|
Shimrra Jamaane posted:Isn’t the model fundamentally flawed because it outright says its projecting based on the assumption of a nationwide adherence to strict social distancing measures which is bullshit outside of some states? People so seem to be mostly complying all over the country, including the Republicans. How much the morons everywhere (and not just South and Southie) make a difference is indeed a good question, though.
|
# ? Apr 6, 2020 14:46 |
|
|
# ? May 30, 2024 16:57 |
|
Shimrra Jamaane posted:Isn’t the model fundamentally flawed because it outright says its projecting based on the assumption of a nationwide adherence to strict social distancing measures which is bullshit outside of some states? They try to factor in the measures each state is taking then sum the result to get national numbers, as far as I can tell. Owlofcreamcheese posted:it's not a simulation that takes "fed" data. It's a math model with formulas. Models are systems of equations you input (or feed) data into.
|
# ? Apr 6, 2020 14:48 |
|
Owlofcreamcheese posted:Like, if the number of cats in a city doubles every 5 years and that is a good model, you can graph out and pretty accurately say how many cats there is in 20 years. But you can't say exactly how many cats there is march 20th 2024, and the number is going to be wrong basically every single day. The graph saying 140 and the real answer being 160 is not shocking. They aren't going to go back and wiggle the graph for you. they will update the factors that make the graph with new data and it will make a new graph, but it's not going to pass through each data point exactly, even past ones. How did you come to the conclusion that the number of cats in a city doubles every five years? Let’s say the human society takes a census. What if, instead of working out the doubling period on paper and hardcoding it into the formula, you instead let the computer work out the doubling period? Now imagine if you let the computer continually look at new data as it becomes available, and make its own refinements of the doubling period. Would it not be important for the computer to get the right figures for the cat census? If it gets the wrong figures, isn’t it going to get the doubling period wrong as well? quote:On two occasions I have been asked, — "Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?" In one case a member of the Upper, and in the other a member of the Lower, House put this question. I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question.
|
# ? Apr 6, 2020 14:55 |
|
Looks like they wrote a long effort-post explaining the updates they've made: http://www.healthdata.org/covid/updates
|
# ? Apr 6, 2020 14:57 |
|
I think it's foolish to compare NYC's lockdown to Italy's lockdown. Italy locked down the entire country. On March 22, Lombardy banned outdoor physical activity entirely. Italians need to fill out a form to leave their house. Only one member of the house can make one trip outside per day. Meanwhile, traffic is still flowing in and out of NYC and people are crowding the parks and public transit. Subways are packed rear end to elbow. Delivery drivers are zooming around the city bringing takeout (lmao america) with a possible side order of COVID to thousands of households a day. OddObserver posted:People so seem to be mostly complying all over the country, including the Republicans. How much the morons everywhere (and not just South and Southie) make a difference is indeed a good question, though. As of this weekend, there are still nine states without Stay At Home orders. Religious gatherings are permitted in several states. Pretty much every state lets people go outside as many times as they want per day. Remember that funny video with all the Italian mayors yelling at people for going outside all the time, walking their dog and running? That's the status quo in America. America is painfully relearning what China and Italy already know: a half-assed lockdown isn't enough.
|
# ? Apr 6, 2020 14:58 |
|
Platystemon posted:
Okay? what if you did do that? And the computer spit out that it would double every 4.876 days. That might be a shockingly powerful prediction, but it also would still make a line that did not pass through each and every daily data point, past or future.
|
# ? Apr 6, 2020 15:02 |
|
There is going to be a huge explosion of new cases above projections aprox 10 days after Easter Sunday.
|
# ? Apr 6, 2020 15:02 |
|
Shimrra Jamaane posted:There is going to be a huge explosion of new cases above projections aprox 10 days after Easter Sunday. Churches that are closed now will still be closed then. I'm not aware of any churches scheduling Easter services despite the quarantine.
|
# ? Apr 6, 2020 15:08 |
|
Owlofcreamcheese posted:Okay? what if you did do that? And the computer spit out that it would double every 4.876 days. That might be a shockingly powerful prediction, but it also would still make a line that did not pass through each and every daily data point, past or future. We’re not concerned about the line not passing through all the points. We are concerned that the computer spit out “4.876” when it should have spit out “5.420” because the numbers we gave it to work out are wrong. Now it’s liable to miss all of the points because there is systematic error in the formula. It’s not a good line with a cloud of messy real data surrounding it. It’s a bad line that veers away from the point cloud entirely.
|
# ? Apr 6, 2020 15:08 |
|
Owlofcreamcheese posted:Okay? what if you did do that? And the computer spit out that it would double every 4.876 days. That might be a shockingly powerful prediction, but it also would still make a line that did not pass through each and every daily data point, past or future. okay but if you calculated the doubling period based on wrong data, why would you expect the correct doubling period to come out and if the doubling period is incorrect, the prediction gets more inaccurate over time, not less
|
# ? Apr 6, 2020 15:08 |
|
Deteriorata posted:Churches that are closed now will still be closed then. I'm not aware of any churches scheduling Easter services despite the quarantine. Florida is one state that specifically is allowing churches to be open as an exception to their shutdown.
|
# ? Apr 6, 2020 15:09 |
|
https://twitter.com/sfchronicle/status/1246959336859086849 Yikes on this - not sure of the reliability of serological testing at this point and what it means. Are we even close to being sure that this means you can't get further infections? Is there a risk of second infections that are asymptomatic? Thinking of the old coronavirus test in the 90s where test subjects did show reinfection on later coronavirus tests but without symptoms, not sure what that does to the infectiousness, though. But outside of that, it's going to create a two-tier workforce and I can imagine the effect on the labor market for this as companies will have preferences for COVID-19 survivors. Some people will gain advantages in the market, being uninfected isn't a protected class, certain groups will be discriminated against in new ways, and people without jobs are going to see some perverse incentives. To quote Twerk, "poz my unemployable rear end"
|
# ? Apr 6, 2020 15:10 |
|
As the saying goes, all models are wrong, but some models are useful. This IMHE model has so far not been useful as anything other than a source of optimism (which can prove to be extremely counter-productive in this particular context).
|
# ? Apr 6, 2020 15:10 |
|
enraged_camel posted:https://twitter.com/ScottGottliebMD/status/1247139249016713216 Maaaaybe? I mean, are NY's and Italy's lockdowns similar in nature? Like you said, it'll depend on the numbers, but I somehow have a feeling NY's going to have deaths increase for at least a little bit more. But then that's just a feeling based on what I'm hearing from the state about how the virus is spreading rather than hard numbers. As for Cuomo and Trump: Probably, but it'll more be in terms of Trump shouting at Cuomo and the latter paying lipservice to the guy, as per their interactions. Cuomo hates his guts, but he also knows Trump is a narcissistic blowhard that's only sending out ventilators etc to states who kiss his rear end or he needs to win in the election.
|
# ? Apr 6, 2020 15:11 |
|
friendly 2 da void posted:I think it's foolish to compare NYC's lockdown to Italy's lockdown. Keep in mind that religious gatherings will almost certainly involve singing hymns, which is a super effective means of spreading the virus.
|
# ? Apr 6, 2020 15:12 |
|
I don’t know how useful it is to look at trends in Italy and Spain to try and predict how the US will go because of how many things they’ve done right and conversely how many things we have and continue to do wrong.
|
# ? Apr 6, 2020 15:15 |
|
friendly 2 da void posted:I think it's foolish to compare NYC's lockdown to Italy's lockdown. Yeah, this lies at the heart of the problem with IMHE's model. They say this: quote:--At the time of our first release on March 26, the only location where the number of daily deaths had already peaked was Wuhan City. These data from Wuhan formed the basis of our estimation of the time from implementation of social distancing policies to the peak day of deaths. Since then, an additional seven locations in Italy and Spain with large coronavirus epidemics appear to have reached the peak number of daily deaths (see below): two in Spain (Castile-La Mancha and Madrid), and then five in Italy (Emilia-Romagna, Liguria, Lombardy, Piedmont, and Tuscany). I mean this is just laughably bad.
|
# ? Apr 6, 2020 15:20 |
|
Shimrra Jamaane posted:I don’t know how useful it is to look at trends in Italy and Spain to try and predict how the US will go because of how many things they’ve done right and conversely how many things we have and continue to do wrong. In the US a lot of people still have to go to work every day because nobody’s going to guarantee you income
|
# ? Apr 6, 2020 15:20 |
|
https://twitter.com/MarkLevineNYC/status/1247156159896748032
|
# ? Apr 6, 2020 15:24 |
|
United States of America, year of our Lord 2020, burying our dead... in parks.
|
# ? Apr 6, 2020 15:29 |
|
This is horrifying.
|
# ? Apr 6, 2020 15:29 |
|
Owlofcreamcheese posted:Okay? what if you did do that? And the computer spit out that it would double every 4.876 days. That might be a shockingly powerful prediction, but it also would still make a line that did not pass through each and every daily data point, past or future. You are misunderstanding the most basic thing here. This model is pretty much starting new on this day. The first entry should not be a prediction, because this is a known plot point. Take any other model, like the weather. We are quite good at calculating the weather for short periods and long time trends. (mid-period predictions are pretty much impossible) Now if we want to make a model of the temperature in a country in 20 years we have reasonable ways to predict that. But you have to give an input value of the current system, say the average temperatures of the last year. Is it not obvious that the result would be vastly different if your input would be incorrect? Or if you want to model pandemics. If you are assuming that the amount of infected people doubles every two weeks it makes a huge difference if the current amount is 1000, or 1100.
|
# ? Apr 6, 2020 15:31 |
|
Mass graves in NYC, Jesus. When do the riots start?
|
# ? Apr 6, 2020 15:31 |
|
FlamingLiberal posted:I agree, those Spain/Italy numbers have people in total lockdown In NYC the subways are still packed standing room only with daily commuters. It’s bad.
|
# ? Apr 6, 2020 15:32 |
|
Looking at that Twitter model thread I don't think I'm going to be able to take years of virus deniers.
|
# ? Apr 6, 2020 15:33 |
|
keep punching joe posted:Mass graves in NYC, Jesus. when can we riot while observing social distancing?
|
# ? Apr 6, 2020 15:34 |
|
enraged_camel posted:As the saying goes, all models are wrong, but some models are useful. This IMHE model has so far not been useful as anything other than a source of optimism (which can prove to be extremely counter-productive in this particular context). My understanding is that this is the whole drat point of the model but the media and policymakers (and you all) are misusing it. In a situation where policy changes rapidly AND we have a lot of random fluctuations and unknown, models aren’t going to be magic oracles predicting the future. If they’re parametric like the IMHE model, they’re not going to be magic oracles predicting the past. All they can do is say “given our understanding of this disease and this model that’s simple enough to fit robustly, IF we as a society and policy-makers act in a particular fashion then here’s a range of plausible results.” Even beyond the uncertainty in the underlying models and numbers there is no THE model, there’s just models conditioned on different responses. Having a model the says “here’s what we could expect if we do everything right from here on out” is incredibly useful if it’s used properly - ie along other models as a tool to drive policy and show the public why we should follow the advice and instructions of our policy-makers and public health specialists. That said, the website’s presentation could do a lot more to encourage it to be used properly by our dumbass president and the media - feature giant disclaimers about proper interpretation, add multiple models to the same page, only provide national-level results if those are the intended goals of the model and city-level models are too crude to be useful, etc.
|
# ? Apr 6, 2020 15:34 |
|
enraged_camel posted:Yeah, this lies at the heart of the problem with IMHE's model. They say this: Are they seriously arguing that a shutdown in say Florida is equivalent to that of one of the Italian provinces and that a direct parallel can be made between implementation of both shutdowns and when they reach peak deaths?
|
# ? Apr 6, 2020 15:36 |
|
Shimrra Jamaane posted:In NYC the subways are still packed standing room only with daily commuters. It’s bad. every day i see more and more reasons to never live within ten miles of a city i bitched about rural living for years and it turns out i'm exactly where i want to be for this
|
# ? Apr 6, 2020 15:36 |
|
Any model right now are probably qualified guesstimates with the level of unknowns we still have. https://www.newscientist.com/articl...ource=editorial
|
# ? Apr 6, 2020 15:39 |
|
Ikonoklast posted:Any model right now are probably qualified guesstimates with the level of unknowns we still have. Given the uncertainties with current numbers based on the limited amount of testing, anyone who expects any models to be particularly accurate is delusional. Current models are useful as a rough approximation for near-term planning. Those plans will need to change with the output of the model in some cases, but for most purposes a rough model is good enough. I don't really get this obsession with the absolute accuracy of the model. It's the best we can do given what we know at the moment, no more and no less.
|
# ? Apr 6, 2020 15:43 |
|
NofrikinfuN posted:every day i see more and more reasons to never live within ten miles of a city Small towns are going to suffer especially terribly. Their time just hasn't arrived yet.
|
# ? Apr 6, 2020 15:45 |
|
Deteriorata posted:Churches that are closed now will still be closed then. I'm not aware of any churches scheduling Easter services despite the quarantine. https://www.cnn.com/2020/04/02/us/stay-at-home-order-religious-exemptions-states-coronavirus/index.html are the states to watch
|
# ? Apr 6, 2020 15:46 |
|
NofrikinfuN posted:every day i see more and more reasons to never live within ten miles of a city Rural areas have hardly any hospitals. Good luck
|
# ? Apr 6, 2020 15:47 |
|
I just find it uncomfortable that so much weight is being put on a model that spits out a number based around assumptions of “if you do basically everything perfectly exactly when they need to happen then”.
|
# ? Apr 6, 2020 15:47 |
|
Deteriorata posted:I don't really get this obsession with the absolute accuracy of the model. It's the best we can do given what we know at the moment, no more and no less. Citation needed. That is the point people are arguing.
|
# ? Apr 6, 2020 15:47 |
|
Shimrra Jamaane posted:I just find it uncomfortable that so much weight is being put on a model that spits out a number based around assumptions of if you do basically everything perfectly exactly when they need to happen then. that is not what that site says. it's "if you do basically everything perfectly STARTING RIGHT NOW" not at an ideal point in the past. it's "the best time to plant a tree was 10 years ago, the next best time is today." it's still crazy optimistic for many reasons which have been said in this thread but the reason it got so much rosier is that italy & spain show a PROPER lockdown can work faster than it did in wuhan, so if we start today things actually might look better than we'd thought.
|
# ? Apr 6, 2020 15:50 |
|
Shimrra Jamaane posted:I just find it uncomfortable that so much weight is being put on a model that spits out a number based around assumptions of “if you do basically everything perfectly exactly when they need to happen then”. This is how heterodox economists do it and they're never wrong. lol. But really, if I have any beef it's with the assumptions being made. lovely data is a fact of life and if you hold your breath for perfect data you'll never actually produce a model. Sometimes "close" is acceptable.
|
# ? Apr 6, 2020 15:51 |
|
charity rereg posted:that is not what that site says. it's "if you do basically everything perfectly STARTING RIGHT NOW" not at an ideal point in the past. it's "the best time to plant a tree was 10 years ago, the next best time is today" That’s only slightly less unrealistic though. Because we continue to do nothing perfectly even in the states where we’re honestly trying. Edit: ok saw your new edit. Yeah that makes sense but I wish the website then made it far more clear that these numbers aren’t based of of how things currently are but how things should be starting right now. Shimrra Jamaane fucked around with this message at 15:55 on Apr 6, 2020 |
# ? Apr 6, 2020 15:52 |
|
|
# ? May 30, 2024 16:57 |
|
Shimrra Jamaane posted:That’s only slightly less unrealistic though. Because we continue to do nothing perfectly even in the states where we’re honestly trying. Oh absolutely but in some micro areas I think it's possibly right, like in WA, LA (county, and maybe CA overall) maybe MA that have sort of kind of gotten their poo poo a little together, maybe it won't be quite so rosy but closer to that than doomsday. But it's way wrong about states like TN, FL, LA, etc. I'm sure.
|
# ? Apr 6, 2020 15:54 |