[EM] Advanced Voting Systems: the Dirty Little Secret

[Abd ul-Rahman Lomax]

Well, we have a huge body of work examining the performance of voting 
systems under various conditions, but what may be the most common and 
most influential condition that real voting systems face, 
particularly in political applications, but also elsewhere.

Voter ignorance. Normal, non-reprehensible voter ignorance.

The models used in simulations don't take it into account, usually. 
The methods proposed as reforms may, in the presence of common voter 
ignorance, be no better than Plurality. The poster boy for this is 
IRV, of course, for nonpartisan elections.

"Voter ignorance" here means that the voter doesn't have sufficient 
knowledge of the candidates to rank or rate them all, except perhaps 
by donkey voting (as is claimed to take place in Australia where 
mandatory full ranking is required; the claim is somewhat 
questionable, because the Australian elections are partisan, and 
voters may not have heard of the candidate, but *have* heard of the 
party, and are likely to have an opinion about the party which then 
is easily attached to the candidate, it's right there on the ballot).

Lewis Carroll, in his booklet on what we now call Asset Voting, after 
Warren Smith, though it was earlier proposed as Candidate Proxy by 
Mike Ossipoff and Forest Simmons, noted that many voters would only 
know enough to vote for their favorite. (There are other causes for 
this besides ignorance, to be sure, but it is undoubtedly a strong 
effect.) Carroll was considering Single Transferable Vote, used to 
elect parliamentary seats. How could these "bullet voters" 
participate meaningfully, and be represented, should they happen to 
favor an eliminated candidate?

His solution just could make advanced voting systems moot, 
intellectual curiosities, unusual of application. Allow the first 
preference candidate on the ballot to "own" the votes, to be 
reassigned at the discretion of this candidate, "as if it were their 
own property." Smith used the "asset" metaphor, which is the same. 
Candidate proxy, though, is more descriptive, that's what it is, so 
here I will use that term.

Candidate proxy could be implemented within an STV system used for 
proportional representation. In a full application, where the entire 
assembly elected has one district, the entire jurisdiction it covers, 
and with easy registration of candidates, and with appropriate rules 
(Hare quota, probably, with special representation for the dregs), it 
would enable practically *perfect* proportional representation, 
limited only by the size of the assembly. Political parties would 
become almost moot (they would still exist for other purposes). 
Voters would vote, indeed, for a candidate they know and trust. Why 
vote for anyone else? Strategic voting: moot, useless.

I am not here describing how the electors or proxies (they are 
essentially public voters) would put together seats. It's a 
communication problem, and there are simple systems that would make 
it easy, but the *legal* system need not be concerned with this, but 
only with basic rules for registration of the recast votes, the 
actual formal election process of a seat, where named electors assign 
specified votes to the seat. (I'd recommend that they be specified by 
precinct that they come from, so voters will generally know where 
*their* vote went.) Many ballots could be used, to be sure, but there 
are better ways. This is a problem of voluntary coordination, more 
than it is an election problem.

Asset Voting, of course, would make runoffs easy, if one wished to 
insist on runoff processes. There are known, defined, electors, who 
vote publicly, so web voting becomes quite simple and the security 
not difficult; errors and hacks will be detected and fixed.

But Asset Voting is really a step into a whole new world, a world of 
representative/direct democracy, a world where public participation 
in every aspect of the political system becomes facilitated, 
effective, and voluntary. Want to vote in the assembly? Assembly 
votes are public, so you must become a public voter. Register as a 
candidate and vote for yourself. Even if you don't get another vote, 
you have a vote which, in a mature Asset system, can be cast on any 
matter before the Assembly. But it would be, I'm sure, you would 
find, quite inconvenient to routinely participate in that way. So, if 
you can't get proxies or whatever it takes from others, you'd pass 
your default right to vote on to someone with a seat, or at least 
someone holding a lot of votes who has, therefore, more motivation to vote.

But most people would not register as public voters, and most people 
who did would not actually vote except rarely, so most votes in the 
Assembly would be cast by those who hold seats.

The function of the seats is twofold: they have deliberative rights. 
They can rise and speak to the assembly. They can enter motions. They 
can also vote, but their votes are theoretically subject to 
deweighting depending on how many of their "constituent" electors 
vote. My prediction is that the direct voting would only rarely 
change results. The direct voting would take place in real time, 
i.e., electors would vote remotely, having access to full video of 
committees, and full documentation, the same as any seat. But I don't 
care to go too far into specifying details.

Electors would not merely serve, in a mature system, as appendages or 
election devices. They would become a penumbra around the assembly, a 
penumbra with real power, if they are functionally connected and can 
communicate directly with each other. They could recall seats. They 
could overrule the assembly; but such a thing would be rare, I'm 
sure, if it *ever* happens. When the people and their assembly are as 
closely connected as Asset Voting makes possible, there will be, I'd 
predict, far more tendency to seek consensus than we presently see. 
Those with seats would be dependent only on the continued trust of 
their constituents; their constituents would be represented, to them, 
by a manageable number of electors, who will advise them and who will 
be advised by them. Seats would not be dependent upon donations to be 
re-elected; the danger of bribery and graft doesn't disappear, except 
that its effect would be blunted. "Why did you vote for that project? 
Almost none of us want it!" "Uh, but ... it's really good for the 
economy. Yeah, the economy!" "John, that's bullshit and you know it. 
The only "economy" that it's good for is HugeCo. John? John? Are you 
there?" Suddenly a few electors, holding the right to vote for most 
of those involved in John's seat, take a very hard look at John's 
voting practices. They might ask for an investigation of contacts 
between John and HugeCo. They start voting routinely, and they 
initiate the recall process, which could be very quick. John gets 
some continued deliberative rights, perhaps, which is harmless, but 
loses most of his voting power right away. And all of it is removed, 
except for what he might hold directly from voters (i.e., as an 
elector himself), in short order.

The whole structure shifts when direct representatives of the people, 
freely chosen, can communicate directly with the representatives in 
an assembly. I don't think that John's abuse would be common at all, 
much less common than what is routine -- and even legal, if it 
involves campaign donations --; and it would be so risky that it 
would be cheaper and more effective for HugeCo to focus on 
efficiently serving the public, instead of trying leverage graft.

It can be done. The suggestion for how to have hybrid 
direct/representative democracy, even when the scale is very large, 
has been around for a long time. Direct democracy is still considered 
impossible by experts, but they have overlooked the hybrids, that get 
very, very close to direct democracy. It could be argued that Asset, 
fully implemented, *is* direct democracy, because anyone who cares to 
vote in the Assembly, to be a public voter, can do it. (Direct 
democracy with secret ballot is a very, very bad idea. The only 
functioning direct democracies that I know of, New England Town 
Meeting governments, have public voting at Town Meeting. Ancient 
direct democracies, likewise, had public voting, to participate, you 
had to attend the meeting and stand and be counted -- or shout, as in 
the Spartan implementation, which Smith calls Range Voting. Maybe.)

Back to the beginning of this post: Asset Voting harnesses and uses, 
to full effect, the ability of voters to determine a favorite.

Because votes aren't wasted, it becomes possible for that favorite to 
be much closer to the voters, and, I'd suggest, in a mature system, 
voters would not normally vote for someone they can't visit or call 
up and talk to. Would you? Why? I would argue that your vote is 
*more* powerful if assigned to someone you trust, and especially 
someone you can talk with.

I've encountered people who said, when told about this system or the 
similar Delegable Proxy, "Oh, I could never entrust my vote to 
anyone." Right. As if.

As if she ever had a choice. Already, she has people voting for her. 
On the town council or Board of Selectman, as the case happened to be 
for her, in the state senate, in the state House, in the U.S. House, 
in the U.S. Senate, people vote on her behalf. Or do they? Does she 
really know? Do they vote on her behalf, or on behalf of those who 
give political contributions. (Probably a bit of both....)

With Asset, she would periodically vote for the single person she 
trust most, of all those who make themselves available. She'd be 
assigning him or her voting power, one vote. And if she *really* 
"could never trust anyone with her vote," well, then, she can abstain 
from voting entirely, or she can register and vote as an elector. 
It's up to her. But I don't think she would have the present excuse 
that her vote doesn't count anyway. It would count, and how it 
counted would be visible. If she concludes that she made a mistake, 
she'll be able to fix it, next general election.

(It's possible to imagine continuous Asset, but I'm avoiding that for 
the moment, because of security issues. Traditional ballot security 
would be sufficient for secret ballot asset, where the canvassing is 
distributed and difficult to manage wholesale. Having some central 
web site where every voter gets an account.... I worry not only about 
hacking, but about "who watches the watcher?" The problem might be 
soluble. I just think we don't *need* to solve it.)