Main video: vnclip.net/video/tUX-frlNBJY/video.html Post with spreadsheet download: www.cgpgrey.com/blog/the-sneaky-plan-to-subvert-the-electoral-college
I just wrote a 15 page paper on why the electoral college is actually great ;) thanks grey for going into a lot of nitty gritty for me, it really allowed me to get my ideas out there I think the problem ultimately comes down too; 1) do you like seperation of powers 2) do you like representatives 3) should states matter? for 1) theres the British system, one legislature, I think they do matter though 2) plebiscite clauses have their place, not for literally everything though... 3) state should matter! regional issues are a thing!! if you believe in one, you believe in the electoral college, and not the popular vote :D
I had an idea to illustrate the apportionment of seats by showing the states climbing up the priority list (in Javascript), a couple of years ago really. Seeing this inspired me to get it done! www.gogococo.co.uk/apportion
...the data scientist in me, partly appreciates the ingenuity, but another part me looks in horror. I just think of my co-workers I support doing this...
Your such a nerd you made a 42 minute video explaining Apportionment and how you're not going to explain apportionment! And then added each of those states one by one. NERD!!!!
And then I sat here and watch the entire thing... I need better things to do with my life.... oh well when's the next video? :)
So I can't help but wonder about the systems used before the Huntington Hill method, and how final votes might have changed if we redistributed the electoral votes from those elections using the Huntington Hill method. How might history have been different if we had better math, earlier?
Spreadsheets are the most used programming environment in the world. And they're not just functional, but functional-reactive! Who says functional programming is hard?
38:00 Grey mentioning hattricks got me thinking, how did hattricks get their name, it feels like there's a story behind that name. I looked it up, so, random fact: Hat-trick once referred to a magician consecutively taking 3 rabbits out of his hat. The term was used in sports for the first time in a cricket game in 1858
Huge fan of CGP Grey, but I must admit even I started to skip after 25 minutes... one can't help but think you could have found a faster macro programme - I recommend WinAutomation Pro... doh, but you use a Mac ;-)
Gotta say I really enjoyed this. I also realized once again that it is no wonder your videos take so long. Not that that is a bad thing in itself, you have a passion and excitement for learning and understanding which really shines through. And whilst I really enjoy your scripted content, I would love to see more of this type of ramble-y, non-animated stuff as well to tie me over. More please!
Your method is a bit naive. It considers only the inequity in apportionment before adding the next seat, not the inequity after adding it. The Huntington-Hill method considers both.
Think of it sort of like this. If you increase something from 5 to 10, you have doubled it--a relative increase of 100%. But if you decrease something from 10 to 5, you have halved it--a relative decrease of 50%. How can we describe the "relative difference" between 5 and 10? There are multiple ways to do it. We might say there is relative difference of 100% or of 50%, but neither is satisfactory for this purpose. We could also take an average approach: instead of dividing the difference (10-5=5) by the initial amount (5) or by the final amount (10), we divide by the average ((5+10)/2 = 7.5). Then we get 5/7.5 = 67% relative change. In fact though, the optimal method from the perspective of game theory here, and the conceptually correct method in general, is to use not the arithmetic mean but the geometric mean. We consider the number of voters per rep both before and after adding the next seat and take the geometric mean. And it is state with THIS maximum value that most deserves the next seat.
Okay, I was wondering how older elections would have gone using the proposed NPVIC. Would it have changed anything other than wildly changing the winner's number of electoral votes? Would it make Reagan win by more? Would it have made Nixon beat Kennedy or would everything be the same? There's a video to put in a spreadsheet, all the elections from 1952 till 2016 (the post WWII elections) in NPVIC vs way they are normally done vs proportionate allotment (this last is for seeing how many third parties would get electors).
So, at 2 minutes and 30 seconds you are showing how many people there are per EC vote. How does that compare to number of people per House of Representative? Is there similar comparisons to which state is most represented and which is least? EDIT: Okay, I should watch more of the video.
Quick thought/question, in't the formula for this wrong? In proportional ares indeed its close enough to right, but in first past the post gets everything, wouldn't the fomula be closer to (population/2) / number of ec
Just a newbie here, but the spreadsheet got me to wondering how predicting the state with the most likely increase in population (using population models) can identify states most likely to move up on priority. There must be a way to tag these differences (future predicted population minus future constant population) and see the descending order. Those states at the top are the most likely place to gain a new seat and should be prioritized for political advantage, based on predicted population shifts. I guess then you could target more swing/undecided districts in those states to see where to target marketing/message (target rich). I'm just saying the chart might be useful for political considerations, outside of my poorly worded example. Great video - wonderful to watch.
What if the “minimum number” started at 3 and accounted for senate seats in these calculations? Would that change anything for the mid-population states?
I'm sorry Grey. I love your work. I've watched the Tesla feature twice. At nine minutes in I just can't do it anymore. Thanks for all the wonderful work you do.
Also, I know he'll never see this, but eh fuck it, whatever.
Huntingon-Hill method is not significantly different from just picking who is "the worst off"... A/sqrt(a(a+1)) ≈ A/sqrt(a^2) = A/a, which is the simple method. The two methods turn out about the same in the end, only two changes larger than one seat if we were to switch to simple method (TX -2 and CA -3). I did some spreadsheeting myself. If we switched to apportionment using the simple method and the 2010 census data, these changes would occur (with all states not listed having no change in seats): CA -3, DE +1, FL -1, GA -1, ID +1, IA +1, LA +1, MO +1, MT +1, NY -1, OK +1, OR +1, SD +1, TX -2, WA -1
It would have been nice to see how the result differs from what you'd get by following the first algorithm you proposed, just gviving the next seat to whoever has the least per capita.
If you watched this whole video, you'd love the Center for Election Science. We're a 501(c)3 dedicated to improving the way we vote. We recently passed Approval Voting in Fargo, ND, and have a campaign in St. Louis, MO for the 2020 ballot. We're also a collection of voting scientists, and love discussion the various benefits and intricacies of how we currently vote, and how we could vote for a more optimal outcome. Check out electionscience.org
This really should have just been a simple python script, not an ugly spreadsheet. Spreadsheets look simple to do at first, but they are really limiting and force you to do ugly hacks. Just code this in python, and solve it automatically. I bet that would have taken at most an hour.
Here's an example to show why grey's method of giving the next seat to the least represented state doesn't hold up to mathematical scrutiny in some cases. Imagine the following scenario. State A has 1 person and 1 seat. Ratio is 1 seat per person. State B has 300 person and 301 seats. Ratio is a tiny bit over one seat per person. Better representation than state A. State C has 300 person and 302 seats. Ratio is a tiny bit higher than B's and also better than A's. The seats per person ratio is the worst in state A but if you give them a new seat (as per cgpgrey's suggestion) it makes it way worse than giving an extra seat to state B instead because state A would jump from one seat per person to two seats! So much better than state B and C.
The method discussed actually makes a lot of sense and there's a simple way to understand how it works. First, you start with the 309,183,463 population and divide it equally between the 435 seats in the house. That means, ideally, every seat should represent 710,767 people. If you take the population of each state and divide it by 710,767, and round to the nearest whole number, you will get the number of seats for each state. When you do that, you get 433 seats filled. The next two seats go to the states who are closest to getting the next seat. At this point, you take the population of the state, divide it by 710,767 and look where the decimal is closest to 0.5 without going over. Those states are Minnesota and Rhode Island. As you noted at the end, the next state that would get an elector is North Carolina. Another way to look at it is every 710,767 people in a state gives them a seat in the House. Then, starting with the state that is closest to getting another seat will get one more seat in the house, continuing down until all the seats are filled. The method is simply a way to mathematically represent this equitable distribution of seats.
Sacre Bleu. A spreadsheet on VNclip without Matt Parker. There is probably a way to view this as a collection of multidimensional points, with an operation that minimises their distances, or perhaps maximum radius, according to some metric. It probably doesn't have to be run incrementally.
Something tells me if you were to ask every congressional representative how this system works, only a small fraction would be able to give you so much as a vague outline. "Look, I just campaign in a popularity contest and either the numbers people tell me I win or they don't."
No wonder Texas wants to leave.
I just wrote a 15 page paper on why the electoral college is actually great ;)
thanks grey for going into a lot of nitty gritty for me, it really allowed me to get my ideas out there
I think the problem ultimately comes down too; 1) do you like seperation of powers 2) do you like representatives 3) should states matter?
for
1) theres the British system, one legislature, I think they do matter though
2) plebiscite clauses have their place, not for literally everything though...
3) state should matter! regional issues are a thing!!
if you believe in one, you believe in the electoral college, and not the popular vote :D
I had an idea to illustrate the apportionment of seats by showing the states climbing up the priority list (in Javascript), a couple of years ago really. Seeing this inspired me to get it done! www.gogococo.co.uk/apportion
...the data scientist in me, partly appreciates the ingenuity, but another part me looks in horror. I just think of my co-workers I support doing this...
What woukd happen if you kept adding seats? Would it ever reach a range smaller than 10,000?
You should monetise more I could easily watch more ads for this level of content
Please put in ads we want you to make money
17:22 okay, now we are talkin *pulls seat closer*
doenst this belong to cgp play?
Your such a nerd you made a 42 minute video explaining Apportionment and how you're not going to explain apportionment! And then added each of those states one by one. NERD!!!!
And then I sat here and watch the entire thing... I need better things to do with my life.... oh well when's the next video? :)
So I can't help but wonder about the systems used before the Huntington Hill method, and how final votes might have changed if we redistributed the electoral votes from those elections using the Huntington Hill method.
How might history have been different if we had better math, earlier?
Spreadsheets are the most used programming environment in the world.
And they're not just functional, but functional-reactive!
Who says functional programming is hard?
38:00 Grey mentioning hattricks got me thinking, how did hattricks get their name, it feels like there's a story behind that name.
I looked it up, so, random fact:
Hat-trick once referred to a magician consecutively taking 3 rabbits out of his hat.
The term was used in sports for the first time in a cricket game in 1858
hey, could you please explain the impeachment process of the POTUS?
Someone needs to make a 50v50 state's fighting tournament video series
“Spreadsheets are a bicycle for the mind”
why didn't you order it by number of wins it would have been easier IMO
Spreadsheet ASMR
They also take votes from states like California and give them to Wyoming
Huge fan of CGP Grey, but I must admit even I started to skip after 25 minutes... one can't help but think you could have found a faster macro programme - I recommend WinAutomation Pro... doh, but you use a Mac ;-)
Gotta say I really enjoyed this. I also realized once again that it is no wonder your videos take so long. Not that that is a bad thing in itself, you have a passion and excitement for learning and understanding which really shines through.
And whilst I really enjoy your scripted content, I would love to see more of this type of ramble-y, non-animated stuff as well to tie me over. More please!
Your method is a bit naive. It considers only the inequity in apportionment before adding the next seat, not the inequity after adding it. The Huntington-Hill method considers both.
Think of it sort of like this. If you increase something from 5 to 10, you have doubled it--a relative increase of 100%. But if you decrease something from 10 to 5, you have halved it--a relative decrease of 50%. How can we describe the "relative difference" between 5 and 10? There are multiple ways to do it. We might say there is relative difference of 100% or of 50%, but neither is satisfactory for this purpose. We could also take an average approach: instead of dividing the difference (10-5=5) by the initial amount (5) or by the final amount (10), we divide by the average ((5+10)/2 = 7.5). Then we get 5/7.5 = 67% relative change.
In fact though, the optimal method from the perspective of game theory here, and the conceptually correct method in general, is to use not the arithmetic mean but the geometric mean. We consider the number of voters per rep both before and after adding the next seat and take the geometric mean. And it is state with THIS maximum value that most deserves the next seat.
Electoral college would be much more proportional if the number of congressional seats wasn't capped arbitrarily last century.
what am I doing watching this? I have an essay due two weeks ago, and I haven't any more than a sentence done.
"Literally nobody cares: the video"
Okay, I was wondering how older elections would have gone using the proposed NPVIC. Would it have changed anything other than wildly changing the winner's number of electoral votes? Would it make Reagan win by more? Would it have made Nixon beat Kennedy or would everything be the same?
There's a video to put in a spreadsheet, all the elections from 1952 till 2016 (the post WWII elections) in NPVIC vs way they are normally done vs proportionate allotment (this last is for seeing how many third parties would get electors).
So, at 2 minutes and 30 seconds you are showing how many people there are per EC vote. How does that compare to number of people per House of Representative? Is there similar comparisons to which state is most represented and which is least?
EDIT: Okay, I should watch more of the video.
It was originally Arr-Kansas, and the state was founded by pirates
Quick thought/question, in't the formula for this wrong?
In proportional ares indeed its close enough to right, but in first past the post gets everything, wouldn't the fomula be closer to (population/2) / number of ec
You really love this states rights thing eh?
Got a a confederate flag up in your bedroom no doubt
The takeaway is that partisan politics, inevitable as it is, is _the_ problem.
Fun! I learned from Martin Shkreli that spreadsheets can be exciting, when they're good... and this one is damn good! thanks :)
Dude i love this video, please make more videos like this
That was long and peaceful, I really liked that.
Serious question: why Numbers rather than Excel? As a spreadsheet enthusiast, what are the pros and cons of each?
"Adoptive state of North Carolina"
You have piqued my interest. As a North Cackalackian, story please? Or if you've done video, linky please?
@Cooper Smithson Music I see then.
Kinkajou1015 pretty sure his family's from there, I think that's where he went in the "Summer of Grey" series
Just a newbie here, but the spreadsheet got me to wondering how predicting the state with the most likely increase in population (using population models) can identify states most likely to move up on priority. There must be a way to tag these differences (future predicted population minus future constant population) and see the descending order. Those states at the top are the most likely place to gain a new seat and should be prioritized for political advantage, based on predicted population shifts. I guess then you could target more swing/undecided districts in those states to see where to target marketing/message (target rich). I'm just saying the chart might be useful for political considerations, outside of my poorly worded example. Great video - wonderful to watch.
That is an awesome spreadsheet. I am defiantly going to be downloading this and playing around with it.
What if the “minimum number” started at 3 and accounted for senate seats in these calculations? Would that change anything for the mid-population states?
Could you possibly do a video on the possible effect of having more than 435 house seats, and the effect it would have on politics as a whole?
Grey incrementing values in a spreadsheet ASMR.
I'm sorry Grey. I love your work. I've watched the Tesla feature twice. At nine minutes in I just can't do it anymore. Thanks for all the wonderful work you do.
Also, I know he'll never see this, but eh fuck it, whatever.
Huntingon-Hill method is not significantly different from just picking who is "the worst off"... A/sqrt(a(a+1)) ≈ A/sqrt(a^2) = A/a, which is the simple method. The two methods turn out about the same in the end, only two changes larger than one seat if we were to switch to simple method (TX -2 and CA -3).
I did some spreadsheeting myself. If we switched to apportionment using the simple method and the 2010 census data, these changes would occur (with all states not listed having no change in seats):
CA -3, DE +1, FL -1, GA -1, ID +1, IA +1, LA +1, MO +1, MT +1, NY -1, OK +1, OR +1, SD +1, TX -2, WA -1
Oddly compelling to watch..
It would have been nice to see how the result differs from what you'd get by following the first algorithm you proposed, just gviving the next seat to whoever has the least per capita.
I always appreciate the New Mexico flag love, Grey!
Love this broken-in-some-ways-and-beautiful-in-many-others state.
The 435 rule is stupid. They should just set a minimum at 1 and add members until the relative population approaches the least-populated state.
In the state vs state region of the sheet, the green cell is always the left one. D:
If you watched this whole video, you'd love the Center for Election Science. We're a 501(c)3 dedicated to improving the way we vote. We recently passed Approval Voting in Fargo, ND, and have a campaign in St. Louis, MO for the 2020 ballot. We're also a collection of voting scientists, and love discussion the various benefits and intricacies of how we currently vote, and how we could vote for a more optimal outcome. Check out electionscience.org
a 43 Minute "footnote"? lol
TL;DW
This really should have just been a simple python script, not an ugly spreadsheet. Spreadsheets look simple to do at first, but they are really limiting and force you to do ugly hacks. Just code this in python, and solve it automatically. I bet that would have taken at most an hour.
@CGP Grey screw the spreadsheet (ok, actually not, but,) I want the source for those two papers! please?
Here's an example to show why grey's method of giving the next seat to the least represented state doesn't hold up to mathematical scrutiny in some cases.
Imagine the following scenario.
State A has 1 person and 1 seat. Ratio is 1 seat per person.
State B has 300 person and 301 seats. Ratio is a tiny bit over one seat per person. Better representation than state A.
State C has 300 person and 302 seats. Ratio is a tiny bit higher than B's and also better than A's.
The seats per person ratio is the worst in state A but if you give them a new seat (as per cgpgrey's suggestion) it makes it way worse than giving an extra seat to state B instead because state A would jump from one seat per person to two seats! So much better than state B and C.
The method discussed actually makes a lot of sense and there's a simple way to understand how it works. First, you start with the 309,183,463 population and divide it equally between the 435 seats in the house. That means, ideally, every seat should represent 710,767 people. If you take the population of each state and divide it by 710,767, and round to the nearest whole number, you will get the number of seats for each state. When you do that, you get 433 seats filled. The next two seats go to the states who are closest to getting the next seat. At this point, you take the population of the state, divide it by 710,767 and look where the decimal is closest to 0.5 without going over. Those states are Minnesota and Rhode Island. As you noted at the end, the next state that would get an elector is North Carolina.
Another way to look at it is every 710,767 people in a state gives them a seat in the House. Then, starting with the state that is closest to getting another seat will get one more seat in the house, continuing down until all the seats are filled. The method is simply a way to mathematically represent this equitable distribution of seats.
"I'm a big fan of spreadsheets, and a fan of big spreadsheets!"
Why I'm watching this if I'm not even american? lol
Sacre Bleu. A spreadsheet on VNclip without Matt Parker.
There is probably a way to view this as a collection of multidimensional points, with an operation that minimises their distances, or perhaps maximum radius, according to some metric.
It probably doesn't have to be run incrementally.
Did CP already cover this apportionment previously? or was it somebody else?
Something tells me if you were to ask every congressional representative how this system works, only a small fraction would be able to give you so much as a vague outline. "Look, I just campaign in a popularity contest and either the numbers people tell me I win or they don't."