[EM] Concerns of KM & RF. Approval, Condorcet & ICT strategy. Reform schedule.

Dave Ketchum davek at clarityconnect.com
Thu May 17 16:58:47 PDT 2012


On May 17, 2012, at 2:09 AM, Michael Ossipoff wrote:

> Kristofer:
>
> You expressed concern about uncertainty about how to vote in  
> Approval. Let me re-word what I was trying to say about that:
>
> First, for simplicity let’s say that you belong to a faction that  
> all prefer and vote as you do. What you object to is that, in  
> Approval, you don’t know the way of voting by
> which your faction can get the best result possible. But what you  
> _do_ know (if you like and choose strategic voting) is the way of  
> voting that will maximize your expectation, based on what your  
> expectation already is. (I’ve already said much about the better- 
> than-expectation strategy of Approval)
>
> That’s good enough. You can’t expect to know exactly what ballot  
> marks will give the best outcome for you.

But the voting method will affect your proper thinking:

If Approval you are expressing equal liking for all that you vote for  
- thus should not vote for those liked too little.

If Condorcet you are expressing amount of liking via ranking and can  
include voting for all wanted if all liked better are not elected.
>
> I emphasize that, in Condorcet, you don’t know either of those  
> things, _even if it’s a u/a election_. Especially if it’s a u/a  
> election.
>
> And do you really think that our elections don’t have unacceptable  
> candidates who could win? …or two sets of candidates such that the  
> merit differences within
> the sets are negligible compared to the merit difference between the  
> sets?
>
> In contrast, Approval’s u/a strategy (as is its non-u/a strategy) is  
> not only known, but is the simplest there is: Approve (only) all the  
> acceptables.

That sounds simple - until I try to apply it to actual voting as to a  
candidate who I see as on the edge between acceptable and non-
.     Wrong if non- but would have deserved winning.
.     Wrong if voted for and wins without deserving.
>
> Yes, Condorcet has the consolation or compensation that, if the  
> election isn’t u/a, and if you don’t much care about the results,  
> then you can rank the candidates at as many rank positions as you  
> want to. But sincere ranking would be a big mistake in a u/a  
> election. And often in a non-u/a election too, if the result  
> matters. Condorcet’s supposed strategy-free-ness is only a  
> sometimes, maybe, thing. And when it isn’t that “sometime”, then you  
> _really_ don’t know what to do. Approval voting is incomparably  
> easier and simpler.

And this is simpler, for I can rank less-liked below all I like better.

> Richard expressed the concern that, if Approval were enacted, then  
> maybe people wouldn’t be willing to later change to something else,  
> and those who would like to go to something better wouldn’t have the  
> opportunity. In other words, if you don’t enact Condorcet instead of  
> Approval, before Approval, then you’ll never get an opportunity to  
> enact Condorcet.
>
> A valid concern. Valid to ask about, but not valid to be concerned  
> about.
>
> Forgive me for repeating this: If Approval were enacted, there would  
> be changes in government and society, such that the media would be  
> incomparably more free and open. Campaign laws and ballot-access  
> laws would be more fair. Political debates would be more inclusive.  
> These things would result from a government that is more what the  
> voters want. Also, for one thing, after the results of the first  
> Approval election, it would no longer be as easy to exclude non- 
> Republocrats from ballots, debates, editorial letters, articles,  
> news coverage, airtime, etc.

Enacting Approval or Condorcet, or most any other true improvement,  
would help as described above.  Would also include freer selection  
among candidates:
.     As always, vote against the worst of such as Republocrats.
.     Can also vote for the most desirable candidates.
>
> And it would be well-established that voting system improvement is  
> possible, because it would be an observed fact. People would be open  
> to it. People would know the subject of voting systems better than  
> they do now.
>
> The environment would be _much_ more favorable to rank-balloting  
> than it now is.
>
> Now, make no mistake: Condorcet, or rank-methods in general won’t be  
> easy, any time. But if they’re ever do-able at all, then it will be  
> when Approval has started the improvement and shown that there is  
> such a thing as good voting system reform.   …and shown the  
> numerical importance of non-Republocrat candidates, voters and  
> parties.
>
> So no, getting a fancier method won’t be harder after Approval.  
> That’s when it will be possible, if it ever will.
>
> Any improvement would be a step. That’s the spirit of the  
> Declaration. That’s why none of us should oppose eachother’s  
> proposals in any publication or forum other than here at EM.  
> Fighting eachother and opposing eachother’s proposals in public  
> would be counterproductive, a ruinous hindrance to reform.
>
> Does it sound self-serving when I say that Approval should be the  
> first step, the first proposal? I’ve told many times why Approval is  
> so much more proposable,  acceptable, enactable. Approval is  
> literally the minimal change from Plurality that will get rid of  
> Plurality’s big fault. Approval is nothing more than the repeal of  
> Plurality’s ridiculous rule, the rule that makes Plurality a  
> blatantly, bizarrely undemocratic points rating system.
>
> The desirability of that change is so obvious, as is the fact that  
> it’s an improvement and nothing other than an improvement, that  
> Approval  might even be enacted (Plurality’s funny rule repealed)  
> via a voting rights court ruling.
>
> Even if you think that Condorcet is enactable, you must agree that  
> Approval’s expected enactment time is much less than that of  
> Condorcet. Do you really think that there’s a reason to do without  
> reform for a longer time?
>
> You want something more than Approval. So do I. I’ve been saying  
> that all along. I want defection-resistance. It can be had in ICT,  
> and probably in Kristofer’s CC/ABE method proposals (but of course  
> they don’t meet FBC, but ICT does).
>
> So, when Approval has made genuine open discussion possible, you can  
> propose Condorcet, and I’ll propose ICT. You don’t seem me proposing  
> ICT now, do you? It would be something to ask for later. Though I’d  
> prefer ICT, I’m now only suggesting to propose Approval, because  
> that’s what’s do-able.
>
> When someone said that  favorite-burial isn’t a technically-valid  
> strategy in Condorcet, I replied that favorite-burial is indeed a  
> technically optimal strategy in Condorcet, if it’s a u/a election,  
> and if it seems like Condorcet is the only acceptable who can beat  
> the unacceptables.

Said more clearly and correctly, rank all that you want considered  
acceptable (this is above all unranked).

Rank your favorite, if you have one, at top - to be seen as most- 
wanted.  Rank by desirability, not by what follows here.
>
> If you have no idea which acceptable can beat the unacceptables,  
> then top-rank them all.
>
> If it’s somewhere in between, as will usually be the case, then you  
> don’t know what to do. You don’t know what ballot ranking would give  
> your faction its best result. You don’t even know what ballot  
> ranking would maximize your expectation. No one knows, in Condorcet.
>
> What about ICT? Completely different ballgame. There is no strategic  
> need for favorite-burial, in u/a or non-u/a. Therefore you won’t be  
> wondering whether your best u/a strategy is Compromise alone at top,  
> or all acceptables at top.  Vote all of the acceptables at top.

That is not what I read about ICT:
      Other Condorcet counts for each pair how many voted A>B and how  
many B>A, but not A=B.
      ICT also counts how many A=B - maximized if all acceptables  
voted at same rank.
>
> Ok, I admit that I don’t know whether there’s advantage, in a u/a  
> election, to rank the unacceptables with the more winnable ones  
> nearest the bottom. I would expect that your treatment of the  
> unacceptables should be either that, or not ranking them at all.   
> Are rank methods complicated contraptions or what! But I still like  
> ICT.

This reads as ranking the unacceptables, but low enough to probably  
not win.
>
> But, in any case, you won’t have the lack of known strategy that  
> you’d have in Condorcet.
>
> I’ve always said that you can’t really discuss voting systems in a  
> vacuum. But I’ll call the candidates in this example, A, B, C, D, E,  
> and F.   But, fill in, for yourself, _actual_ candidates for them to  
> represent, so that these situations aren’t being discussed in a  
> vacuum.
>
> F is the unacceptable. There could be more, but it’s simpler, and  
> just as good, to have just one.

Seems simpler to simply leave Fs unranked.
>
> A , B, C, D, and E are the acceptables. A is your favorite. But you  
> like B, C, D and E a lot too.
>
> Oh, one other thing, you don’t like E as much as B, C, and D.  E has  
> some very good policy proposals, but is also somewhat lacking. Just  
> to give you an example, E’s platform says that we should adequately  
> fund NPR (National Public Radio) because of its excellent news and  
> commentary  :-|   :-)
>
> But I’m not naming any names.
>
> So, it’s u/a. How would you vote in Approval? Of course you’d  
> approve A, B, C, D, and E.
>
> But maybe that could be complicated by a C/D problem. Maybe the E  
> people aren’t as co-operative as you are, and maybe everything that  
> they’ve said indicates that they will not vote for anyone but E,  
> because they want E to win instead of A, B, C, or D.
>
> I should look up and re-post my 5 suggestions for dealing with C/D  
> in Approval. It needn’t be a problem. For now, I’ll just say that  
> some of those 5 solutions might call for not approving E, depending  
> on the circumstances.
>
> C/D is solvable in Approval, but it definitely involves strategy   
> (as it would in Condorcet too).
>
> What if the method is ICT? If there isn’t any C/D problem, then I  
> suggest that the optimal u/a strategy is to vote as in Approval. As  
> I said, that makes ICT voting a lot simpler and easier than  
> Condorcet voting. Just top-rank A, B, C, D, and E.

Telling the counters you have equal desire for all?

DWK
>
> Since F is the only unacceptable, you’re done with your ranking.
>
> (If there’s been more than one unacceptable, then maybe you should  
> rank them below top, in reverse order of their (guessed) ability to  
> beat an acceptable).
>
> But what if there’s a C/D situation? Rank E in second place, instead  
> of at top. Because of ICT’s defection-resistance, it’s pretty safe  
> to do that.
>
> But of course, in a u/a election, it’s really bad strategy to not  
> top-rate all of the acceptables. And the strategy I’ve suggested  
> here is to downrank E. That constitutes a little bit of a C/D  
> problem. I’ll call it a u/a C/D problem. It’s much less than  
> Approval’s or Condorcet’s C/D problem, because it’s unlikely that F  
> will win because you ranked E in second place.
>
> ICT’s  u/a C/D problem is a minor one.
>
> Mike Ossipoff


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20120517/6c0410aa/attachment-0002.htm>


More information about the Election-Methods mailing list