Thanks Professor, for this 'substack'-informing post! While, I desire to be "informed" so as to expand my puny little head, I'm easily overwhelmed by a non-stop signal-and-noise broadcast. With you the pointman filtering the nosie, I'm a happy subscriber!
On the moderation issues, and the "oh, this is a Nazi bar now" problem -- Yishan Wong attempted to describe this problem in a long Twitter thread ( https://twitter.com/yishan/status/1514938507407421440 ) back when Elon first announced his bid for Twitter. A number of my more libertarian friends just steadfastly refused to believe this problem exists, but do any of them want to go hang out on Parler, or Truth Social, with the nutters? No, they want to exercise their freeze peach on Twitter or Facebook, where there's an audience, which is only there because it's _not_ overwhelmed by Nazis and insane 8chan trolls.
Substack Notes is the single worst idea these people have come up with. The consensus among my reader-commenters about it is that we came to Substack BECAUSE it isn't Twitdom, FleeceBlock, or any of the rest of the Bullshit Social Meeeeeeedia waste-of-time internet.
If these guys continue to let Notes become what it is already becoming (it took less than 72 hours for the first "non-positive" post to appear), it clearly demonstrates that they're just another pair of fucking morons, a category Silly Con Valley is overflowing with.
Zeynep Tufekci: The policy question is not well posed. It should be: what was the correct basis for recommending wearing of masks in different circumstances at different times.
It seems to me that modeling could have provided the basics. Subjects would be tested for how much virus was emitted and spread to different distances under different ventilation conditions with and without masks worn correctly or not. This when combined with a measurement of prevalence at a specific place and time would give good estimate of the benefits of mask wearing if the recommendation was acted on by differing proportions of people within a population of different levels of vulnerability of infection based on age, vaccine status etc.
This could have led to nuanced information on benefits, changing with circumstance, which would have permitted individuals and policy makers to combine with estimates of cost of alternative measure to arrive at cost effective measures to carry out.
Hi, new subscriber--found you via Notes for better or worse.
Anyway, very minor question, in Briefly Noted item #10 Langchain do you have a link? The other items do, so curious if there is a link. I have been playing with it and I am still working to get an understanding of the sixth step. I have the MIT video up to watch too--though I suspect it will be more theoretical than applied.
The incompetent-management view.… The incompetent-customer view…. Incompetent regulators…. Ill-advised macroeconomic policies…. The result is a bit like the conclusion of Agatha Christie’s Murder on the Orient Express: everyone did it
No, macroeconomic policies were not (or only at the margin) “ill advised.” Higher than target inflation was an appropriate target (the “flexibility” in FAIT, flexible average inflation targeting) in response to recovery from COVID and the huge shift from services to goods (and now back), concomitant supply chain constraints, and the Putin-petroleum price shock. But at some point the Fed needed to start bringing inflation back down toward target and that was going to mean higher interest rates. THAT was not “ill advised.” Now one can argue that the Fed should have started “fighting inflation” sooner (which might have meant a slower run up of interest rates, maybe not having to haver raised them as high as it has) and that would have made bank managements’, customers’ and regulators’ lives easier. But the delay is not “mismanagement” of the same sort as that of the managers and regulators.
Slashdot commentary has turned more toward classic hard right wing thought over the 10 years (from a bias toward naive techno-libertarianism) so it was surprising to me to find a fairly honest discussion there over the last two days about Twitter's engagement algorithm. In which a number of commenters openly admitted what has become increasingly clear as Melon Husk has stripped away layers of moderation from Twitter and published some (but not all) of Twitter's ranking algorithms: that Twitter has been deliberately boosting obnoxious and in some cases purely evil people/personas/accounts for a long time now - certainly predating Husk - maybe 5 years or more. The reason it is doing this is not so much to drive additional views and re-publishes from the followers of the obnoxious people, who probably were doing so anyway, but to stir up outrage and dunking-fests among otherwise reality-based members leading to exponentially higher clicks among the larger percentage of the user base and, via links, even from viewers who aren't Twitter members. I have to question whether all social media entities, including Substack, are doing the same thing.
Apropos of nothing and everything - everything being the ability of generative AI to make all human production appear as a flyspeck on the ever expanding supernova of generated heat and maybe light. Consider giving the narcissists who suggest (oh so humbly) that they may be Gods with a capital G and about to create super beings. That's boorish. Since the universe is amazing I won't call it impossible. I will call it improbable and absolutely not worth looking at twice. Please consider how many writers are saying generative AI is amazing. It's a machine doing what it's programmed to do.
Probably we humans are also. I'll venture that evolution has written thousands or millions of times as much good code as all the AI workers to date. Is Augmented Inference a good tool? I use it a bit when Google picks phrases to complete my sentences. If it discovers cancers, great. But when it makes the world's writings fake as opposed to just combative, I've had more than enough.
If you're talking about AI at your next social get together, just stop it. It's not intelligent. It's bad enough that Microsoft will use it to confuse the next generation, at least teach the younguns that it's just a game. Consider "https://en.wikipedia.org/wiki/Why_the_Sea_is_Salt". Is life salty enough yet?
Regarding Josh Barro's comment that "only those paying for the SubStack service infrastructure should be allowed to post or reply", why not go further: if someone who has a Substack newsletter posts a note, why not restrict people from replying to that note unless they've paid for a subscription to that person's newsletter? I subscribe to a number of newsletters without paying for them, and I have no expectation of being able to comment on those posts. Why should Notes be an exception to that policy?
Thanks Professor, for this 'substack'-informing post! While, I desire to be "informed" so as to expand my puny little head, I'm easily overwhelmed by a non-stop signal-and-noise broadcast. With you the pointman filtering the nosie, I'm a happy subscriber!
On the moderation issues, and the "oh, this is a Nazi bar now" problem -- Yishan Wong attempted to describe this problem in a long Twitter thread ( https://twitter.com/yishan/status/1514938507407421440 ) back when Elon first announced his bid for Twitter. A number of my more libertarian friends just steadfastly refused to believe this problem exists, but do any of them want to go hang out on Parler, or Truth Social, with the nutters? No, they want to exercise their freeze peach on Twitter or Facebook, where there's an audience, which is only there because it's _not_ overwhelmed by Nazis and insane 8chan trolls.
Substack Notes is the single worst idea these people have come up with. The consensus among my reader-commenters about it is that we came to Substack BECAUSE it isn't Twitdom, FleeceBlock, or any of the rest of the Bullshit Social Meeeeeeedia waste-of-time internet.
If these guys continue to let Notes become what it is already becoming (it took less than 72 hours for the first "non-positive" post to appear), it clearly demonstrates that they're just another pair of fucking morons, a category Silly Con Valley is overflowing with.
Ferrell seem to be addressing the wrong problem:
“For decades, economists have argued that state policy makers lack the requisite knowledge to intervene appropriately in the economy.”
Only extreme Libertarians argue this way.
“ some goals of modern industrial policy are in principle impossible to solve through purely market mechanisms.”
Again, I’m unaware of anyone that thinks they can be.
When a writer starts off building a straw man, it ought to make one suspicious.
“We lack the kinds of expertise that we need to achieve key goals of industrial policy, or to evaluate the tradeoffs between them.”
This could be true, but first we need to identify what those goals are and understand the precise ways that market actors would not achieve them.
Zeynep Tufekci: The policy question is not well posed. It should be: what was the correct basis for recommending wearing of masks in different circumstances at different times.
It seems to me that modeling could have provided the basics. Subjects would be tested for how much virus was emitted and spread to different distances under different ventilation conditions with and without masks worn correctly or not. This when combined with a measurement of prevalence at a specific place and time would give good estimate of the benefits of mask wearing if the recommendation was acted on by differing proportions of people within a population of different levels of vulnerability of infection based on age, vaccine status etc.
This could have led to nuanced information on benefits, changing with circumstance, which would have permitted individuals and policy makers to combine with estimates of cost of alternative measure to arrive at cost effective measures to carry out.
Hi, new subscriber--found you via Notes for better or worse.
Anyway, very minor question, in Briefly Noted item #10 Langchain do you have a link? The other items do, so curious if there is a link. I have been playing with it and I am still working to get an understanding of the sixth step. I have the MIT video up to watch too--though I suspect it will be more theoretical than applied.
Looking forward to my time here!
Eichengreen: A Bank Murder Mystery (fa-mag.com)
The incompetent-management view.… The incompetent-customer view…. Incompetent regulators…. Ill-advised macroeconomic policies…. The result is a bit like the conclusion of Agatha Christie’s Murder on the Orient Express: everyone did it
No, macroeconomic policies were not (or only at the margin) “ill advised.” Higher than target inflation was an appropriate target (the “flexibility” in FAIT, flexible average inflation targeting) in response to recovery from COVID and the huge shift from services to goods (and now back), concomitant supply chain constraints, and the Putin-petroleum price shock. But at some point the Fed needed to start bringing inflation back down toward target and that was going to mean higher interest rates. THAT was not “ill advised.” Now one can argue that the Fed should have started “fighting inflation” sooner (which might have meant a slower run up of interest rates, maybe not having to haver raised them as high as it has) and that would have made bank managements’, customers’ and regulators’ lives easier. But the delay is not “mismanagement” of the same sort as that of the managers and regulators.
Slashdot commentary has turned more toward classic hard right wing thought over the 10 years (from a bias toward naive techno-libertarianism) so it was surprising to me to find a fairly honest discussion there over the last two days about Twitter's engagement algorithm. In which a number of commenters openly admitted what has become increasingly clear as Melon Husk has stripped away layers of moderation from Twitter and published some (but not all) of Twitter's ranking algorithms: that Twitter has been deliberately boosting obnoxious and in some cases purely evil people/personas/accounts for a long time now - certainly predating Husk - maybe 5 years or more. The reason it is doing this is not so much to drive additional views and re-publishes from the followers of the obnoxious people, who probably were doing so anyway, but to stir up outrage and dunking-fests among otherwise reality-based members leading to exponentially higher clicks among the larger percentage of the user base and, via links, even from viewers who aren't Twitter members. I have to question whether all social media entities, including Substack, are doing the same thing.
Apropos of nothing and everything - everything being the ability of generative AI to make all human production appear as a flyspeck on the ever expanding supernova of generated heat and maybe light. Consider giving the narcissists who suggest (oh so humbly) that they may be Gods with a capital G and about to create super beings. That's boorish. Since the universe is amazing I won't call it impossible. I will call it improbable and absolutely not worth looking at twice. Please consider how many writers are saying generative AI is amazing. It's a machine doing what it's programmed to do.
Probably we humans are also. I'll venture that evolution has written thousands or millions of times as much good code as all the AI workers to date. Is Augmented Inference a good tool? I use it a bit when Google picks phrases to complete my sentences. If it discovers cancers, great. But when it makes the world's writings fake as opposed to just combative, I've had more than enough.
If you're talking about AI at your next social get together, just stop it. It's not intelligent. It's bad enough that Microsoft will use it to confuse the next generation, at least teach the younguns that it's just a game. Consider "https://en.wikipedia.org/wiki/Why_the_Sea_is_Salt". Is life salty enough yet?
Regarding Josh Barro's comment that "only those paying for the SubStack service infrastructure should be allowed to post or reply", why not go further: if someone who has a Substack newsletter posts a note, why not restrict people from replying to that note unless they've paid for a subscription to that person's newsletter? I subscribe to a number of newsletters without paying for them, and I have no expectation of being able to comment on those posts. Why should Notes be an exception to that policy?