4 ms·

"Why wouldn't people game the system by entering temperatures as low or as high as allowed?" Because it is median, not average. That does nothing. Demonstrate

"Why wouldn't people game the system by entering temperatures as low or as high as allowed?" Because it is median, not average. That does nothing. Demonstrated here: https://pianop.ly/voting/median.html https://pianop.ly/voting/median.html I think you are wrong about parties forming, there would be no benefit whatsoever. There is no reason anyone would have any incentive to do anything other than vote for their preferred value.

hirundo 7y ago
> There is no reason anyone would have any incentive to do anything other than vote for their preferred value. I think people would quickly come to see voting for their preferred value as wasting their vote, as is often the case with presidential elections. Good point about median vs average, I missed that. But there would be benefit in converging the votes on popular numbers to make them the median. E.g. if I want 70 but think the contest will come down to 72 vs 76, it may make perfect sense to change my vote to 72.
- robbrown451 7y ago
  It may make perfect sense psychologically to do that, but it accomplishes nothing for you. Some people may do things like this, but those things tend to cancel out. They aren't general tendencies, they are just individual quirks. Also, why would it come down to 72 or 76? It's not going to bounce around like that, it will converge on a center value. Regardless, it would make no sense for you to pick anything other than your preferred value. Any value you pick is going to move it the same amount (if it moves it at all). And if you pick 72 when you prefer 70, you are taking a risk of your moving the result in the opposite direction of where you want it to go (as would be the case if the median was around 71) "But there would be benefit in converging the votes on popular numbers to make them the median." "Popular numbers" don't matter with a numerical vote like this. We aren't voting for discrete "candidate values". Again, the best strategy is to vote for you preferred value, full stop. Trying to guess how others are voting would accomplish exactly nothing. It is very different from presidential votes where it may well be wasting your vote to vote for who you actually want. All of the things you say apply there. This is not that.
  - hirundo 7y ago
    > It's not going to bounce around like that, it will converge on a center value. There's no reason for temperature preference to form a normal distribution. In fact I'd think that sexual dimorphism would generate a bimodal distribution at a minimum. Women tend to prefer it warmer than men. That's an incentive for two parties at least. Any form of syndication will fracture the normal distribution, and we are political animals. We frequently vote at odds with our own interests in politics. I would expect the same behavior when voting to set the thermostat.
    - robbrown451 7y ago
      It doesn't need to have a normal distribution for this to work. And there is no incentive for parties to form even if it was a 100% polarized electorate. However, let's imagine a slightly more realistic case where 45% of the room prefers exactly 65 degrees, 46% prefers exactly 75 degrees, and the other 9% are evenly spread in the in-between range. That's far from a normal distribution curves, it is two big spikes and a very low area connecting them. The median preference is probably going to be around 70/71. A reasonable middle ground. The only way it won't arrive at that result is if people are consistently acting against their own interests. If people attempt to get strategic about it, and form parties and agree to vote in a particular way, it won't change the result at all. I'm not sure where you are thinking these incentives come from, but I'm saying they don't exist. What is the incentive for an individual voter of doing anything other than voting for their exact preference? If you are one of those closer to the middle, preferring, say, 72 degrees, your vote will have exactly as much upward pull if you vote for 72 compared to voting for 75. Joining the 75 party doesn't help you (but it potentially hurts you, if you were wrong about the median being lower than your preference). I'll refer again to this, where I built a demonstration app you can play with, or you can just watch the video. If you want it done with polarized distribution, let me know. (it would be easy to hack it to do so) https://pianop.ly/voting/median.html https://pianop.ly/voting/median.html
      - hirundo 7y ago
        > 45% of the room prefers exactly 65 degrees, 46% prefers exactly 75 degrees, and the other 9% are evenly spread in the in-between range ... The median preference is probably going to be around 70/71. That does not sound right. The average would be around 70/71 but the median would be 75. So instead of a moderate compromise, 46% of the room would be happy and (at least) 45% unhappy.

hirundo 7y ago

> There is no reason anyone would have any incentive to do anything other than vote for their preferred value. I think people would quickly come to see voting for their preferred value as wasting their vote, as is often the case with presidential elections. Good point about median vs average, I missed that. But there would be benefit in converging the votes on popular numbers to make them the median. E.g. if I want 70 but think the contest will come down to 72 vs 76, it may make perfect sense to change my vote to 72.

robbrown451 7y ago
It may make perfect sense psychologically to do that, but it accomplishes nothing for you. Some people may do things like this, but those things tend to cancel out. They aren't general tendencies, they are just individual quirks. Also, why would it come down to 72 or 76? It's not going to bounce around like that, it will converge on a center value. Regardless, it would make no sense for you to pick anything other than your preferred value. Any value you pick is going to move it the same amount (if it moves it at all). And if you pick 72 when you prefer 70, you are taking a risk of your moving the result in the opposite direction of where you want it to go (as would be the case if the median was around 71) "But there would be benefit in converging the votes on popular numbers to make them the median." "Popular numbers" don't matter with a numerical vote like this. We aren't voting for discrete "candidate values". Again, the best strategy is to vote for you preferred value, full stop. Trying to guess how others are voting would accomplish exactly nothing. It is very different from presidential votes where it may well be wasting your vote to vote for who you actually want. All of the things you say apply there. This is not that.
- hirundo 7y ago
  > It's not going to bounce around like that, it will converge on a center value. There's no reason for temperature preference to form a normal distribution. In fact I'd think that sexual dimorphism would generate a bimodal distribution at a minimum. Women tend to prefer it warmer than men. That's an incentive for two parties at least. Any form of syndication will fracture the normal distribution, and we are political animals. We frequently vote at odds with our own interests in politics. I would expect the same behavior when voting to set the thermostat.
  - robbrown451 7y ago
    It doesn't need to have a normal distribution for this to work. And there is no incentive for parties to form even if it was a 100% polarized electorate. However, let's imagine a slightly more realistic case where 45% of the room prefers exactly 65 degrees, 46% prefers exactly 75 degrees, and the other 9% are evenly spread in the in-between range. That's far from a normal distribution curves, it is two big spikes and a very low area connecting them. The median preference is probably going to be around 70/71. A reasonable middle ground. The only way it won't arrive at that result is if people are consistently acting against their own interests. If people attempt to get strategic about it, and form parties and agree to vote in a particular way, it won't change the result at all. I'm not sure where you are thinking these incentives come from, but I'm saying they don't exist. What is the incentive for an individual voter of doing anything other than voting for their exact preference? If you are one of those closer to the middle, preferring, say, 72 degrees, your vote will have exactly as much upward pull if you vote for 72 compared to voting for 75. Joining the 75 party doesn't help you (but it potentially hurts you, if you were wrong about the median being lower than your preference). I'll refer again to this, where I built a demonstration app you can play with, or you can just watch the video. If you want it done with polarized distribution, let me know. (it would be easy to hack it to do so) https://pianop.ly/voting/median.html https://pianop.ly/voting/median.html
    - hirundo 7y ago
      > 45% of the room prefers exactly 65 degrees, 46% prefers exactly 75 degrees, and the other 9% are evenly spread in the in-between range ... The median preference is probably going to be around 70/71. That does not sound right. The average would be around 70/71 but the median would be 75. So instead of a moderate compromise, 46% of the room would be happy and (at least) 45% unhappy.