia.org/wiki/Intergovernmental_Panel_on_Climate_Change"><img src = "http://math.ucr.edu/home/baez/ecological/jamieson-lane/IPCC_report.jpg" alt = ""/></a>` I also moved the pictures around a bit, and changed one. I thought that since your myth mentions "burning lumber", a picture of burning Amazon rainforest would be nicer than a picture of an oil well. I then changed the words to "burning the forests"... with a link to the Wikipedia article on greenhouse gas emissions. I think this article is great! It's ready except: 1. It needs some more references (or else: delete the part where it says more references are coming). 2. It needs Vanessa's companion article.
  • 13.
    edited January 2014
    Correct format for pictures duly noted- sorry about that. There is much to learn.

    References have now been added- Hopefully they are in the right format. I'm a little uncertain about some of the less standard article things (which don't have page number etc) but have tried to stick to the official forum style as much as possible.

    I think that's pretty much everything.

    Awesome.
    Comment Source:Correct format for pictures duly noted- sorry about that. There is much to learn. References have now been added- Hopefully they are in the right format. I'm a little uncertain about some of the less standard article things (which don't have page number etc) but have tried to stick to the official forum style as much as possible. I think that's pretty much everything. Awesome.
  • 14.
    edited January 2014

    Thanks for the references! I fixed up the format a bit and made sure all links point to freely available resources.

    (This is an Azimuth thing: since we're aiming at a wide audience that includes nonacademics, there's not much point referring to papers that you have to belong to a university with a journal subscription to read. Luckily all but one reference had a freely available version.)

    I will copy this post over to the blog and wait until Vanessa has finished her companion piece before publishing it. She said she'll get less busy after January 15th.

    Thanks again!

    Comment Source:Thanks for the references! I fixed up the format a bit and made sure all links point to _freely available_ resources. (This is an Azimuth thing: since we're aiming at a wide audience that includes nonacademics, there's not much point referring to papers that you have to belong to a university with a journal subscription to read. Luckily all but one reference had a freely available version.) I will copy this post over to the blog and wait until Vanessa has finished her companion piece before publishing it. She said she'll get less busy after January 15th. Thanks again!
  • 15.
    edited January 2014

    I changed the title of this blog article to "Stochastic cross-impact balance analysis", inserting a hyphen — because as David Tanzer pointed out, that makes it clearer.

    Luckily, that also seems to be the way most people in the subject write this term.

    Comment Source:I changed the title of this blog article to "Stochastic cross-impact balance analysis", inserting a hyphen &mdash; because as David Tanzer pointed out, that makes it clearer. Luckily, that also seems to be the way most people in the subject write this term.
  • 16.
    Oh- I thought if we attached the link to article title itself it needed freely available, and if it was on the journal title it could be pay to enter- Hence that was the convention I tried to stick to.
    Didn't realize freely available versions were floating around- should have done more exploring.

    Hmmm... I clearly have a lot to learn about writing blog posts. :-/
    Comment Source:Oh- I thought if we attached the link to article title itself it needed freely available, and if it was on the journal title it could be pay to enter- Hence that was the convention I tried to stick to. Didn't realize freely available versions were floating around- should have done more exploring. Hmmm... I clearly have a lot to learn about writing blog posts. :-/
  • 17.

    Alastair wrote:

    Oh- I thought if we attached the link to article title itself it needed to be freely available, and if it was on the journal title it could be pay to enter. Hence that was the convention I tried to stick to.

    Yes, that's the convention. But if we're just giving someone a pile of references to read, we might as well give them references that are free, unless the non-free ones are especially important. So, I got rid of one non-free reference and found a lot of free ones.

    I'm waiting for Vanessa's blog post to appear... I don't want to pester her yet, but maybe you can.

    Comment Source:Alastair wrote: > Oh- I thought if we attached the link to article title itself it needed to be freely available, and if it was on the journal title it could be pay to enter. Hence that was the convention I tried to stick to. Yes, that's the convention. But if we're just giving someone a pile of references to read, we might as well give them references that are free, unless the non-free ones are especially important. So, I got rid of one non-free reference and found a lot of free ones. I'm waiting for Vanessa's blog post to appear... I don't want to pester her yet, but maybe you can.
  • 18.

    I've changed the title of this blog article (and this thread) to "Markov models of social change (part 1)". I'll publish it really soon.

    Comment Source:I've changed the title of this blog article (and this thread) to "Markov models of social change (part 1)". I'll publish it really soon.
  • 19.
    edited February 2014

    I published this article on Monday:

    Alastair: it would be really great if you and/or Vanessa could respond to comments. Part of the advantage of blogging over traditional publication is the conversations that arise. Admittedly none of the comments so far have been very deep.

    Comment Source:I published this article on Monday: * Alastair Jamieson-Lane, [Markov models of social change (part 1)](http://johncarlosbaez.wordpress.com/2014/02/24/markov-models-of-social-change-part-1), _Azimuth_, 24 February 2014. Alastair: it would be really great if you and/or Vanessa could respond to comments. Part of the advantage of blogging over traditional publication is the conversations that arise. Admittedly none of the comments so far have been very deep.
Sign In or Register to comment.