[FRIAM] the slow red-pill
David Eric Smith
desmith at santafe.edu
Sun Jul 18 07:04:52 EDT 2021
Thanks for this Glen; very helpful.
I didn’t realize there was a term for impredicative construction, but will be glad to have it. Will save me long typed harangues trying to describe a thing I don’t have a good understanding of. I assume that a lot of work on questions of this sort is done in “reflection” or whatever the technical name is in computer science. When executables contain a characterization of themselves in some interpreter.
Also good to have the Klein interview and the book.
Eric
> On Jul 17, 2021, at 12:01 AM, uǝlƃ ☤>$ <gepropella at gmail.com> wrote:
>
> Re: impredicative construction, building a system with the system being built - Can *we* animals do this whereas a *typical* machine cannot? By "typical", of course, I mean *tools*, things with an a priori passthrough accountability. So, my question is an example of itself, circular, petitio principii. It seems like part of the psychologist's job is to delineate which parts of the machine are impredicative and which parts are analytically reducible ... I guess a bit like Tononi's IIT.
>
> Re: institutional lifting - In the podcast where Ezra Klein interviews this guy:
> https://linkprotect.cudasvc.com/url?a=https%3a%2f%2fbookshop.org%2fbooks%2fwork-a-deep-history-from-the-stone-age-to-the-age-of-robots%2f9780525561750&c=E,1,rsODS_cXowZxQ--tmVu4e6wr2Djv6tqi1zAoVNvAAA2bEIiEGKmJ9nzoxuxOd8Y_9tLub3_POs5rZD1eVn_4MQW_lyI-8k8dxFq3FNKO2IyBEpYGYeuwcTQlzYMw&typo=1
> they talk quite a bit about needs vs. wants. What rang a bell for me was this idea that I have of the meaning of life. Each of us (animals and plants) are here to explore regions of the "space" that no other individual is exploring. It is our purpose to do that. As Marcus points out, some of those regions of the space cannot be reached without some *thing* creating a huge gravity well of convex space. E.g. our Federal government gifts us with things like interstate highways and nationwide laws that make my region of space reachable ... particularly ARPANet.
>
> So your (and Harriot's) path toward The Country's responsibility/accountability can be tied to this lifting. And I think we can do the same with social media companies, maybe by a modification of the logic that gave us the Citizens United ruling. But we can't get there through predicative, analytical legal calculus. Somehow, we have to infuse these artificial persons (like Facebook, incorporated corporations) with a similar self-referencing bootstrapped construction to that we animals have.
>
> I've been calling this (pseudo) domain "artificial morality". But that's as far as I've gotten. 8^D
>
> On 7/15/21 5:28 PM, David Eric Smith wrote:
>> This is a nice thread.
>>
>> Like any good narrow question, it quickly makes itself insoluble because it gets entangled with the whole wider world. In particular, the final paragraph, which I like, requires us to get back into the morass of “why punish”.
>>
>>> On Jul 16, 2021, at 8:21 AM, uǝlƃ ☤>$ <gepropella at gmail.com <mailto:gepropella at gmail.com>> wrote:
>>>
>>> I have a rebuttal to the Matthias argument that I purposefully did not link to, just in case my trawl led to an interesting catch. (Know 10 things, say 9.) And although Frankenstein's monster is a good start, it pales in comparison to the modern questions of explainable/interpretable AI and *ethical* AI. Can we *read through* an algorithm to effectively blame an algorithm's author? Or, if not the author, the post-authorship *user* of the author's product? Or, as a postmodernist might argue, should we take the author's product as a stigmergic naturfact and treat the author and her artifact as *excused* ... they were only being creative ... as with gun manufacturers ... the blame lies with the user of the artifact. Algorithms don't kill people. People kill people!
>>
>> This is one where I am very much _not_ like a professional working psychologist, who might be inclined to say “because I can’t derive its mechanism analytically within a philosophical system that I can articulate, I am inclined to believe there is no there there.” I am strongly on-board with the psychologist, in what I would call the view that the language our society uses to get to punishment choices doesn’t stand up well as an analytic language of concepts and mechanisms. But my response would be to say “it’s all part of the structure of message-passing within the many-actor, many exchange system, arrived at through whatever filtering on the behaviors of those actors and exchanges; it may not be a description _of_ itself as a concept (even though it presents itself as being that), but that doesn’t make it uninformative _about_ whatever its nature really is, if we could find a language to describe it”. (And of course, I recognize that that is also just what the
>> psychologist is after as well.)
>>
>> But to get back to the point: whatever reason we use for punishing people is simply non-sequitur for artifacts that are brought into existence and are of a completely different nature.
>>
>> I did hear one thing that struck me as a nice framing, on a somewhat related topic. Al Franken did an interview with Michael Harriot
>> https://linkprotect.cudasvc.com/url?a=https%3a%2f%2falfranken.com%2flisten%2fmichael-harriot-senior-writer-for-theroot-com-the-nations-largest-black-online-newspaper&c=E,1,d9LjgT6dvddYRlZVPVrIAcn5GeYKYoSJNCRLf4XOge_Xt5dRyR527HqTVNmAP-6b48GEx6ZORLV9Sy0GWeO0OlVDixmokcUJLlNhkcgrHVorZKwiIp0,&typo=1 <https://linkprotect.cudasvc.com/url?a=https%3a%2f%2falfranken.com%2flisten%2fmichael-harriot-senior-writer-for-theroot-com-the-nations-largest-black-online-newspaper&c=E,1,5iMDhx3hHWKLv8J8JsJg-i8dfVgwUUrAOEI7jzk6V-ZWhlQRaHwr5G03AhiN5_9W4aQkerXegVenhoIy_cgIWuA6ulH6XTUhF7k5xPAHXw,,&typo=1>
>> who I guess is a senior writer for The Root.
>>
>> Harriot has a nice take on reparations, which is both practically great for getting around many of the dodges that people would use to try to exempt themselves, and also rather a nice principle from which to reason. He says reparations are owed to black people by “The Country” as a political and civil institution, the same way as national defense, lawmaking, a currency, public health, regulation of truth in advertising or contract, etc. etc. are the remit of The Country. It works very well with the rest of the whole conversation. To the extent that racism is systemic, that is precisely because it can systematically harm people without specific intent by one or another actor. The reason favored groups can hold and accumulate wealth at faster rates than exploited groups, even if the favored groups are not descendants of slaveholders and the disfavored groups not descendants of slaves (both, e.g., being more recent immigrants), is that the complex fabric of law, institution,
>> and norms ensnares them all into certain situations or roles. So reparations are best framed the way one would frame a budgetary responsibility or a development goal or any other national priority.
>>
>> I wonder if there is some variant on that overall approach that could be the basis of law for these social media providers. Clearly the Volokh article can’t be used if it omits the role of recommendation, which although not identical to editing is certainly every bit as active. (The other article handled that from the start, so no problem of omission overall.) But I feel like trying to “reduce” (in the way non-physicists usually mean the term as an expression of disdains for physicists) the legal jeopardy to the builder, to one or another content-provider or end user, etc., looks like it fails to recognize the existence of a de facto layer of institutionally that is emerging out of these technical capabilities.
>>
>> Eric
>
> --
> ☤>$ uǝlƃ
> - .... . -..-. . -. -.. -..-. .. ... -..-. .... . .-. .
> FRIAM Applied Complexity Group listserv
> Zoom Fridays 9:30a-12p Mtn GMT-6 bit.ly/virtualfriam
> un/subscribe https://linkprotect.cudasvc.com/url?a=http%3a%2f%2fredfish.com%2fmailman%2flistinfo%2ffriam_redfish.com&c=E,1,pScN9B67a9aMFGjrBPtvqKMRkJDFKBla_nADfn-ftvhfMBQ_7RTVSR4ubxrc3TNzNTiJYh9T2P0jKPAGL7zf7bt6F46dvNJdLRg55kwsiql2ezIOdA,,&typo=1
> FRIAM-COMIC https://linkprotect.cudasvc.com/url?a=http%3a%2f%2ffriam-comic.blogspot.com%2f&c=E,1,kXTGUd_0lxEXqH8BA2JTqOibHUqh6TY1MpVhcg2NNa8k_P1WzdS_OWiwzuxYz0WVuSZvmE5aR7XJJVXuS3jW93PAP4iTcjPrtiVgxxknAGhJaySw8W2Y16YqCUs,&typo=1
> archives: http://friam.471366.n2.nabble.com/
More information about the Friam
mailing list