Yeah, I read a bit, some documentation... it’s a representation of high dimensional data in low dimension, but is the vector basis the best one? Because, of course, they have to estimate the bases through algorithms, which, I don’t know how they do it, obviously.
There is one piece of, could be magic like any sufficiently advanced technology is, which is how it finds that biggest vector.
I’m just wondering what the metric on that graph...?
OK.
There’s actually a metric in there.
Here is where I am a bit sad not to speak Mandarin, but [inaudible 17:52]. [laughs]
How does it reward them?
That would be...
Yeah, the problem is, does it scale to 20 or 60 million. [laughs]
Hmm?
It works well in practice?
That’s one of the things that’s really interesting. I looked how to set it up and...
How do we get that? We get that by a public debate. You’ve been using Pol.is?
That’s the goal of POP, is to say even if we use RSV -- we’re considering it -- if we want to have the legitimacy of 10,000 people choosing for the rest, then the only way to have that is to have the rest informed, it turns out.
Yeah, but letting some group of people chosen randomly choose for the rest? It wouldn’t be a referendum. It would be...
What?
There’s probably a sweet spot before people have election fatigue, but also not too long away between two votes so that they don’t feel disenfranchised or left out. The problem is, do you think that people will be ready to accept that?
If you ask 10,000 people each time, and they’re random, in Taiwan, they will get one question every one year, two years. It doesn’t matter. You can actually fiddle with it so that people have a reasonable amount.
The problem is that for online voting, people could be convinced, I’m guessing. The fundamental of RSV is that we take a sample group. You take 10,000 people. That’s what allows you to have policy questions every day if you want.
That would be surprising. It would.
Anyway the margins -- the compounded errors -- less than 0.1 percent, so it’s nothing.
I don’t think we have that. I think it’s too late to implement it. The election is in a few days.
Which was...?
That is nice. When did you implement that?
France has one of the best systems. We really are secure and everything. In the end, we still have about 10 percent of our counting offices reporting errors. But it’s each time one or two ballots, so really doesn’t change anything. People really overestimate the correctness...
Do you know the error rates in the paper ballot voting?
As a way to do that, it could be the future. The problem that we have is that how do you get people to trust that?
Yeah, you just tell the people, "Well, if you want, you can vote online. If you can’t, you can ask anyone in your family, or even come to the City Hall, and anyone will help you vote, and will not know what you voted for, and you will be able ...
The thing is that you can vote through a proxy but without actually telling them what you voted for, and while being able to verify that they voted later. Voting by proxy could be a very nice way to access the last 20 or 30 percent.
At least we got the security down. The thing that we tried recently in practice -- and it works quite well -- is that you can reach the people who don’t have Internet. We were at a conference, and a lot of people there didn’t have Internet.
The mathematical feature is that it’s mathematically secure, that you can’t hack the system. If you manage to hack the system, the best you can get -- and it’s hard to get -- is the identity of the people who vote, and not what they voted for, and you can’t ...
Really nice mathematical feature. Are you familiar with ThreeBallot?
Because the thing is that, actually, to answer some of your questions about RSV, that’s one of the nice features is it’s an online system that doesn’t have to be online. I don’t think you have time to read the documentation on that because it’s long.
Yeah. [laughs]
Yeah, but even then, you remember the team, the...
The problem is that you do have the security critical point...You know, what happened in Estonia. Their voting system is catastrophically vulnerable but they’re still using it.
Not about how it’s used, whether it’s for policies or for people, the very interface. The fact that the voting could be online. Do you think it would be dangerous with regards to...It has two big problems. The first is that 20 percent people who don’t really have Internet access ...
That’s the idea of POP, that even if you vote for a party or a person, the agenda sets only part will correspond to what you want. Then, once they get in power or not, they have to change their agendas to fit with other politicians. Then again, maybe you ...
Quick question. What do you think of online voting? Voting for general elections?
It unites the will.
That is problematic when you want a public debate that happens only online.
Has broadband or...?
78% of what?
How much of the population, you know?
That comes also from the self-awareness? There is a strong public desire for that.
Trying to create a public discussion.
It seems that a lot of what you’re doing is open data and...
The French equivalent, you mentioned her recently, and her predecessor, Fleur Pellerin. They mostly seem to make the link between the government and technology companies, including small startups and stuff like that. It’s an economy-focused position. This doesn’t seem to be your case.