Celeste Kidd - How to Know

classic Classic list List threaded Threaded
13 messages Options
Reply | Threaded
Open this post in threaded view
|

Celeste Kidd - How to Know

Roger Critchlow-2
This talk was mentioned on hacker news this week and inspired my babbling at Saveur this morning.  https://slideslive.com/38921495/how-to-know.  The talk was delivered at Neural IPS on December 9 and discusses recent research on how people come to believe they know something.

This paper https://www.mitpressjournals.org/doi/full/10.1162/opmi_a_00017 describes the Amazon Mechanical Turk experiment on people becoming certain they understood the boolean rule they were being taught by examples.

-- rec --


============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove
Reply | Threaded
Open this post in threaded view
|

Re: Celeste Kidd - How to Know

Steven A Smith

REC -

Good find!

I am not closely following the development and results of GAN work, but it seems like this kind of study explicates at least ONE GOOD REASON for worrying about AI changing the nature of the world as we know it (even if it isn't a precise existential threat).   Convolved with Carl's offering around "weaponizing complexity", it feels more and more believable (recursion unintended) that the wielders of strong AI/ML will have the upper hand in any tactical and possibly strategic domain (warfare, public opinion, markets, etc.).   

I don't know how deeply technical the presumed election-manipulation of 2016 (now 2020) is, but it *does* seem like the work you reference here implies that with the information venues/vectors like streaming video (TV, Movies, Clips, attendant advertising) and social media (FB/Insta/Twit...) the understanding and tools are already in place to significantly manipulate public opinion.  Based on my anecdotal experience about people's *certainty*, this article is very on-point.   And this doesn't even reference the technology of "deep fakes".   

- Steve


On 12/27/19 8:21 PM, Roger Critchlow wrote:
This talk was mentioned on hacker news this week and inspired my babbling at Saveur this morning.  https://slideslive.com/38921495/how-to-know.  The talk was delivered at Neural IPS on December 9 and discusses recent research on how people come to believe they know something.

This paper https://www.mitpressjournals.org/doi/full/10.1162/opmi_a_00017 describes the Amazon Mechanical Turk experiment on people becoming certain they understood the boolean rule they were being taught by examples.

-- rec --


============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove
Reply | Threaded
Open this post in threaded view
|

Re: Celeste Kidd - How to Know

Marcus G. Daniels

Steve writes:

< I don't know how deeply technical the presumed election-manipulation of 2016 (now 2020) is, but it *does* seem like the work you reference here implies that with the information venues/vectors like streaming video (TV, Movies, Clips, attendant advertising) and social media (FB/Insta/Twit...) the understanding and tools are already in place to significantly manipulate public opinion.  Based on my anecdotal experience about people's *certainty*, this article is very on-point.   And this doesn't even reference the technology of "deep fakes".   >

What would be some fun deep fakes?   He’s decided the wall was all a big mistake and that now families in the south will be required to give up their living room floor and extra bedrooms for families that cross the border?   The especially vulnerable, like gender-conflicted teens are at the front of the line.    Once a week will be (mandatory) take your visitor to work day, where you train them to do your job.   Another good one might be the announcement of a  white people tax, which will pay reparations to native Americans and black people.     There’s a long list of fun taxes to announce, like 5 dollars per gallon for gasoline for each mpg less than 25mpg.    Oh, and how about tax incentives for mixed race couples in order to diversify the gene pool.   Concurrent with this, all other dependent tax deductions will be eliminated.   In fact, because of global warming, all children of non-mixed races couples will be subject to a $20,000 a year tax.  Churches will no longer be tax exempt. 

Marcus


============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove
Reply | Threaded
Open this post in threaded view
|

Re: Celeste Kidd - How to Know

Steven A Smith

Marcus -

I do like the idea that a good "comedy team" might effectively de-weaponize deep-fakes...   like The Yes Men or SNL perhaps...   there is the risk that such "normalizes" deep fakes, but to the extent that it is already on it's way...  comedy-ifying may be the best (least-worst) alternative?

After Trump's ascendancy showed it's worst true-colors I had any number of "comedy-centric" ideations to generate a crowd-sourced reaction.   The first (dark) one was called "take a Dump for Trump" and involved variations on the old teenager's bad joke of placing a burning bag of dogshit on a rival's porch to be "stomped out".   The second was more campy and involved collecting donations to be put into a fund to pay out "bounties" on pie-throwers.   Think of an Iraq-War style pack of cards with faces/names/bounties on Trump and his inner-mid-outer circles.   Maybe starting with a $1 bounty for anyone who has shaken hands with him, spiraling in toward those working/living in the White House...  

It would surely play havoc for the Secret Service but more entertaining I can just imagine the distortion of security at Mara Lago for him and his.   Would his aides who got booed out of restaurants a few years back instead have gotten a pie in the face (and a bounty big enough to fund future operations paid to the pie-thrower?).   Imagine all the pies being confiscated at the entry to one of his rallys?   With a $1 bounty on any rally attender, you might see a very different style and texture of counter-protest.

While there would surely be civil and legal consequences... there *would* also be a comic-relief and perhaps de-escalation of self-seriousness consequence as well.

I wonder if anyone has a (meta?) model of this kind of "changing the game"?

- Steve

Steve writes:

< I don't know how deeply technical the presumed election-manipulation of 2016 (now 2020) is, but it *does* seem like the work you reference here implies that with the information venues/vectors like streaming video (TV, Movies, Clips, attendant advertising) and social media (FB/Insta/Twit...) the understanding and tools are already in place to significantly manipulate public opinion.  Based on my anecdotal experience about people's *certainty*, this article is very on-point.   And this doesn't even reference the technology of "deep fakes".   >

What would be some fun deep fakes?   He’s decided the wall was all a big mistake and that now families in the south will be required to give up their living room floor and extra bedrooms for families that cross the border?   The especially vulnerable, like gender-conflicted teens are at the front of the line.    Once a week will be (mandatory) take your visitor to work day, where you train them to do your job.   Another good one might be the announcement of a  white people tax, which will pay reparations to native Americans and black people.     There’s a long list of fun taxes to announce, like 5 dollars per gallon for gasoline for each mpg less than 25mpg.    Oh, and how about tax incentives for mixed race couples in order to diversify the gene pool.   Concurrent with this, all other dependent tax deductions will be eliminated.   In fact, because of global warming, all children of non-mixed races couples will be subject to a $20,000 a year tax.  Churches will no longer be tax exempt. 

Marcus


============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove
Reply | Threaded
Open this post in threaded view
|

Re: Celeste Kidd - How to Know

Marcus G. Daniels


Sent from my iPhone

On Dec 28, 2019, at 10:26 AM, Steven A Smith <[hidden email]> wrote:



Marcus -

I do like the idea that a good "comedy team" might effectively de-weaponize deep-fakes...   like The Yes Men or SNL perhaps...   there is the risk that such "normalizes" deep fakes, but to the extent that it is already on it's way...  comedy-ifying may be the best (least-worst) alternative?

After Trump's ascendancy showed it's worst true-colors I had any number of "comedy-centric" ideations to generate a crowd-sourced reaction.   The first (dark) one was called "take a Dump for Trump" and involved variations on the old teenager's bad joke of placing a burning bag of dogshit on a rival's porch to be "stomped out".   The second was more campy and involved collecting donations to be put into a fund to pay out "bounties" on pie-throwers.   Think of an Iraq-War style pack of cards with faces/names/bounties on Trump and his inner-mid-outer circles.   Maybe starting with a $1 bounty for anyone who has shaken hands with him, spiraling in toward those working/living in the White House...  

It would surely play havoc for the Secret Service but more entertaining I can just imagine the distortion of security at Mara Lago for him and his.   Would his aides who got booed out of restaurants a few years back instead have gotten a pie in the face (and a bounty big enough to fund future operations paid to the pie-thrower?).   Imagine all the pies being confiscated at the entry to one of his rallys?   With a $1 bounty on any rally attender, you might see a very different style and texture of counter-protest.

While there would surely be civil and legal consequences... there *would* also be a comic-relief and perhaps de-escalation of self-seriousness consequence as well.

I wonder if anyone has a (meta?) model of this kind of "changing the game"?

- Steve

Steve writes:

< I don't know how deeply technical the presumed election-manipulation of 2016 (now 2020) is, but it *does* seem like the work you reference here implies that with the information venues/vectors like streaming video (TV, Movies, Clips, attendant advertising) and social media (FB/Insta/Twit...) the understanding and tools are already in place to significantly manipulate public opinion.  Based on my anecdotal experience about people's *certainty*, this article is very on-point.   And this doesn't even reference the technology of "deep fakes".   >

What would be some fun deep fakes?   He’s decided the wall was all a big mistake and that now families in the south will be required to give up their living room floor and extra bedrooms for families that cross the border?   The especially vulnerable, like gender-conflicted teens are at the front of the line.    Once a week will be (mandatory) take your visitor to work day, where you train them to do your job.   Another good one might be the announcement of a  white people tax, which will pay reparations to native Americans and black people.     There’s a long list of fun taxes to announce, like 5 dollars per gallon for gasoline for each mpg less than 25mpg.    Oh, and how about tax incentives for mixed race couples in order to diversify the gene pool.   Concurrent with this, all other dependent tax deductions will be eliminated.   In fact, because of global warming, all children of non-mixed races couples will be subject to a $20,000 a year tax.  Churches will no longer be tax exempt. 

Marcus


============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove
============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove
Reply | Threaded
Open this post in threaded view
|

Re: Celeste Kidd - How to Know

Roger Critchlow-2
In reply to this post by Steven A Smith
I thought she was arguing that very mechanisms that google, facebook, twitter, etc. are using right now to engage people's interest online are already engendering and entrenching all sorts of weird beliefs.  6-9 minutes of activated charcoal advocacy videos and you're probably certain that black smoothies are okay, maybe even good for you.  There are no neutral platforms, because the order in which content is presented is never neutral, and it is especially biased if its goal is to keep you clicking.  Whether this allows focused election manipulation seems dubious, but it does allow for thousands of bizarre theories to be injected into the public consciousness at low cost, and some of them even make money.  Hey, some of them, bizarre as they are, might turn out to be correct, not that the platforms have any interest in that aspect, because that wouldn't be neutral.

Andrew Gelman linked this paper, https://core.ac.uk/download/pdf/81961261.pdf, earlier this week, too.  It's about the genesis of the distinction(s) between objective and subjective probabilities in the 19th century.  Several writers started distinguishing objective and subjective probabilities writing in German, French, and English at about the same time.  True to Kidd's scatterplots of concept variability, none of them appear to be making the same distinction even when they claim to be agreeing.  Part of the problem may have been that objective and subjective had just barely adopted the meanings which we more or less use to this day.  Prior to this time:
The objective in this context referred to the objects of thought, and the subjective to objects in themselves [35, A.2.a]. This (to modern ears) inverted sense survived well into the 18th century; witness, for example, the entry for "Objective/objectivus" in the 1728 edition of Chamber's Dictionary: "Hence a thing is said to exist OBJECTIVELY, objectivè, when it exists no otherwise than in being known; or in being an Object of the Mind" [6, 649]. The meanings of the terms had, however, already branched and crisscrossed in the 17th century in both Latin and in various vernaculars, although "objective" still generally modified thoughts rather than external objects. A famous example can be found in the Meditationes (1641) of René Descartes, in which he contrasted the "objective reality" of an idea--whether it represents its cause by perfection and/or content--with its "formal reality"--whether it corresponds to anything external to the mind [15, 40-42; 8, 136-137; 33]
Over the 18th century we -- or at least some of us -- swapped Platonic objects for Empirical objects.  The dictionaries attribute the change to Kant, but the author notes that the new concept was sort of a Cartesian-Kantian-wild-type hybrid, not exactly anything that anyone had exactly proposed.

-- rec --

On Sat, Dec 28, 2019 at 10:23 AM Steven A Smith <[hidden email]> wrote:

REC -

Good find!

I am not closely following the development and results of GAN work, but it seems like this kind of study explicates at least ONE GOOD REASON for worrying about AI changing the nature of the world as we know it (even if it isn't a precise existential threat).   Convolved with Carl's offering around "weaponizing complexity", it feels more and more believable (recursion unintended) that the wielders of strong AI/ML will have the upper hand in any tactical and possibly strategic domain (warfare, public opinion, markets, etc.).   

I don't know how deeply technical the presumed election-manipulation of 2016 (now 2020) is, but it *does* seem like the work you reference here implies that with the information venues/vectors like streaming video (TV, Movies, Clips, attendant advertising) and social media (FB/Insta/Twit...) the understanding and tools are already in place to significantly manipulate public opinion.  Based on my anecdotal experience about people's *certainty*, this article is very on-point.   And this doesn't even reference the technology of "deep fakes".   

- Steve


On 12/27/19 8:21 PM, Roger Critchlow wrote:
This talk was mentioned on hacker news this week and inspired my babbling at Saveur this morning.  https://slideslive.com/38921495/how-to-know.  The talk was delivered at Neural IPS on December 9 and discusses recent research on how people come to believe they know something.

This paper https://www.mitpressjournals.org/doi/full/10.1162/opmi_a_00017 describes the Amazon Mechanical Turk experiment on people becoming certain they understood the boolean rule they were being taught by examples.

-- rec --


============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove
============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove
Reply | Threaded
Open this post in threaded view
|

Re: Celeste Kidd - How to Know

gepr
Ha!  "There's a fun sub-result, which is, if you have a very deviant concept ... if you have a very weirdo concept that other people don't share, you're actually much more likely to be aware that you have a deviant concept."

At least I *know* I'm a deviant.


On 12/29/19 8:43 AM, Roger Critchlow wrote:
> I thought she was arguing that very mechanisms that google, facebook, twitter, etc. are using right now to engage people's interest online are already engendering and entrenching all sorts of weird beliefs.  6-9 minutes of activated charcoal advocacy videos and you're probably certain that black smoothies are okay, maybe even good for you.  There are no neutral platforms, because the order in which content is presented is never neutral, and it is especially biased if its goal is to keep you clicking.  Whether this allows focused election manipulation seems dubious, but it does allow for thousands of bizarre theories to be injected into the public consciousness at low cost, and some of them even make money.  Hey, some of them, bizarre as they are, might turn out to be correct, not that the platforms have any interest in that aspect, because that wouldn't be neutral.
> [...]

>
> On Sat, Dec 28, 2019 at 10:23 AM Steven A Smith <[hidden email] <mailto:[hidden email]>> wrote:
>
>     REC -
>
>     Good find!
>
>     I am not closely following the development and results of GAN work, but it seems like this kind of study explicates at least ONE GOOD REASON for worrying about AI changing the nature of the world as we know it (even if it isn't a precise existential threat).   Convolved with Carl's offering around "weaponizing complexity", it feels more and more believable (recursion unintended) that the wielders of strong AI/ML will have the upper hand in any tactical and possibly strategic domain (warfare, public opinion, markets, etc.).   
> [...]
>
>     On 12/27/19 8:21 PM, Roger Critchlow wrote:
>>     This talk was mentioned on hacker news this week and inspired my babbling at Saveur this morning.  https://slideslive.com/38921495/how-to-know.  The talk was delivered at Neural IPS on December 9 and discusses recent research on how people come to believe they know something.
>>
>>     This paper https://www.mitpressjournals.org/doi/full/10.1162/opmi_a_00017 describes the Amazon Mechanical Turk experiment on people becoming certain they understood the boolean rule they were being taught by examples.

--
☣ uǝlƃ

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove
uǝʃƃ ⊥ glen
Reply | Threaded
Open this post in threaded view
|

Re: Celeste Kidd - How to Know

Roger Critchlow-2
If you're deviant and you know it, clap your hands!

The sub-fact I liked, which might be in the Daxxy paper, is that people are very good at evaluating their certainty with respect to facts about the physical environment, but that same feeling of certainty is all over the place respecting the metaphysical environment.  I guess we've known that for a while.

-- rec --

On Mon, Dec 30, 2019 at 2:45 PM uǝlƃ ☣ <[hidden email]> wrote:
Ha!  "There's a fun sub-result, which is, if you have a very deviant concept ... if you have a very weirdo concept that other people don't share, you're actually much more likely to be aware that you have a deviant concept."

At least I *know* I'm a deviant.


On 12/29/19 8:43 AM, Roger Critchlow wrote:
> I thought she was arguing that very mechanisms that google, facebook, twitter, etc. are using right now to engage people's interest online are already engendering and entrenching all sorts of weird beliefs.  6-9 minutes of activated charcoal advocacy videos and you're probably certain that black smoothies are okay, maybe even good for you.  There are no neutral platforms, because the order in which content is presented is never neutral, and it is especially biased if its goal is to keep you clicking.  Whether this allows focused election manipulation seems dubious, but it does allow for thousands of bizarre theories to be injected into the public consciousness at low cost, and some of them even make money.  Hey, some of them, bizarre as they are, might turn out to be correct, not that the platforms have any interest in that aspect, because that wouldn't be neutral.
> [...]

>
> On Sat, Dec 28, 2019 at 10:23 AM Steven A Smith <[hidden email] <mailto:[hidden email]>> wrote:
>
>     REC -
>
>     Good find!
>
>     I am not closely following the development and results of GAN work, but it seems like this kind of study explicates at least ONE GOOD REASON for worrying about AI changing the nature of the world as we know it (even if it isn't a precise existential threat).   Convolved with Carl's offering around "weaponizing complexity", it feels more and more believable (recursion unintended) that the wielders of strong AI/ML will have the upper hand in any tactical and possibly strategic domain (warfare, public opinion, markets, etc.).   
> [...]
>
>     On 12/27/19 8:21 PM, Roger Critchlow wrote:
>>     This talk was mentioned on hacker news this week and inspired my babbling at Saveur this morning.  https://slideslive.com/38921495/how-to-know.  The talk was delivered at Neural IPS on December 9 and discusses recent research on how people come to believe they know something.
>>
>>     This paper https://www.mitpressjournals.org/doi/full/10.1162/opmi_a_00017 describes the Amazon Mechanical Turk experiment on people becoming certain they understood the boolean rule they were being taught by examples.

--
☣ uǝlƃ

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC
http://friam-comic.blogspot.com/ by Dr. Strangelove

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove
Reply | Threaded
Open this post in threaded view
|

Re: Celeste Kidd - How to Know

gepr
I'm struggling to reconcile something she said from the presentation with what's said in the paper. In the presentation, she said (my probably flawed transcription) "The original vision was: we'd ask about concrete things. And we'd ask about abstract things. And we were expecting to see more agreement for things like cups and bowls and maybe more disagreement for abstract concepts like love and war. Instead what we found was a surprising amount of variance for both abstract and concrete concepts, though people do agree more for some concepts than others. ... even in the same context, people's concepts can vary quite a bit."

And in the paper they say something like "Thus, while our certainty might be a useful guide with regard to perceptual decisions, such as trying to locate a friend yelling for help in the middle of the woods, it may be misleading in higher-level domains, such as deciding whether to see a chiropractor versus a medical doctor."

So, in the talk, the contrast is between concrete and abstract, whereas in the paper, the contrast is between perceptual versus higher-level.

I worry that your contrast (physical vs. metaphysical) might well be orthogonal to both of those other contrasts. Even if by "physical", you intend something like "perceptual", your contrast with metaphysical evokes the abstract (e.g. Platonic forms or whatever). Since I don't really understand what your contrast means, my question is more about her 2:

  1) concrete vs. abstract, and
  2) perceptual vs. higher-level.

In the talk, she says there's similar concept-mismatching variation across (1). In the paper, they say accuracy of certainty is distinct within (2) (more accurate with perceptual concepts). This is either something paradoxical and I'm missing the resolution. *Or* there's a counter intuitive result lurking. According to (1), my certainty about your concept of "cup" should be just as inaccurate as my certainty about your concept of "centroid". But according to (2), the former should be more accurate than the latter. What am I missing?

On 12/30/19 1:53 PM, Roger Critchlow wrote:
> The sub-fact I liked, which might be in the Daxxy paper, is that people are very good at evaluating their certainty with respect to facts about the physical environment, but that same feeling of certainty is all over the place respecting the metaphysical environment.  I guess we've known that for a while.

--
☣ uǝlƃ

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove
uǝʃƃ ⊥ glen
Reply | Threaded
Open this post in threaded view
|

Re: Celeste Kidd - How to Know

Roger Critchlow-2
Physical vs Metaphysical is probably dragging my own deviant concept of Objective vs Subjective into the foreground, as well as my rhetorical style.  

I think that concrete vs. abstract is identifying categories of conceived object, while perceptual vs. higher-level is identifying the amount of reasoning involved in the conception.

The Daxxy experiment was directly manipulating the perceptual vs. higher-level contrast, measuring certainty and performance as the rules varied from the purely perceptual (Daxxy is Red) to higher-level distinctions (Daxxy is (Red and Square and Small) or (Green and Triangular and Large)), and found that certainty was a good guide to performance at perceptual level and increasingly a crap shoot as rules got more convoluted.  

The "Is X more like Y or Z?" experiment looked for variance in conceptions.  They thought they would find less variance in concepts of everyday objects than in concepts of political leadership.  I'd like to see that paper, but it's still in preparation.


I don't think Dr. Kidd is anywhere near resolving all these issues, many more entertaining experiments are coming.

-- rec --

On Tue, Dec 31, 2019 at 9:19 AM uǝlƃ ☣ <[hidden email]> wrote:
I'm struggling to reconcile something she said from the presentation with what's said in the paper. In the presentation, she said (my probably flawed transcription) "The original vision was: we'd ask about concrete things. And we'd ask about abstract things. And we were expecting to see more agreement for things like cups and bowls and maybe more disagreement for abstract concepts like love and war. Instead what we found was a surprising amount of variance for both abstract and concrete concepts, though people do agree more for some concepts than others. ... even in the same context, people's concepts can vary quite a bit."

And in the paper they say something like "Thus, while our certainty might be a useful guide with regard to perceptual decisions, such as trying to locate a friend yelling for help in the middle of the woods, it may be misleading in higher-level domains, such as deciding whether to see a chiropractor versus a medical doctor."

So, in the talk, the contrast is between concrete and abstract, whereas in the paper, the contrast is between perceptual versus higher-level.

I worry that your contrast (physical vs. metaphysical) might well be orthogonal to both of those other contrasts. Even if by "physical", you intend something like "perceptual", your contrast with metaphysical evokes the abstract (e.g. Platonic forms or whatever). Since I don't really understand what your contrast means, my question is more about her 2:

  1) concrete vs. abstract, and
  2) perceptual vs. higher-level.

In the talk, she says there's similar concept-mismatching variation across (1). In the paper, they say accuracy of certainty is distinct within (2) (more accurate with perceptual concepts). This is either something paradoxical and I'm missing the resolution. *Or* there's a counter intuitive result lurking. According to (1), my certainty about your concept of "cup" should be just as inaccurate as my certainty about your concept of "centroid". But according to (2), the former should be more accurate than the latter. What am I missing?

On 12/30/19 1:53 PM, Roger Critchlow wrote:
> The sub-fact I liked, which might be in the Daxxy paper, is that people are very good at evaluating their certainty with respect to facts about the physical environment, but that same feeling of certainty is all over the place respecting the metaphysical environment.  I guess we've known that for a while.

--
☣ uǝlƃ

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC
http://friam-comic.blogspot.com/ by Dr. Strangelove

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove
Reply | Threaded
Open this post in threaded view
|

Re: Celeste Kidd - How to Know

Roger Critchlow-2
In reply to this post by Roger Critchlow-2
Following up Daston's paper on the origins of objective and subjective probability, one of the files that ended up in my Downloads folder was http://www.fitelson.org/probability/ramsey.pdf, a collection of three essays by Frank P. Ramsey on probability.  HackerNews came up with a link to Cheryl Misak's biography of Frank Ramsey this morning, https://hnn.us/article/174250.

Ramsey's first essay commences with these epigrams:

To say of what is that it is not, or of what is not that it is, is false, while to say of what is that it is and of what is not that it is not is true.
 
-- Aristotle.
 
When several hypotheses are presented to our mind which we believe to be mutually exclusive and exhaustive, but about which we know nothing further, we distribute our belief equally among them .... This being admitted as an account of the way in which we actually do distribute our belief in simple cases, the whole of the subsequent theory follows as a deduction of the way in which we must distribute it in complex cases if we would be consistent.
 
-- W. F. Donkits.
 
The object of reasoning is to find out, from the consideration of what we already know, something else which we do not know. Consequently, reasoning is good if it be such as to give a true conclusion from true premises, and not otherwise.
 
-- C. S. Peirce.
 
Truth can never be told so as to be understood, and not be believed.
 
-- W. Blake. 

The epigram by W. F. Donkits in this paper is apparently the only place his name appears on the internet.

What follows to the end of the section is almost entirely based on the writings of C. S. Peirce. [Especially his
"Illustrations of the Logic of Science", Popular Science Monthly, 1877 and 1878, reprinted in Chance Love and Logic
(1923).]

Back to Popular Science again!

-- rec -- 

On Sun, Dec 29, 2019 at 11:43 AM Roger Critchlow <[hidden email]> wrote:
I thought she was arguing that very mechanisms that google, facebook, twitter, etc. are using right now to engage people's interest online are already engendering and entrenching all sorts of weird beliefs.  6-9 minutes of activated charcoal advocacy videos and you're probably certain that black smoothies are okay, maybe even good for you.  There are no neutral platforms, because the order in which content is presented is never neutral, and it is especially biased if its goal is to keep you clicking.  Whether this allows focused election manipulation seems dubious, but it does allow for thousands of bizarre theories to be injected into the public consciousness at low cost, and some of them even make money.  Hey, some of them, bizarre as they are, might turn out to be correct, not that the platforms have any interest in that aspect, because that wouldn't be neutral.

Andrew Gelman linked this paper, https://core.ac.uk/download/pdf/81961261.pdf, earlier this week, too.  It's about the genesis of the distinction(s) between objective and subjective probabilities in the 19th century.  Several writers started distinguishing objective and subjective probabilities writing in German, French, and English at about the same time.  True to Kidd's scatterplots of concept variability, none of them appear to be making the same distinction even when they claim to be agreeing.  Part of the problem may have been that objective and subjective had just barely adopted the meanings which we more or less use to this day.  Prior to this time:
The objective in this context referred to the objects of thought, and the subjective to objects in themselves [35, A.2.a]. This (to modern ears) inverted sense survived well into the 18th century; witness, for example, the entry for "Objective/objectivus" in the 1728 edition of Chamber's Dictionary: "Hence a thing is said to exist OBJECTIVELY, objectivè, when it exists no otherwise than in being known; or in being an Object of the Mind" [6, 649]. The meanings of the terms had, however, already branched and crisscrossed in the 17th century in both Latin and in various vernaculars, although "objective" still generally modified thoughts rather than external objects. A famous example can be found in the Meditationes (1641) of René Descartes, in which he contrasted the "objective reality" of an idea--whether it represents its cause by perfection and/or content--with its "formal reality"--whether it corresponds to anything external to the mind [15, 40-42; 8, 136-137; 33]
Over the 18th century we -- or at least some of us -- swapped Platonic objects for Empirical objects.  The dictionaries attribute the change to Kant, but the author notes that the new concept was sort of a Cartesian-Kantian-wild-type hybrid, not exactly anything that anyone had exactly proposed.

-- rec --

On Sat, Dec 28, 2019 at 10:23 AM Steven A Smith <[hidden email]> wrote:

REC -

Good find!

I am not closely following the development and results of GAN work, but it seems like this kind of study explicates at least ONE GOOD REASON for worrying about AI changing the nature of the world as we know it (even if it isn't a precise existential threat).   Convolved with Carl's offering around "weaponizing complexity", it feels more and more believable (recursion unintended) that the wielders of strong AI/ML will have the upper hand in any tactical and possibly strategic domain (warfare, public opinion, markets, etc.).   

I don't know how deeply technical the presumed election-manipulation of 2016 (now 2020) is, but it *does* seem like the work you reference here implies that with the information venues/vectors like streaming video (TV, Movies, Clips, attendant advertising) and social media (FB/Insta/Twit...) the understanding and tools are already in place to significantly manipulate public opinion.  Based on my anecdotal experience about people's *certainty*, this article is very on-point.   And this doesn't even reference the technology of "deep fakes".   

- Steve


On 12/27/19 8:21 PM, Roger Critchlow wrote:
This talk was mentioned on hacker news this week and inspired my babbling at Saveur this morning.  https://slideslive.com/38921495/how-to-know.  The talk was delivered at Neural IPS on December 9 and discusses recent research on how people come to believe they know something.

This paper https://www.mitpressjournals.org/doi/full/10.1162/opmi_a_00017 describes the Amazon Mechanical Turk experiment on people becoming certain they understood the boolean rule they were being taught by examples.

-- rec --


============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove
============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove
Reply | Threaded
Open this post in threaded view
|

Re: Celeste Kidd - How to Know

Roger Frye-4


On Fri, Feb 14, 2020 at 8:59 AM Roger Critchlow <[hidden email]> wrote:
 
When several hypotheses are presented to our mind which we believe to be mutually exclusive and exhaustive, but about which we know nothing further, we distribute our belief equally among them .... This being admitted as an account of the way in which we actually do distribute our belief in simple cases, the whole of the subsequent theory follows as a deduction of the way in which we must distribute it in complex cases if we would be consistent.
 
-- W. F. Donkits.
 
The epigram by W. F. Donkits in this paper is apparently the only place his name appears on the internet.

The proper attribution is
W. F. Donkin, Prof of Astronomy, Oxford 
May 1851 Article XLVII 
Phil. Mag. S. $. Vol. 1. No.5. May 1851.

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove
Reply | Threaded
Open this post in threaded view
|

Re: Celeste Kidd - How to Know

Marcus G. Daniels
W. F. Donkin wrote:

"When several hypotheses are presented to our mind which we believe to be mutually exclusive and exhaustive, but about which we know nothing further, we distribute our belief equally among them .... This being admitted as an account of the way in which we actually do distribute our belief in simple cases, the whole of the subsequent theory follows as a deduction of the way in which we must distribute it in complex cases if we would be consistent."

In another context, Eric mentioned the concept of branching structures.   In mixed integer branch & cut solvers, the decisions concerning how to repeatedly separate a problem into sub-spaces is one of the most crucial to get right.   There's a significant literature on it.   Some involve lookahead, others use information theoretic techniques, others do aggregation of variables into simpler forms.   Which one works the best, as far as I can tell, is problem dependent.   It is some analogue to No Free Lunch, I suspect.   It is not unreasonable for a solver to compete them, given the compute resources, however the conclusion from that competition should not be that one policy is better than the other.   Also it reminds me of Glens' advocacy of parallax.

Marcus

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove