Google Trends Electionland • Hacker News Board

Re: Google Trends Electionland

#101

Post by TallGuyShort » Tue, Nov 08, 2016, 5:50 PM UTC

This is cool, but seems to me to be very especially vulnerable to the Observer Effect. Any fluctuation in these search terms that shows up here will cause people to try and find out more by searching for those terms themselves. What may have started out as random noise gets fed through a feedback loop and amplified, while not necessarily being signal.

I saw an increase in searches for voter intimidation near where I live. I immediately went to DuckDuckGo, and "!n voter intimidation". Now I'm part of the problem, apparently :)

Re: Google Trends Electionland

#102

Post by arcticfox » Tue, Nov 08, 2016, 5:51 PM UTC

folksinger wrote: post #91

nostromo wrote: post #89

Earlier quoted context omitted.

https://xkcd.com/1138/

Yeah but the maps I'm showing don't follow general population density so the point being made by this comic doesn't apply in this instance.

Unless I'm reading something on your graph incorrectly it's almost perfectly correlated with Martha Stewart Living subscribers. Looks like it applies almost exactly to me.

Re: Google Trends Electionland

#103

Post by myowncrapulence » Tue, Nov 08, 2016, 5:54 PM UTC

mangeletti wrote: post #87

komali2 wrote: post #84

Earlier quoted context omitted.

I know it sucks but it's good that you at least voted provisionally. If you get any kind of receipt, etc, it can help if there's class actions following the election. Given the state of this election... there very well might be.

Definitely. EDIT I hrhmmm, hope there is some evidence of my ballot.

A lot of states have "track my ballot" systems in place for voters to ensure their votes were counted. Maybe try looking it up in your state.

Re: Google Trends Electionland

#104

Post by TallGuyShort » Tue, Nov 08, 2016, 5:54 PM UTC

AznHisoka wrote: post #64

I just really want to know how many estimated ppl voted for each candidate in each state. Couldn't they just extrapolate that based on search/browsing history (IE. if someone visited BreitBart, they're probably voting for Trump, if someone visited 538, most likely Clinton)

I would think that's extremely unreliable. If it was true, it would be really bad. I want to know what Trump and Clinton are saying regardless of who I vote for. I visit BreitBart but voted for relatively few Republicans. I think one of the main problems with our political process is that our news feeds are turning into echo chambers, polarizing everyone who doesn't go out of their way to really think critically and expand their horizons.

Re: Google Trends Electionland

#105

Post by ethanbond » Tue, Nov 08, 2016, 5:59 PM UTC

Sacho wrote: post #96

bowmessage wrote: post #81

Earlier quoted context omitted.

Voting machine problems though?! There are no voting machines here.

If you dont have voting machines, it would make sense for you to google about them to understand what other voters are talking about, no?

Maybe, but that's kind of the point. A search string + location + change in occurrence tells you literally nothing.

Re: Google Trends Electionland

#106

Post by throwanem » Tue, Nov 08, 2016, 6:01 PM UTC

morganvachon wrote: post #16

kens wrote: post #11

Earlier quoted context omitted.

Why are xkcd mentions on HN always prefixed with Obligatory?

It's not so much about HN, it's a meme that's been around nearly as long as xkcd has. I recall seeing "Obligatory xkcd" on Fark.com as far back as 2007 or so. I've also seen it on Reddit and similar forums over the years; hell I've even seen it in Usenet discussions.

> nearly as long as xkcd has

It's been around a lot longer than that; I first saw it in version 3.0.0 of esr's Jargon File, which was released in 1993, and I doubt it was newly added then.

Re: Google Trends Electionland

#107

Post by deckar01 » Tue, Nov 08, 2016, 6:03 PM UTC

nthitz wrote: post #77

traskjd wrote: post #36

What's the visualisation library they're using here? Looks great (I'm guessing D3, but it's hard to break through the webpack file to figure it out).

React + d3 + Canvas + More. Disclaimer: I helped build it.

You might consider including the BSD licenses those projects were published under in your minified JavaScript file. Is this project's source code published anywhere?

Re: Google Trends Electionland

#108

Post by dominotw » Tue, Nov 08, 2016, 6:05 PM UTC

whybroke wrote: post #74

simonh wrote: post #33

Earlier quoted context omitted.

Come on, Trump very deliberately made voter fraud, of which voter intimidation is a big component, a major campaign issue.

Along with a sign-up form on his web site for 'volunteers' to drive over into 'suspicious' neighborhoods in blue counties. And the mere threat of intimidation will be enough to keep people home googling instead of voting. At any rate, Oregon provides the perfect control for any theoretical 'googling out of idle curiosity' crowd.

>And the mere threat of intimidation will be enough to keep people home googling

What does 'intimidation' mean though? what are they going to do?

Re: Google Trends Electionland

#109

Post by folksinger » Tue, Nov 08, 2016, 6:07 PM UTC

arcticfox wrote: post #102

folksinger wrote: post #91

Earlier quoted context omitted.

Yeah but the maps I'm showing don't follow general population density so the point being made by this comic doesn't apply in this instance.

Unless I'm reading something on your graph incorrectly it's almost perfectly correlated with Martha Stewart Living subscribers. Looks like it applies almost exactly to me.

Well I already did the work I'm willing to do on my end. Further research would obviously have to take in to consideration deviations from general population density, but just looking at the map, there's a lot of voter intimidation being reported in the south that clearly has nothing to do with population density.

Also, there's no correlation with voter intimidation and population density for cities in the midwest.

So I guess it's just my word and my graph against your word and, well, nothing else. So you better get to work!

Re: Google Trends Electionland

#110

Post by nickysielicki » Tue, Nov 08, 2016, 6:10 PM UTC

dmix wrote: post #69

Apparently a bunch of states (23?) are using a software system called Interstate Crosscheck to look for "double voters". > Election officials in more than two dozen states have compiled lists of citizens whom they allege could be registered in more than one state – thus potentially able to cast multiple ballots – and eligible to be purged from the voter rolls. The problem is that it often only uses a persons name as…

> So if a person votes with the same first/last name as another person in another state, it's possible that vote could be wiped out. It was even matching names even though there were differences in middle names or had Jr/Sr at the end.

Funny how this exact same problem was a non-issue for the left when it came to "No-fly, No-Buy."