[EM] Possibly more stable consensus government

[Kristofer Munsterhjelm]

On 11/29/2012 09:02 PM, Raph Frank wrote:
> On Tue, Nov 27, 2012 at 9:16 PM, Kristofer Munsterhjelm
>> However, if you need supermajority support for decisions, then you have to
>> have something to put in place when the supermajority support isn't there.
>
> One option is to select 2 PMs.  That is what they do in Northern Ireland.
>
> The cabinet is decided by the d'Hondt method (so proportional) and
> there is 1 PM (actually first minister) from each community.
>
>   So, the vote would work something like
>
>   - vote for PM (including cabinet) combination
>   -- if a candidate gets 60%, he is appointed PM, finish
>
>   - Round 2
>   -- Anyone with more than 1/3 of the vote gets nominated as joint PM
>   -- Keep voting until 2 get 1/3 or more
>   -- If that fails, then if 1 gets 1/3, he can take office, as half a PM
>   (maybe have previous PM as other one)
>   -- Each PM appoints half of the office
>   --- The PM who got the most votes has to option to go first or second
>   --- Each picks a department alternatively
>   -- Department of finance might be different
>
> You could have more departments than cabinet positions.  Each PM gets
> to appoint half the seats to anyone he likes, and then can assign any
> departments he picked any way he likes.
>
> The more departments, the more even the balance of power between the 2 PMs.
>
> You could also split them based on the relative support of the 2 PMs,
> but that would mean constant adjustment as support goes up an down.
>
> Each PM would require 1/3 support to stay in office (voting for both
> would count as 1/2 a vote each)
>
> It might also be required that both submit  their cabinet member
> choices and if either can't get 1/3 support, they are considered to
> have lost confidence.

I see. That's a third option, then: you distill, to use such a term, the 
lines of disagreement or representation blocs into the executive, so 
that the executive has to find consensus rather than having to wait on 
the legislature to do so.

That might work in combination with the idea of Simmons. You could have 
a vote where you ask the members of the assembly for their favorite as 
well as their consensus choice. If the consensus candidate gets more 
than the threshold (say 60%), he gets the task of appointing the other 
ministers, otherwise some PR method is used to elect a small number 
(perhaps only two) "joint PMs".

That sounds better than having a PM chosen by random ballot when the 
consensus choice fails; but the PR method would have to be probabilistic 
to be strategy-proof, I think.

>> So a supermajority requirement upon forming the government and a minority
>> for a vote of no confidence would be a recipe for instability (and probably
>> rule by the bureaucracy).
>
> I was thinking 50% to form after an election and 60% to vote no confidence.

Yes. I'm just saying that 60% to form and <=50% for no confidence would 
definitely not work.

> Another option is that if no government is formed by 60%, the old one
> stays in power and a new election is automatically triggered within 30
> days.
>
> After that election, if nobody has 60%, then 50% is sufficient, but
> maybe if that happens the term is reduced by 50%.
>
> No matter how the government is picked, 60% would be required to
> replace it with a different one.

That would provide an incentive for the slight majority to hold out for 
an election, so that they can reaffirm their slight majority and then 
get through on a 50%. I do see the point, though, because a very slight 
majority couldn't be sure they would stay a majority after the election.

> 1/3: new election
> 2/3: Each legislator nominates a candidate and then a random
> legislator is picked and his choice wins[*]
>
>   [*] could use something like IRV to eliminate very small options (say<  20%)

Perhaps something like multistage Hay voting ( 
http://www.panix.com/~tehom/essays/hay-extended.html ) could be used to 
remove clones while keeping the method strategy-proof, also.

The mathematics is a little too tough for me, so I don't know if one 
could remove very small options in multistage Hay without upsetting the 
resistance to strategy.

> The problem with requiring 60% to take down the government, means you
> have to swing 20% of the house to cause a collapse.  That is a shift
> of power to the executive.

And secondarily, to the faction that managed to get their government 
through, yes. In more general terms: a 60% barrier to no-confidence 
favors the status quo because the status quo can survive on less (40%) 
than any of the alternatives.