It boils down to the fact that most states are red or blue. They are going one way or the other and nothing is going to change that. In the contested states, there are tight votes and there are ones that are "in the bag". in the states with tight elections, it will boil down to a few geographic regions within those states that carry enough weight to swing the entire state one way or the other.
Not always. In the 2004 election, Kerry won in exit polls but lost in the official count. The discrepancy was especially large in Ohio--which, had it gone as predicted by exit polls, would have resulted in President Kerry.
A few days ago there were big gaps between official and exit poll results in Arizona, favoring Clinton over Sanders.
I wouldn't normally link to Democratic Underground, but this post has a great summary of exit poll discrepancies, voter suppression, and the evidence for election fraud both in the 2004 presidential election and in the current Democratic primary.
http://www.democraticunderground.com/12511568043
This is a very urgent problem that I think the tech community can help fix. We need open source, auditable voting machines and professionally run elections. Election fraud, or even just the appearance of fraud, is toxic to democracy.
Also, the deviation in the last elections from the mathematically expected was about 4%...
But the end results was 51% vs 49% meaning that if the result went as expected, the 49% candidate would have won (with 53%).
So, in some situations is not that hard to cheat.
EDIT: also even when there is blatant cheating going on, nothing much can be done about it...
Example, during elections for a minor office in one of the smallest Brazillian states (Acre), the number of invalid votes, intentionally blank votes, and voters that didn't show was exact the same, and the total number of votes was higher than the number of voters.
The losing candidate sued, the government argued that "the computer counts, and thus a recount is not a valid request", and FINED the guy that sued.
And the whole argument stopped there.
As someone else points out, you don't need to hack the whole country, just some key locations. Also, you could hack the source; many firewalls from a major vendor had (have?) security holes because someone checked the necessary code into their revision control system. How secure is Diebold or whoever makes the machines?
> The reported vote count is usually very close to the predicted vote count
Suprise results, where the outcome doesn't match the polls, are not that rare. Recently Sanders' win in Michigan is an example. In the UK, in the last parliamentary election, I believe the conservatives did much better than polls predicted.
> always very close to exit polling results
If true, this is a useful check.
> hacking could not be used to create a very close result that could trigger a recount, or it would need to be combined with actual ballot theft
That assumes there are ballots to steal; many voting machines are electronic.
Yes, polls can be wrong, but I think exit polls are rarely wrong. Another comment pointed out these can be wrong too though, so this might not be a useful point.
Yes, as I mentioned in another comment, if you want to hack a real election you need to take a long-term view and make electronic voting widespread.
This machine then is the target, no?
I'm not aware of this. Are you sure? Do you have any point of reference?
I will try to do more research tomorrow, because this is quite interesting to me. I voted with a paper ballot that I fed into a machine, and at the time I assumed this was the standard. Now I'm not sure.
That said, appears like given your points that I agree a better use of the idea would be to insure an election and reduce costs too. Meaning run a real campaign, stay behind, but close enough to not draw attention, which I'm guessing would reduce costs more than enough to cover the added costs of hacking.
I can't think of a low-risk way to literally hack an election in the US at the moment. You would need a long-term plan to change election laws, homogenize voting machines and methods across the country, and then hack some centralized database.
There is no way to get around the polling issue in a country with an even semi-free press, so you still need a pretty strong candidate that you just boost a little.
Anybody in the real world who needed to rig an election would use indirect methods. Spying, spreading misinformation, ignoring campaign finance laws, gerrymandering, etc. All things that have happened and will continue to happen.
It didn't seem like a problem for Diebold. The 2000 general election had a lot of awfully-convenient-for-Republicans scandals with their voting machines in place.
And outside of squeaking by in Florida with an assist from SCOTUS, the 2000 general election was terrible for Republicans. They lost a few extremely close Senate races (ultimately resulting in a 50-50 tie) that any competent conspiracy could have pushed the other way.
1. I have seen video demonstrations wherein one person with opportunistic unsupervised access to a voting machine can hack it to affect every issue on the ballot within five minutes. Given the security at many polling places in the US, this easily extrapolates to one person able to hack all machines at a given polling place prior to election day, and possibly repeat the effort at several other nearby polling places within counties using the same type of machine.
Other hacks target the central tabulators, and don't require touching the voting machines at all. Opportunistic timing to enjoy unsupervised access between the election and results certification is a bit more critical to these.
2. Exit polls have frequently been "adjusted" in the US to more closely match the actual elections results. In the UK, this is called the Shy Tory Factor[0] or Shy Labour Factor, and has existed since 1992. The US has had the Tom Bradley Effect[1] since 1982.
This is usually explained by presuming that the exit polling result is inaccurate. It is never seriously considered in public that vote fraud may be becoming more severe. Yet I saw a statistical analysis of actual voting results, differentiated by the model of voting machine, that showed a clear, obvious bias towards certain choices of specific issues. The poll adjustment factor can easily be explained by the hypothesis that people remain as honest as ever in exit polls, but the voting process itself has become less honest.
3. Recounts will not uncover fraud if the records necessary for the recount are subject to the same controls as the records used for the first official count. If the voting machine does not leave a non-electronic record that can be verified by the voter while they are still at the polling place, a recount will just be run on whatever records exist, which may be falsified.
4. Thanks to the winner-take-all elections in the US, and the availability of previous census and elections data, it is possible to identify specific counties as keystones, wherein efforts to sway the vote locally will have disproportionately effective results.
For instance, you examine the "battleground counties" in the "swing states", and hack the central tabulator in that county to give your favorite candidate a 3% advantage by switching an opposing vote 1.5% of the time. You don't tell your party you're doing it. You don't even tell your dog that you're doing it. You just do it. If your guy loses, well, you tried.
Multiply by dozens of technically competent yet unconnected and independent supporters, and you get a significant extent that does not require any team.
"It's not the votes that count, it's who counts the votes."If you honestly believe there is nothing any candidate can offer you which is worth more than $100, I propose you should be allowed to take that deal.
( Of course, completely open source hardware and software is the real solution. )
EDIT: I am a little bitter because I'm in a state using machines which don't even keep a paper trail to begin with. https://www.verifiedvoting.org/verifier/
To affect a national election, you'd still need a small army of helpers - "vote smurfs" to inject your votes. Chances are very slim of keeping that quiet.
Individual polling place results are always reported, so just modifying the total count would not work.
2. This is a good point, but again, look at Florida. It was very seriously considered, and is still seriously considered by some people, that real widespread vote fraud occurred.
It's an interesting effect though, and I agree that widespread subtle adjustments might not be caught by looking at polls.
3. Yes, all of this would be much easier if voting was totally electronic. In the US, the vast majority of voting is done with paper ballots that are read by a machine. I think at least 3/4 of votes use paper ballot.
4. Yes, I agree, you wouldn't need to hack the whole country. But on the other hand, battleground counties are also ones the media and the public is paying the most attention to.
I don't think your example is realistic. I find it very difficult to believe that nobody would notice the individual district counts don't add up to the total count in a county. Or maybe that's not quite what you meant, but regardless, the system is not that automated. There are people at every stage of the process who would need to notice that what they're reporting is not what's going through.
1) Hacking voting machines to a significant extent would require a team of people, not just one guy, which dramatically increases the risk of exposure.
2) The reported vote count is usually very close to the predicted vote count, and always very close to exit polling results. The hacked results would need to be very close to the real results, or be combined with control over a huge number of people in the polling industry. Campaigns also run internal polls, so opponent campaigns might become suspicious if their own polls were not matching the results.
3) Recounts are possible and somewhat common. Therefore hacking could not be used to create a very close result that could trigger a recount, or it would need to be combined with actual ballot theft, or control over the officials doing the recount.
These are not theoretical concerns. See Bush v. Gore in Florida for what might happen if someone tried to hack a close election.
It seems to me that hacking an election would only be feasible in an already close race, would require the cooperation of a number of people with huge risks of exposure, and could still be caught by one of the many safeguards that exist.
Obviously I am only talking about hacking voting machines. It would not surprise me if the tactics in this article were common; using Twitter bots, spying on other campaigns, etc.