That's not what flooding the zone is or does, though.
It doesn't impact the people building models.
Flooding the Zone is - by definition - unreliable polling by polling companies that are unreliable or new or obviously biased.
The polls themselves are the zone flooders, and generally don't have credibility to lose.
Among the 'weird' people I mention in the first post is an obsessive fixation on only using 'favorable' polls to their outcome... that's sort of an obverse of flooding the zone -- and another thing that modelers manage when building their models.
... what exactly would you have a modeler bet eating a bug on in a 50/50 prediction?
FWIW, I *don't* think that this kind of modelling will always yield a 50/50 result; it can and will show a candidate winning in a preponderance of scenarios a'la 2016. I'd be curious to see what a historical forecast of 1980 election would have shown using this kind of model. Would you get 99% bug eating forecast? or 85%? or what?
As long as it keeps to that time table, we should be ok -- we should be somewhere south of Cuba by that time. Of course, they're also saying there's a low probability for a tropical storm in the western Caribbean week after next... The Ocean is a fickle...beast.
Good point; that and the simplification of the National Aggregation of who's 'winning' the polls... we have 50 variables that can confound the National Aggregation.
Heh, anyone can build a model -- as far as I can tell all the modelers are working with published data with various flavors of how they do the Data Science on top of the published data. The Data Science is proprietary with various degrees of skirt lifting to show a bit of ankle here or there.
538 (now that it's got it's shit together(?)) has a fun graphic that I assume(?) is regularly updated... as of today, there's a scenario where Harris wins 519 to 19. So... take heart, Saul and Chip... they're sayin' there's a chance.
I think you're trying to make a meta-critique that the published Polls are not measuring voter sentiment correctly? That could be. That's part of the Data Science as well... which includes historical analysis of previous polls by the same company vs. actual results vs. corrections. But that's why the 'Flooding the Zone' nonsense isn't really an issue for the 'serious' modelers because they aren't using those polls and/or are applying adjustments.
Flooding the Zone is a Media play (to get the horserace people to talk about it) not a thing that influences the 'serious' modelers.
I don't think that that's what the people with 'models' vs. simple aggregators are doing.
Modeling is a Data Science project on top of an aggregation project. People are getting mad about the 'modeling' aspect but they are getting mad about the wrong things.
The models aren't predicting the vote percentage of 49.2%, they are saying that the models running multiple cycles show candidate #1 winning 49.2% of the time and therefore candidate #2 winning 50.8% of the time (technically most models are showing ~.2% chance of tie).
The models have, in fact, been on different sides of the 50% divide both among other models and the same model itself at different points of time. A quick google search showed Harris a month ago was favored 58% down to 52% *chance to win* among the modelers. Now the models are even more variable with some showing *Trump* has 51% to win with others showing *Harris* 52% chance to win. Which is why all of them are basically saying the outcome can't be statistically differentiated from 50/50 (with a very small chance of a tie).
What's dumb, IMO, is 'preferring' a model that favors your candidate *as if* it's giving you a better chance to win -- that model has in fact predicted that the OTHER candidate will win in with it's own 'biases' nearly 50% of the time. Sometimes that very model will 'predict' a blow-out win for the candidate who will ultimately lose.
Sometimes the model will even predict the Trumwill 'inside straight' victory for Harris where she loses PA, but wins 7 other toss-up states. This scenario happens for *all* the models... Therefore, if Trumwill is correct, it will have been predicted even by a model that says Trump had a 51% chance to win.
In the end, one candidate will win and vindicate 50% a certain sort of person's assumptions about the models.
The 'weird' thing about the poll wars, at least to me, is the obvious desire for confirmation that your preferred candidate is going to win.
And the 'weirder' thing about this cycle vs. 2016 is that calling the election a statistical dead heat has driven that sort of poll watcher completely mad.
And the 'weirdest' thing of all is that 50% of those sorts of persons will be 100% validated in their unhinged understanding of what polling models do.
I could only see a preview clip, not the full episode, but Cass is an interesting figure to follow; he's deceptively clever and people underestimate his positions and ability to speak on them.
Ultimately, I don't think he's the 'future' of the Republican Party because he's just one guy (ok I think now 4 or 5) at a think tank... but I do think some future right of center faction will pick up his work. And, by and large, that would be a better correction to the Libertarian / Tech-bro right economics that I think we're going to see come after (and in some form during) Trump.
p.s. the small clip isn't a great clip since it's mostly Stewart saying that Trump doesn't follow Cass's policies -- which, yes, but that's not why you have Cass on your show.
We're doing woods-work this weekend getting one of our older hunting spots opened up with better sight lines and cutting back as many honeysuckle bushes and tree of heaven as we possibly can without a skid steer. Now I want a skid steer again; skid steer concupiscence is never far from the heart.
The pendulum will swing back; I'm seeing an opening for those of us who've read widely in prompt engineering.
One project I'm working on my engineer told the LLM, 'you are a marketing professional' and a bunch of other stuff like, don't lie, don't add text after the name of the product, and so on... to help sell cookies.
Oh what an opportunity missed is all I could think. Imagine using the voice of a 'marketing professional' how droll.
You are Fafhrd describing to the Gray Mouser the finest cookie you've ever had; embellish with comparisons to local fauna, and after naming the cookie officially, give it a nickname that would tickle us with whimsey. Post to instagram.
Yes, I think that if we run the election through simulation models multiple times accounting for margins of error and simulate relational effects, sometimes Harris will win and sometimes Trump will win.
You may be right that there won't be 'ticket splitters' but, I think you're wrong in your reasoning; that is, IF there's a reason to split a ticket its to NOT vote for Robinson and NOT vote for Lake, while still NOT voting for Harris... it's exactly because Robinson and Lake are even more odious than Trump, especially in a LOCAL sense.
It's a weird reverse split, but it wouldn't surprise me.
NH going Red is a bold prediction... I have friends in NH that are pretty plugged in to local politics... and just when you think maybe NH will flip red, they never do.
I gave Harris 35% odds of being nominee, so my being wrong was less wrong.
For simplicity's sake I kept PA as the tipping point... but my alternate path to Blue victory goes through NC and AZ where MAGA nutters could potentially impact the Presidential vote in somewhat unpredictable fashion.
There's only one word to describe Trump winning: Inconceivable.
Senate: R+1 (too many Blue seats to defend)
House: Red if Trump wins, maybe even red if Harris wins depending on *how* she wins; that is, if R's quiet quit Trump and split the ticket, then Red; if Harris pulls votes and builds coattails, then potentially it goes Blue.
At this point, I think an assassination would make Vance president... so probably no more assassination attempts.
Wildcard: The Harris Candidacy is unique in that it is both an incumbent and an outsider; it has a record when it wants it and doesn't when it doesn't. I think they are playing this sort of null vector candidacy pretty well... as long as the rest of the world can sit tight for 30 more days, it might just work.
Now, if you'd like to see me act as an official Presidential Elector, then vote as hard as you can for the Pelicans.
If prelude to ground war; the disruption could be decisive. If the goal is to 'send a message' then not so much.
Assuming a ground war, I'm not sure what it accomplishes without support from Lebanon's Maronites, Sunnis and 'civilian' Shia governing coalition members.
On “Open Mic for the week of 10/21/2024”
I have steel targets on my property; we'll practice with pistols with special frangible rounds at close range (~10 paces).
We'll sight in for hunting at 100 yds.
Would be terrified to shoot an AR-15 at less than 100yds at a steel target.
https://sssfonline.org/wp-content/uploads/2020/03/Steel-Resource-Guide-Ltr.pdf
On “Campaign Scratchpad: Known Unknowns”
That's not what flooding the zone is or does, though.
It doesn't impact the people building models.
Flooding the Zone is - by definition - unreliable polling by polling companies that are unreliable or new or obviously biased.
The polls themselves are the zone flooders, and generally don't have credibility to lose.
Among the 'weird' people I mention in the first post is an obsessive fixation on only using 'favorable' polls to their outcome... that's sort of an obverse of flooding the zone -- and another thing that modelers manage when building their models.
... what exactly would you have a modeler bet eating a bug on in a 50/50 prediction?
FWIW, I *don't* think that this kind of modelling will always yield a 50/50 result; it can and will show a candidate winning in a preponderance of scenarios a'la 2016. I'd be curious to see what a historical forecast of 1980 election would have shown using this kind of model. Would you get 99% bug eating forecast? or 85%? or what?
"
As long as it keeps to that time table, we should be ok -- we should be somewhere south of Cuba by that time. Of course, they're also saying there's a low probability for a tropical storm in the western Caribbean week after next... The Ocean is a fickle...beast.
"
Good point; that and the simplification of the National Aggregation of who's 'winning' the polls... we have 50 variables that can confound the National Aggregation.
"
Heh, anyone can build a model -- as far as I can tell all the modelers are working with published data with various flavors of how they do the Data Science on top of the published data. The Data Science is proprietary with various degrees of skirt lifting to show a bit of ankle here or there.
538 (now that it's got it's shit together(?)) has a fun graphic that I assume(?) is regularly updated... as of today, there's a scenario where Harris wins 519 to 19. So... take heart, Saul and Chip... they're sayin' there's a chance.
I think you're trying to make a meta-critique that the published Polls are not measuring voter sentiment correctly? That could be. That's part of the Data Science as well... which includes historical analysis of previous polls by the same company vs. actual results vs. corrections. But that's why the 'Flooding the Zone' nonsense isn't really an issue for the 'serious' modelers because they aren't using those polls and/or are applying adjustments.
Flooding the Zone is a Media play (to get the horserace people to talk about it) not a thing that influences the 'serious' modelers.
"
I'd probably rather not do Caribbean at this point in my life... but for various reasons simplicity of travel was a factor. So I get the sentiment.
"
I don't think that that's what the people with 'models' vs. simple aggregators are doing.
Modeling is a Data Science project on top of an aggregation project. People are getting mad about the 'modeling' aspect but they are getting mad about the wrong things.
The models aren't predicting the vote percentage of 49.2%, they are saying that the models running multiple cycles show candidate #1 winning 49.2% of the time and therefore candidate #2 winning 50.8% of the time (technically most models are showing ~.2% chance of tie).
The models have, in fact, been on different sides of the 50% divide both among other models and the same model itself at different points of time. A quick google search showed Harris a month ago was favored 58% down to 52% *chance to win* among the modelers. Now the models are even more variable with some showing *Trump* has 51% to win with others showing *Harris* 52% chance to win. Which is why all of them are basically saying the outcome can't be statistically differentiated from 50/50 (with a very small chance of a tie).
What's dumb, IMO, is 'preferring' a model that favors your candidate *as if* it's giving you a better chance to win -- that model has in fact predicted that the OTHER candidate will win in with it's own 'biases' nearly 50% of the time. Sometimes that very model will 'predict' a blow-out win for the candidate who will ultimately lose.
Sometimes the model will even predict the Trumwill 'inside straight' victory for Harris where she loses PA, but wins 7 other toss-up states. This scenario happens for *all* the models... Therefore, if Trumwill is correct, it will have been predicted even by a model that says Trump had a 51% chance to win.
In the end, one candidate will win and vindicate 50% a certain sort of person's assumptions about the models.
Or do you mean something else?
"
The 'weird' thing about the poll wars, at least to me, is the obvious desire for confirmation that your preferred candidate is going to win.
And the 'weirder' thing about this cycle vs. 2016 is that calling the election a statistical dead heat has driven that sort of poll watcher completely mad.
And the 'weirdest' thing of all is that 50% of those sorts of persons will be 100% validated in their unhinged understanding of what polling models do.
On “The Election’s Home Stretch”
I could only see a preview clip, not the full episode, but Cass is an interesting figure to follow; he's deceptively clever and people underestimate his positions and ability to speak on them.
Ultimately, I don't think he's the 'future' of the Republican Party because he's just one guy (ok I think now 4 or 5) at a think tank... but I do think some future right of center faction will pick up his work. And, by and large, that would be a better correction to the Libertarian / Tech-bro right economics that I think we're going to see come after (and in some form during) Trump.
p.s. the small clip isn't a great clip since it's mostly Stewart saying that Trump doesn't follow Cass's policies -- which, yes, but that's not why you have Cass on your show.
On “Campaign Scratchpad: Known Unknowns”
Dang, unless things go wildly astray, we'll be in the Caribbean.
"
"I’m going to be on a cruise on election day"
Hey, we are too! I'll bring my OT monocle so you can recognize me if we're on the same ship.
On “Weekend Plans Post: Thinkin’ ’bout Numbers”
April, August, December, Eleven, February and so on.
"
We're doing woods-work this weekend getting one of our older hunting spots opened up with better sight lines and cutting back as many honeysuckle bushes and tree of heaven as we possibly can without a skid steer. Now I want a skid steer again; skid steer concupiscence is never far from the heart.
"
Game patch logic infiltrating business patch logic.
On “From Freddie: The Basics: School Reform”
The pendulum will swing back; I'm seeing an opening for those of us who've read widely in prompt engineering.
One project I'm working on my engineer told the LLM, 'you are a marketing professional' and a bunch of other stuff like, don't lie, don't add text after the name of the product, and so on... to help sell cookies.
Oh what an opportunity missed is all I could think. Imagine using the voice of a 'marketing professional' how droll.
You are Fafhrd describing to the Gray Mouser the finest cookie you've ever had; embellish with comparisons to local fauna, and after naming the cookie officially, give it a nickname that would tickle us with whimsey. Post to instagram.
On “The Joy Of Opening Time Capsules: The 2024 Presidential Election”
Yes, I think that if we run the election through simulation models multiple times accounting for margins of error and simulate relational effects, sometimes Harris will win and sometimes Trump will win.
You may be right that there won't be 'ticket splitters' but, I think you're wrong in your reasoning; that is, IF there's a reason to split a ticket its to NOT vote for Robinson and NOT vote for Lake, while still NOT voting for Harris... it's exactly because Robinson and Lake are even more odious than Trump, especially in a LOCAL sense.
It's a weird reverse split, but it wouldn't surprise me.
"
Agreed... I'm guess most models assume that if NC or GA goes Blue, both will and if both do, then good chance AZ goes, etc.
But that's just generally the case in either direction... that is, if WI goes Red, then MI and or PA may go that way too.
I guess that's why we open the votes and don't just guess at what the crosstabs imply...
"
To be unburdened; unburdened one must be.
"
NH going Red is a bold prediction... I have friends in NH that are pretty plugged in to local politics... and just when you think maybe NH will flip red, they never do.
"
I gave Harris 35% odds of being nominee, so my being wrong was less wrong.
For simplicity's sake I kept PA as the tipping point... but my alternate path to Blue victory goes through NC and AZ where MAGA nutters could potentially impact the Presidential vote in somewhat unpredictable fashion.
"
There's only one word to describe Trump winning: Inconceivable.
Senate: R+1 (too many Blue seats to defend)
House: Red if Trump wins, maybe even red if Harris wins depending on *how* she wins; that is, if R's quiet quit Trump and split the ticket, then Red; if Harris pulls votes and builds coattails, then potentially it goes Blue.
At this point, I think an assassination would make Vance president... so probably no more assassination attempts.
Wildcard: The Harris Candidacy is unique in that it is both an incumbent and an outsider; it has a record when it wants it and doesn't when it doesn't. I think they are playing this sort of null vector candidacy pretty well... as long as the rest of the world can sit tight for 30 more days, it might just work.
Now, if you'd like to see me act as an official Presidential Elector, then vote as hard as you can for the Pelicans.
On “Let’s face it: We knew that Harris would win back in August”
Heh, I wonder what October 4th's post will be...
On “Open Mic for the week of 9/23/2024”
I don't have to say sh*t; you can't make me.
"
Decapitated, for now.
If prelude to ground war; the disruption could be decisive. If the goal is to 'send a message' then not so much.
Assuming a ground war, I'm not sure what it accomplishes without support from Lebanon's Maronites, Sunnis and 'civilian' Shia governing coalition members.
"
I already double dog dared you... what more do you want?