In our new book, we argue that most talk, including mass media news and academic talk, isn’t really about info, at least the obvious base-level info. But to study talk, it helps to think about what it would in fact look like if it
This definitional change doesn't solve the fundamental issue that its is the absolute difference and not the ratio that needs to be used as the central element before multiplication ... of this I am highly confident ( :)) and now that I have bought this to your attention, I am sure you know that you can model this out with dice and cards etc to prove it for yourself.
But the big picture here is you deserve considerable credit for putting together a really interesting question and model so lets not get too hung up on mathematical minutiae, the big point I wanted to make was the value of info literature had addressed a fair amount of this and you are aware of that so we are all good.
With that minor caveat out of the way, I think you are asking some very probing questions here ... why do we have limited regard for the value of information anywhere in society? Why are we a society that doesn't really invest in information at all ... basically why are we so foolish?
my high level response is that we get what we deserve and that as humans our lack of real respect and acceptance of the truth/reality causes us to harm ourselves in very material ways. News being merely one of them. If we were more sincere about finding and acting on the truth/reality then we would probably be more happy and well adjusted beings.
When giving formulas, I had in mind applying them where the news was that a claim had *risen* in probability. Once can of course represent them with the complement event whose probability has lowered, but that is a lot less natural.
With respect to recency, it's easier to get everybody on the same page by focusing heavily on the latest. If a news organization doesn't focus on the latest but instead picks a particular topic, but nobody else picks that same topic, that's going to make it a lot harder to bring up that same topic around non-readers around the water cooler, or to get noticed in search engines.
There are far fewer people reading those alternative sources than the the number of people reading the news. Most people have picked up a newspaper or magazine once or twice. Far fewer read textbooks after they've already graduated college.
The genre one might call "Gladwellian nonfiction" runs on the premise of "shared explanation and justification of many related small claims"--such books cover multiple apparently unrelated claims, united by using the same concepts as support.
(Though I suppose the *explicit* justification is that the various topics all provide support for a central claim)
Perhaps I misunderstood you but when I try apply your definitions/methodology to a real world scenario it doesn't seem to add up to me. Take something that many people care about ... like Trump and the probability he will end his term as president.
Please can you help me reconcile the following scenario per your own definitions and methodology?
Claim: Probability of trump ending his term as president.
Surprise – how low a probability a reader would have previously assigned to this claim. For me its 75%.
Then I hear credible "news" that he has been assassinated ...
Confidence – how high a probability a reader is to assign after reading this news. This is now 0%. No chance.
Based on your calculation the "amount" of information from that news ... the ratio of confidence/surprise which is 0%/75% ... = 0 as in ZERO
Then you go and multiply 0 with importance and then multiply that by commonality etc and you still get 0.
Which basically means credible news of Trump being assassinated, has a value of ... 0?
The ratio is the likelihood ratio, which is what you can combine via multiplication across multiple news items to give total news. You can't do that with the difference.
Robin ... this is an interesting perspective and your central point is very valid There is a lot else missing/wrong from a purely correct theory perspective here but I guess the major point you are trying to make is that news sources provide low information value.
For example:
"Okay, the amount of info that some news gives a reader on a claim is the ratio of its confidence to its surprise"
This is not correct, the information value its the difference between these two ... and information value should not be confused with economic value. (how much its worth in dollars and sense to have this information) And if you are interested I can model this out for you very simply.
In fact there is a large body of work extending basic Shannon information theory to more human questions very much like you have asked here, what is the "value" of information, but most of it is in this 1966 seminal paper by Ronald Howard:
An experience many people can relate to at some point of another is reading a news story where they happen to know a lot about the subject of story, either because of profession or proximity or whatever. Usually, they figure out the stories on that specific topic aren't very good. Ideally, they'd then generalize that experience to suggest most news stories, even the ones on topics they don't know enough about to compare, are just as inaccurate.
In terms of missing information, it's gotten so bad that sometimes stories headline something, then completely fail to actually include any details. Yesterday I happened to see a headline in Google news about how the House was going to have to re-vote on the Tax Bill because Democrats in the Senate had brought up that three provisions in the proposed bill violated Senate rules and forced them to be dropped. Curious about which 3 provisions the Senate parliamentarian had ruled out, I clicked on the first news article to read it.
It wasn't until the 7th article I read on the topic that one of them (NY Times) thought to include any information on the actual provisions being blocked (one of which turned out to be the title of the bill). You really have to go to great lengths to write 10-15 paragraphs in an article on the subject of "3 provisions in Tax Bill blocked by Democrats" without managing to mention what the provisions actually were. Is the reporting just that poor, or is it bias, in the sense of proclaiming the information they thought people would be excited by (Democratic victory on something!) and hiding the information which might tend to discredit the narrative (not much victory!).
Either way, somehow a good framework would need to distinguish the motivation for providing the information, as that seems to override the portions you'd expect to see based on an honest attempt to provide straight information. This isn't just a recent problem, either. It's been a standard issue since the early days of the printing press.
You switch between "chuck" and "chunk" a few times. Neither sounds especially nice, but the former lends itself more to how-the-sausage-gets-made analogies.
It certainly seems generally true that people are less motivated by informational exchange than they purport to be but aren't you biasing the sample by considering only "news". I mean isn't time spent studying an academic discipline, getting various IT certifications and most other professional development activity essentially exactly what you suggest people should be doing: learning information about the world that wasn't just discovered?
Isn't news in some sense just the name for the market for information people want merely because it is current? Or are you suggesting people aren't showing enough interest in history specifically as opposed to news?
This definitional change doesn't solve the fundamental issue that its is the absolute difference and not the ratio that needs to be used as the central element before multiplication ... of this I am highly confident ( :)) and now that I have bought this to your attention, I am sure you know that you can model this out with dice and cards etc to prove it for yourself.
But the big picture here is you deserve considerable credit for putting together a really interesting question and model so lets not get too hung up on mathematical minutiae, the big point I wanted to make was the value of info literature had addressed a fair amount of this and you are aware of that so we are all good.
With that minor caveat out of the way, I think you are asking some very probing questions here ... why do we have limited regard for the value of information anywhere in society? Why are we a society that doesn't really invest in information at all ... basically why are we so foolish?
my high level response is that we get what we deserve and that as humans our lack of real respect and acceptance of the truth/reality causes us to harm ourselves in very material ways. News being merely one of them. If we were more sincere about finding and acting on the truth/reality then we would probably be more happy and well adjusted beings.
When giving formulas, I had in mind applying them where the news was that a claim had *risen* in probability. Once can of course represent them with the complement event whose probability has lowered, but that is a lot less natural.
Yes, there's a disconnect between the official explicit rationale and the actual one.
Sure, but why from an info point of view would it be important to get everyone on the same page?
With respect to recency, it's easier to get everybody on the same page by focusing heavily on the latest. If a news organization doesn't focus on the latest but instead picks a particular topic, but nobody else picks that same topic, that's going to make it a lot harder to bring up that same topic around non-readers around the water cooler, or to get noticed in search engines.
There are far fewer people reading those alternative sources than the the number of people reading the news. Most people have picked up a newspaper or magazine once or twice. Far fewer read textbooks after they've already graduated college.
The genre one might call "Gladwellian nonfiction" runs on the premise of "shared explanation and justification of many related small claims"--such books cover multiple apparently unrelated claims, united by using the same concepts as support.
(Though I suppose the *explicit* justification is that the various topics all provide support for a central claim)
Perhaps I misunderstood you but when I try apply your definitions/methodology to a real world scenario it doesn't seem to add up to me. Take something that many people care about ... like Trump and the probability he will end his term as president.
Please can you help me reconcile the following scenario per your own definitions and methodology?
Claim: Probability of trump ending his term as president.
Surprise – how low a probability a reader would have previously assigned to this claim. For me its 75%.
Then I hear credible "news" that he has been assassinated ...
Confidence – how high a probability a reader is to assign after reading this news. This is now 0%. No chance.
Based on your calculation the "amount" of information from that news ... the ratio of confidence/surprise which is 0%/75% ... = 0 as in ZERO
Then you go and multiply 0 with importance and then multiply that by commonality etc and you still get 0.
Which basically means credible news of Trump being assassinated, has a value of ... 0?
What am I missing Professor?
The ratio is the likelihood ratio, which is what you can combine via multiplication across multiple news items to give total news. You can't do that with the difference.
I'm well aware of the value of info literature.
Robin ... this is an interesting perspective and your central point is very valid There is a lot else missing/wrong from a purely correct theory perspective here but I guess the major point you are trying to make is that news sources provide low information value.
For example:
"Okay, the amount of info that some news gives a reader on a claim is the ratio of its confidence to its surprise"
This is not correct, the information value its the difference between these two ... and information value should not be confused with economic value. (how much its worth in dollars and sense to have this information) And if you are interested I can model this out for you very simply.
In fact there is a large body of work extending basic Shannon information theory to more human questions very much like you have asked here, what is the "value" of information, but most of it is in this 1966 seminal paper by Ronald Howard:
https://www.scribd.com/docu...
Not in the book, no.
Do you discuss Herman and Chomsky's 'propaganda model'?
An experience many people can relate to at some point of another is reading a news story where they happen to know a lot about the subject of story, either because of profession or proximity or whatever. Usually, they figure out the stories on that specific topic aren't very good. Ideally, they'd then generalize that experience to suggest most news stories, even the ones on topics they don't know enough about to compare, are just as inaccurate.
In terms of missing information, it's gotten so bad that sometimes stories headline something, then completely fail to actually include any details. Yesterday I happened to see a headline in Google news about how the House was going to have to re-vote on the Tax Bill because Democrats in the Senate had brought up that three provisions in the proposed bill violated Senate rules and forced them to be dropped. Curious about which 3 provisions the Senate parliamentarian had ruled out, I clicked on the first news article to read it.
It wasn't until the 7th article I read on the topic that one of them (NY Times) thought to include any information on the actual provisions being blocked (one of which turned out to be the title of the bill). You really have to go to great lengths to write 10-15 paragraphs in an article on the subject of "3 provisions in Tax Bill blocked by Democrats" without managing to mention what the provisions actually were. Is the reporting just that poor, or is it bias, in the sense of proclaiming the information they thought people would be excited by (Democratic victory on something!) and hiding the information which might tend to discredit the narrative (not much victory!).
Either way, somehow a good framework would need to distinguish the motivation for providing the information, as that seems to override the portions you'd expect to see based on an honest attempt to provide straight information. This isn't just a recent problem, either. It's been a standard issue since the early days of the printing press.
Oops; thanks, fixed.
You switch between "chuck" and "chunk" a few times. Neither sounds especially nice, but the former lends itself more to how-the-sausage-gets-made analogies.
It certainly seems generally true that people are less motivated by informational exchange than they purport to be but aren't you biasing the sample by considering only "news". I mean isn't time spent studying an academic discipline, getting various IT certifications and most other professional development activity essentially exactly what you suggest people should be doing: learning information about the world that wasn't just discovered?
Isn't news in some sense just the name for the market for information people want merely because it is current? Or are you suggesting people aren't showing enough interest in history specifically as opposed to news?