> I was left with the impression that it is the paper records in this story that led to the unravelling of an attempt to forge the results.
The manual tallying of paper records is what lead to the attempt to forge the results in the first place. If the results were electronically tallied to generate an official result, then they wouldn't need to recount the whole election to verify the result, just doing a statistically significant random sampling of the polls to recount would be enough.
> If the results were electronically tallied to generate an official result
Electronic voting doesn't make bad politicians less bad. In this instance, the bad guys were prepared to deliberately remove CCTV so when they sent their goons out at night to shoot protestors there would be no evidence.
"Electronic tallies" are never going to give a free and fair election if those in power are prepared to go that far. Safer to stick with paper ballots and election observers equipped with Mark I eyeballs.
By looking at the receipts printed by the ballot machines.
Ballot machines print either a final tally at the end of the day, or print every single vote and automatically drop it into a physical ballot, depending on the threat model of the country in question. Either way the you have partial or total recount.
> By looking at the receipts printed by the ballot machines.
Let's the clear, you're not really "recounting" the ballots at that point. If the machine is compromised - and we're discussing a situation in which we know CCTV was removed and people were then shot - you have no real idea if the receipt corresponds to the voter's original intent. Or, indeed, if all the receipts from all the voters make it as far as the recount (?)
> Ballot machines print either a final tally at the end of the day, or print every single vote and automatically drop it into a physical ballot, depending on the threat model of the country in question.
How is reprinting the final automated tally supposed to represent a "recount" of the original automated tally?
> Either way the you have partial or total recount.
You really don't. Bits of paper and Mark I eyeballs all the way.
As Tom Scott puts it, "The key point is not is that paper voting is perfect - it isn't - but attacks against it don't scale well"[0].
> How is reprinting the final automated tally supposed to represent a "recount" of the original automated tally?
If you want to detect tampering in the central totalling, then all you need is the end of day receipt of each ballot. Exactly like in OP's case.
If you want to detect tampering in a ballot, then you manually recount the individual printed paper votes inside that ballot. That is something that you should do to a random sample of ballots, plus ballots with unusual totals.
> As Tom Scott puts it, "The key point is not is that paper voting is perfect - it isn't - but attacks against it don't scale well"[0].
That is simply not true, large scale paper ballot tampering scales very well to the point of turning elections, and is much easier to pull off because it happens in the fringe where no one is looking (while tampering the electronic system would require pulling your heist in the IT room where everyone is looking).
> large scale paper ballot tampering scales very well to the point of turning elections, and is much easier to pull off because it happens in the fringe where no one is looking
In many countries, there are many tens of thousands of individual polling stations. A conspiracy to tamper with enough of them to make a difference isn't going to stay secret for very long because it would have to involve too many people. Tampering with paper ballots just doesn't scale, and in most places, election observers with their old-fashioned Mk I eyeballs are allowed to watch what's going on at every stage.
> (while tampering the electronic system would require pulling your heist in the IT room where everyone is looking)
How would we propose that an average human election observer is supposed to detect whether any particular system involved in electronic voting is - or isn't - in the process of "pulling a heist"?
On the contrary, eletronic voting doesn't create the paper trail necessary to dig up frauds like this. You can simply program or hack the system to report any vote total you want.
First of all, hacking the electronic system is much much harder than hacking the paper process. In the case at hand the paper tallying process was the one hacked.
And second, electronic systems can create a paper trail, just make the electronic machine spit out a paper receipt. Then you have the best of both worlds, you can have instant electronic totals, and then do some random sampling recounts of the receipts to validate the result.
Scaling an attack against paper is incredibly difficult, and requires coordination in a level that is almost sure to trigger the law enforcement much before it can change some national-level numbers.
Scaling an attack against a computer system is almost the same as doing an attack against a computer system. Few attacks don't scale.
But yeah, if you just print the vote and push it into an urn (while the voter can read it), you'll get the best of both worlds.
You introduce technology to increase transparency and fight corruption. You increase transparency by having video recordings of human counting votes linked to the electronic record of the totals.
When you introduce technology to eliminate manual counting and paper trails, then transparency is eliminated and you give a green light to fraud, corruption, very juicy contracts and death.
