4 ms·

> Rather than conjecture, why couldn't they just phone up each packer and ask if that's what's happening? Why do we need pontification in this case when we coul

by ke88y 3y ago

> Rather than conjecture, why couldn't they just phone up each packer and ask if that's what's happening? Why do we need pontification in this case when we could just go look? We know meat packers do this. The question is whether this behavior explains price asymmetry. There are reasons why it might not (e.g., other market actors in the beef market, goods substitution, cost of demand smoothing, shelf life, etc.) The details matter, and no sane packer is going to release the models/algorithms/decision criteria they use to determine when and how to release stored inventory. > The language in the paper just makes me think the whole thing should be ignored. E.g. there are sentences like: I hate this. It's researchers accurately reporting their findings. They aren't "pontificating". They are explaining the nuances of what they discovered when they tried to answer this question. These days, you can't just ask a real questions in plain English and then accurately report your findings. If it's not TED Talkified bullshit, you'll get pilloried by the commentariate for being mealy mouthed. Our appetite for nuance is dead, and with it our appetite for truth. Also, given the point you're trying to make, you picked a weird sentence for focus on when this sentence is in the intro: > We do not reject the null hypothesis of symmetric responses to shocks at other points in the distribution chain when models are estimated using scanner retail price data. Asking a question and finding that the answer is "yes, but not as bad as previously believed" is a contribution to our knowledge about the true state of the world.

sdwr 3y ago
> It's researchers accurately reporting their findings In a low-confidence, hedged way. A "good" article has the balls to come in with a real question, and use statistics as tools to answer it. If the answer is messy or nuanced, so be it, but at least try. Leading with the methods used and prior studies etc. etc. is the data equivalent of cowering behind cover.
- crazygringo 3y ago
  A hundred times no. Science and economics isn't about having big swinging balls, and thank god it's not. It's about slow, incremental progress towards understanding. Yes, being low-confidence and hedged at first because that's truthfully accurate. Methods and prior studies place your work into context. The idea that this is "cowering behind cover" is completely incorrect. Science, and the social sciences, aren't sports competitions to win the championship -- and again, thank god they're not. It's not that you're supposed to be swinging for the fences or else you're worthless -- what a counterproductive idea. It's about slowly building up knowledge.
  - yourapostasy 3y ago
    > It's about slowly building up knowledge. In other words, if one doesn't care for the grind hip deep in the nuance for years, decades, even centuries, then don't look for a TikTok'ified revelation in science, and wait for the swing the fences headlines in the mainstream news. Flashes of leaps forward in science are exceedingly rare; we don't say "on the shoulders of giants" for nothing, those giants are the towering innumerable generations that came before us, it is utterly humbling to see how long basic technologies like mastering control of fire took to spread across the world [1]. I see this outsized appetite for the easy button in a lot of programmers as well. There comes a time when one will have to grind through small shavings of parsimoniously-parceled out pico packets of understanding to progress in one's engineering maturity evolution. Categorically denying that by demanding there is an easy-quick-simple way or the ones doing the grinding are somehow character deficient is to close off one possible way to grow. Our work is already challenging enough, so I'm always open to any way to grow myself, though YMMV and one could even do better closing off that avenue to grow for all I know. Just to perform junior-level programming acceptably well takes a stupendous amount of grinding that most non-programmers cannot fathom doing themselves through interacting with software, though many will happily grind just as much if not more at other pursuits. I cannot for example, fathom watching so much sports that I can quote on-demand specific games and winning shots/goals on specific days by specific players. And yet, there is an NBA star who is a walking encyclopedia of every game they've ever played in their NBA career, who not only knows winning shots but from where on the court, who was covering them, the relative position of nearby players, who they were, etc. The amount of grind to get into the game so to speak are penny-ante table stakes compared to the grind the pros put in. There are more such personal eras of grinding to accomplish successive levels of mastery in these pursuits, and programming is no exception in my personal experience. This isn't some personal character issue, it is a cognitive bias (or two or more depending upon how you look at it) causing us to overestimate what we can accomplish in the short term and underestimate what we can accomplish in the long term. Projection Bias [2] and Planning Fallacy [3] are my favorites to explain this effect. [1] [[https://en.wikipedia.org/wiki/Control_of_fire_by_early_humans][Control https://en.wikipedia.org/wiki/Control_of_fire_by_early_human... of fire by early humans - Wikipedia]] [2] https://en.wikipedia.org/wiki/Projection_bias https://en.wikipedia.org/wiki/Projection_bias [3] https://en.wikipedia.org/wiki/Planning_fallacy https://en.wikipedia.org/wiki/Planning_fallacy
- davidcuddeback 3y ago
  Wow, no. Just no. That's not how research is done. What you're describing sounds more like writing a clickbait article than writing a research paper. Writing a research paper in that manner would damage the researcher's credibility. A good research paper does lead with a discussion of related research and methodology and is careful to not accidentally make unfounded conclusions. In fact, some papers will include a "threats to validity" section, which I think is a good exercise in skepticism.
CraigJPerry 3y ago
>> Asking a question and finding that the answer is "yes, but not as bad as previously believed" is a contribution to our knowledge Wait, that is not an accurate characterisation of the paper. "maybe yes, but maybe no" is more accurate: Maybe yes: - BLS data supports the proposition, but possible BLS data flaws are highlighted. Maybe no: - using other data instead of BLS allows a different interpretation - perhaps asymmetric in favour of feedlot owners ("this effect is negligible, or cannot be captured by our impulse response-based test") - perhaps symmetric ("Our preferred interpretation is simpler, with the relationship being symmetric"). >> It's researchers accurately reporting their findings. They aren't "pontificating". They are explaining the nuances of what they discovered when they tried to answer this question. I'm not convinced that's what's going on here. Could it be that someone(s) has decided the existing literature weights too heavily in favour of a view they don't like. If the existing literature is broadly correct, one thing you could do is publish another paper which broadly agrees but leaves room for some doubt. If it broadly agrees, could it be less likely to come under robust scruitiny - that could lead to outright public rejection by others. Introducing a small amount of doubt with the broad agreement could have 2 effects: 1. there now exists some doubt when summarising multiple papers on the topic, a contrarian view has now been injected into the discourse. The contrarian view is not backed up by hard facts, it's reasonable conjecture but nonetheless, when summarising the available papers on a topic, its views should be included in that summarisation, thus unfairly skewing the summarisation view. A view from conjecture is summarised next to views more directly evidenced in underlying data. 2. you can count on someone misrepresenting your findings. In our conversation you said "yes, but not as bad as previously believed", something the paper doesn't actually give hard evidence for but does try to suggest. To be clear, i don't claim that this is what's going on here, but going back to my original comment, the lack of weight of evidence for these findings does lead me to dismiss the paper. >> our appetite for truth I can only speak for myself here but my comment - misguided or otherwise - was in search of truth.

by ke88y 3y ago

sdwr 3y ago

> It's researchers accurately reporting their findings In a low-confidence, hedged way. A "good" article has the balls to come in with a real question, and use statistics as tools to answer it. If the answer is messy or nuanced, so be it, but at least try. Leading with the methods used and prior studies etc. etc. is the data equivalent of cowering behind cover.

crazygringo 3y ago
A hundred times no. Science and economics isn't about having big swinging balls, and thank god it's not. It's about slow, incremental progress towards understanding. Yes, being low-confidence and hedged at first because that's truthfully accurate. Methods and prior studies place your work into context. The idea that this is "cowering behind cover" is completely incorrect. Science, and the social sciences, aren't sports competitions to win the championship -- and again, thank god they're not. It's not that you're supposed to be swinging for the fences or else you're worthless -- what a counterproductive idea. It's about slowly building up knowledge.
- yourapostasy 3y ago
  > It's about slowly building up knowledge. In other words, if one doesn't care for the grind hip deep in the nuance for years, decades, even centuries, then don't look for a TikTok'ified revelation in science, and wait for the swing the fences headlines in the mainstream news. Flashes of leaps forward in science are exceedingly rare; we don't say "on the shoulders of giants" for nothing, those giants are the towering innumerable generations that came before us, it is utterly humbling to see how long basic technologies like mastering control of fire took to spread across the world [1]. I see this outsized appetite for the easy button in a lot of programmers as well. There comes a time when one will have to grind through small shavings of parsimoniously-parceled out pico packets of understanding to progress in one's engineering maturity evolution. Categorically denying that by demanding there is an easy-quick-simple way or the ones doing the grinding are somehow character deficient is to close off one possible way to grow. Our work is already challenging enough, so I'm always open to any way to grow myself, though YMMV and one could even do better closing off that avenue to grow for all I know. Just to perform junior-level programming acceptably well takes a stupendous amount of grinding that most non-programmers cannot fathom doing themselves through interacting with software, though many will happily grind just as much if not more at other pursuits. I cannot for example, fathom watching so much sports that I can quote on-demand specific games and winning shots/goals on specific days by specific players. And yet, there is an NBA star who is a walking encyclopedia of every game they've ever played in their NBA career, who not only knows winning shots but from where on the court, who was covering them, the relative position of nearby players, who they were, etc. The amount of grind to get into the game so to speak are penny-ante table stakes compared to the grind the pros put in. There are more such personal eras of grinding to accomplish successive levels of mastery in these pursuits, and programming is no exception in my personal experience. This isn't some personal character issue, it is a cognitive bias (or two or more depending upon how you look at it) causing us to overestimate what we can accomplish in the short term and underestimate what we can accomplish in the long term. Projection Bias [2] and Planning Fallacy [3] are my favorites to explain this effect. [1] [[https://en.wikipedia.org/wiki/Control_of_fire_by_early_humans][Control https://en.wikipedia.org/wiki/Control_of_fire_by_early_human... of fire by early humans - Wikipedia]] [2] https://en.wikipedia.org/wiki/Projection_bias https://en.wikipedia.org/wiki/Projection_bias [3] https://en.wikipedia.org/wiki/Planning_fallacy https://en.wikipedia.org/wiki/Planning_fallacy
davidcuddeback 3y ago
Wow, no. Just no. That's not how research is done. What you're describing sounds more like writing a clickbait article than writing a research paper. Writing a research paper in that manner would damage the researcher's credibility. A good research paper does lead with a discussion of related research and methodology and is careful to not accidentally make unfounded conclusions. In fact, some papers will include a "threats to validity" section, which I think is a good exercise in skepticism.

CraigJPerry 3y ago

>> Asking a question and finding that the answer is "yes, but not as bad as previously believed" is a contribution to our knowledge Wait, that is not an accurate characterisation of the paper. "maybe yes, but maybe no" is more accurate: Maybe yes: - BLS data supports the proposition, but possible BLS data flaws are highlighted. Maybe no: - using other data instead of BLS allows a different interpretation - perhaps asymmetric in favour of feedlot owners ("this effect is negligible, or cannot be captured by our impulse response-based test") - perhaps symmetric ("Our preferred interpretation is simpler, with the relationship being symmetric"). >> It's researchers accurately reporting their findings. They aren't "pontificating". They are explaining the nuances of what they discovered when they tried to answer this question. I'm not convinced that's what's going on here. Could it be that someone(s) has decided the existing literature weights too heavily in favour of a view they don't like. If the existing literature is broadly correct, one thing you could do is publish another paper which broadly agrees but leaves room for some doubt. If it broadly agrees, could it be less likely to come under robust scruitiny - that could lead to outright public rejection by others. Introducing a small amount of doubt with the broad agreement could have 2 effects: 1. there now exists some doubt when summarising multiple papers on the topic, a contrarian view has now been injected into the discourse. The contrarian view is not backed up by hard facts, it's reasonable conjecture but nonetheless, when summarising the available papers on a topic, its views should be included in that summarisation, thus unfairly skewing the summarisation view. A view from conjecture is summarised next to views more directly evidenced in underlying data. 2. you can count on someone misrepresenting your findings. In our conversation you said "yes, but not as bad as previously believed", something the paper doesn't actually give hard evidence for but does try to suggest. To be clear, i don't claim that this is what's going on here, but going back to my original comment, the lack of weight of evidence for these findings does lead me to dismiss the paper. >> our appetite for truth I can only speak for myself here but my comment - misguided or otherwise - was in search of truth.