[EM] Staggering elections (was: Re: Holding byelections with PR-STV)

[Kristofer Munsterhjelm]

Raph Frank wrote:
> On Thu, Sep 10, 2009 at 9:22 PM, Juho <juho4880 at yahoo.co.uk> wrote:
>> The hysteresis function may increase the strategic opportunities since
>> voters could trust that old representatives will be elected in any case and
>> they could try free riding. But in real life small hysteresis may well not
>> be too problematic.
> 
> Maybe you could play around with the quota.  For example, the Droop
> quota could be used for sitting representatives and the Hare quota for
> everyone else.
> 
> Staggering of elections, so that there isn't a single election day,
> may or may not be a good thing.  Would it mean that the government is
> always in campaign mode or would it never be in campaign mode.

One idea of mine, with regards to governments (executives) is this: if 
the government has been in power for n time units, the people can 
(through initiative-type methods) propose a recall which succeeds if a 
fraction f(n) of the voters agree with it. If they do, a new election is 
held. Then set f(n) to a function that's 1 at 0 time units and 0 (new 
election no matter what) at some predefined constant time x.

Thus, a popular president or governor would last longer because a 
smaller minority would agree with the recall, and thus one has to wait 
until f(n) passes below that point. Yet he's relatively safe at the 
beginning of his term, so (hopefully) the parties wouldn't constantly be 
in election mode.

To protect against brute populism or ever-changing governments, the 
function f might decay more slowly than linear for the first half of its 
shape. An ordinary reelection scheme would be equivalent to one where 
f(n) is 1 all the way until x, when it drops to 0.

Possible problems with this idea: unstable governments (could perhaps be 
fixed by "constructive motion of no confidence", i.e. that the people 
have to prove they won't do the same thing to the person they elect the 
next time) and "neverending referenda" as minority groups try again and 
again to get the recall to get above f(n) support. The latter could be 
fixed by having a grace period since the last proposal for recall, so 
that, for instance, if a recall fails, one has to wait at least two 
months (or half a year, or whatnot) before trying again. However, then 
the supporters of the incumbent can strategize by running a recall when 
the government's support is just above f(n) so as to block an early recall.


The same idea could be applied to laws with automatic sunsets. Popular 
laws stay on the books for a longer time than unpopular ones before they 
have to be re-approved. An advantage of the automatic sunset, with or 
without the gradual decay mentioned above, is that pointless laws (of 
the sort like "you can't drive a car unless you have a person next to 
you holding a flag so as not to scare the horses away") would quickly 
disappear once no longer relevant.