[HN Gopher] Will the next German parliament have a gigantic size...
___________________________________________________________________
Will the next German parliament have a gigantic size? A law and
coding challenge
Author : unripe_syntax
Score : 68 points
Date : 2021-09-24 09:57 UTC (13 hours ago)
(HTM) web link (www.r-bloggers.com)
(TXT) w3m dump (www.r-bloggers.com)
| jonchang wrote:
| This should be updated to link to the original website, rather
| than to R-bloggers, which is a blog aggregator.
|
| http://skranz.github.io/r/2021/09/24/bundestag2021.html
| Denvercoder9 wrote:
| Is there any electoral system that has both representatives for
| districts and fully proportional representation that doesn't have
| this problem? It seems this kind of problem is inherent in
| combining the two.
| lambertsimnel wrote:
| I believe localized-list proportional representation avoids the
| problem of variable-sized and potentially very large
| legislatures, while maintaining fully proportional
| representation and representatives for districts:
| https://en.m.wikipedia.org/wiki/Localized_list
|
| As far as I can tell:
|
| 1) each voter expresses a single preference for a candidate in
| a single-member district
|
| 2) when the votes are counted, each party is allocated a share
| of seats proportional to its candidates' votes combined
|
| 3) that many of its candidates are allocated their districts,
| such that they are the party's best performers (not necessarily
| plurality winners in their own districts)
|
| I believe biproportional apportionment can also satisfy these
| criteria:
| https://en.m.wikipedia.org/wiki/Biproportional_apportionment
|
| I'm not sure these are the only systems that satisfy these
| criteria. Scandinavian-style systems have a broadly
| proportional district system and a small proportion of
| "levelling" seats to make the results more proportional. I
| don't have detailed knowledge of any Scandinavian system, but I
| believe they could be tweaked to be more proportional and/or
| localized, and levelling seats could be allocated to districts
| according to where the parties winning levelling seats are most
| under-represented. This is more of an approximation than the
| preceding two, but:
|
| 1) the district system for German federal elections is very far
| from proportional, so the other part of the system has more to
| compensate for than in scandinavian-style systems
|
| 2) proportionality is necessarily approximate under any system
| where each party has an integer number of representatives, with
| achieving outcomes "within quota" being the best that can be
| hoped for
|
| 3) most proportional systems don't even produce results within
| quota, with the Dutch system a notable exception
|
| My impression isn't that the variable number of seats in German
| federal elections is intended to limit accidental
| disproportionality, but that it's intended to prevent gaming
| the system by splitting a party into a district party and a
| list party, in order to be allocated list seats even if
| overrepresented in district seats. I doubt would that kind of
| manipulation would work if each district was multi-member with
| proportional representation.
| inglor_cz wrote:
| One of the weaknesses of the German system is that people cannot
| express their preference on the Liste. So whatever sequence of
| the Liste candidates the parties decide, in that sequence they
| obtain the mandates.
|
| In Czechia, you can circle up to four candidates on the list as
| your preference. As a result, several important politicians flew
| out of the parliament even when they were #1 or #2, because
| people from lower ranks outcompeted them.
|
| Parties hate this one weird trick... but fortunately cannot agree
| on abolishing it.
| jszymborski wrote:
| This seems great to me, but does this make the ballot more
| confusing? That's a large list of candidates, no?
| inglor_cz wrote:
| You have a set of ballots, each with a list for a single
| party. You throw away all the ballots save the one that is
| your choice, then make your preference circles there.
|
| They are recycled, so no waste. (The politicans are often
| recycled too, some having 4-5 party affiliations in their
| CV.)
| Dagonfly wrote:
| We have a similar system in Germany for local elections. In
| cities it can get crazy. Here is a a ballot from Frankfurt: h
| ttps://www.fr.de/bilder/2021/02/11/90200944/25026648-28-lis..
| .
| AnssiH wrote:
| Finland also has voter-decided party list order, but here you
| still vote for a single candidate - the party list gets
| ordered according to votes received by individual members.
|
| Seats are distributed to party lists according to their vote
| share, and then within a party list according to the
| individual vote amounts.
|
| And you just write the candidate number on the ballot (see
| e.g. photo in this news article: https://yle.fi/uutiset/osast
| o/news/yle_news_explains_voting_...), so it is quite simple.
| The voting booth has a list of candidate numbers if you don't
| know yours.
| speeder wrote:
| Here in Brazil it works this way and unfortunately we found
| out it is kind of bad idea.
|
| We had a couple corrupt parties convince inexperienced
| people that were wildly popular to run for them, these
| people (for example one guy is a circus clown that wrote
| some music that went viral) then get massive amounts of
| votes, and drag with them to the parliament all the corrupt
| politicians.
| karatinversion wrote:
| Sounds like proportional representation working as
| intended?
| barbazoo wrote:
| Having had to experience the Canadian federal election this week
| and seeing a resulting parliament that has very little to do with
| what the people actually voted for, I yearn for Germany's system
| and it's kind of challenges.
|
| Monday's election resulted in 5.9% of people being completely
| unrepresented in parliament and the now (again) governing
| Liberals representing 32.6% of the populace, even less than it's
| Conservative competitor [0]. Sure they need the NDP for a
| majority vote but it's nothing like the coalitions in a
| Government like Germany where parties come up with a formal
| coalition agreement.
|
| Seeing how people's research leading up to the election consisted
| of figuring out how to vote strategically for the opposing party
| in their riding to be defeated, instead of comparing party
| platforms and deciding who they agree with instead, made me want
| to barf honestly. There is a significant chance that we'll just
| get called to another election a year or 18 months from now, for
| what, another $600m minor shuffle of seats [1]?
|
| [0]
| https://newsinteractives.cbc.ca/elections/federal/2021/resul...
|
| [1]
| https://elections.ca/content.aspx?section=res&dir=rep/off/co...
| adsche wrote:
| Tangential point but Germany's parliament has a 5% public vote
| threshold for parties to enter parliament (some exceptions
| apply).
|
| According to current polls, that means between of 6%-9% of
| votes (for minor parties) will end up without parliamentary
| representation (5% at the last federal election).
|
| NB: The number of not represented people is of course much
| higher, even discounting non-voters and underage citizens,
| since most(all?) foreign residents don't have the right to vote
| in federal elections.
| barbazoo wrote:
| That's true and probably exacerbated by the huge number of
| political parties. 5.9% in Canada is not that bad then maybe?
| I don't know.
|
| Same applies here, only citizens of age are allowed to vote.
| adsche wrote:
| I don't know either, therefore I considered it a pretty
| tangential point. It _feels_ like the choice is a bit more
| diverse here because there are six(+1*) parties all but
| guaranteed to enter parliament, so you wouldn 't waste a
| vote if you voted for any of them.
|
| *) seven actually, one party (CSU) only standing in Bavaria
| instead of the conservative party CDU.
| vkou wrote:
| > Sure they need the NDP for a majority vote but it's nothing
| like the coalitions in a Government like Germany where parties
| come up with a formal coalition agreement.
|
| Coalition agreements are 'formal' in the sense that they last
| until they get broken. And they get broken whenever it is
| politically convenient for the participants therein. They are
| alliances of convenience, not suicide pacts.
|
| The libs will now govern as long as they have support of at
| least one other party in parliament. It's not that different
| from a formal coalition.
|
| Trudeau promising electoral reform in his first election, and
| delivering not a damn thing is the real travesty, here. Had he
| done that, the country would be ran by a Lib-NDP coalition from
| now until the end of time.
| neals wrote:
| So what's the maximum amount of seats that would be possible?
| dependsontheq wrote:
| If one party won all districts 299, and had 0% of the party
| vote. You would come up with 897 seats in the Bundestag...
| which is not far from the expected seat number new week.
|
| So it is pretty much a worst case already... I have some
| friends who might become Members of the Bundestag because of
| this, but even they are kind of irritated by that and know that
| nobody they will be talking to about their new seat will be
| happy about it.
| adsche wrote:
| Hm, as I understand it, if one party won all 299 direct
| mandates at 0% 2nd vote share, the other parties' seats would
| have to be increased until 299 == 0%. Depending on how you
| round, that would be at least 30,000 seats (there's some
| tolerance allowed that I did not consider here but that's the
| ballpark for that unlikely scenario).
|
| The linked site lists a (more realistic) scenario already,
| where the number of seats comes out to more than 900. How did
| you end up at your 897 seat upper bound?
| adsche wrote:
| Actually, I should add that even in the unlikely extreme
| scenario, the upper bound would be the number of candidates
| on the parties' lists (Landeslisten). Even if a party would
| have a right to more seats, they cannot fill more than they
| have approved candidates on their lists. One could
| determine this actual upper bound by looking at each
| state's lists, which are public of course, but I haven't
| quickly found them compiled in one place, so I won't
| bother.
|
| (Berlin's pirate party once won a surprisingly high share
| in the state parliament but wasn't able to fill all seats,
| because they were polling way lower when compiling their
| list.)
| Dagonfly wrote:
| Apparently, this might actually happen this time.
| (https://twitter.com/mneunho/status/1435887122649686024).
|
| If the Bundestag becomes as massive as predicted (800+),
| the CDU and AfD lists are too short.
| Dunedan wrote:
| > (Berlin's pirate party once won a surprisingly high
| share in the state parliament but wasn't able to fill all
| seats, because they were polling way lower when compiling
| their list.)
|
| That's not correct. They had 15 candidates and won 15
| seats.
|
| For all the nitty-gritty details check paragraph 4.5 on
| page 67 of the official election results [1].
|
| [1]: https://www.berlin.de/wahlen/historie/berliner-
| wahlen/ergebn...
| ThePadawan wrote:
| As a funny aside - if the current Bundestag is 709 seats...
| where the hell are they planning to fit all these chairs?
|
| (I know that lots of politicians never or rarely show up to
| parliament anyway, but I think the idea is that they _have_ a
| fixed seat such that it becomes clear which segment of the
| parliament is physically taken up by which party)
| sbierwagen wrote:
| http://europedirect.cut.ac.cy/wp-
| content/uploads/sites/4/201...
|
| Fill in the visitor's balcony, I guess.
| ThePadawan wrote:
| It's been a while since I visited in school, but IIRC the
| whole balcony is in a different security area. Also, I'm
| not sure the press would appreciate getting uninvited
| either.
| sbierwagen wrote:
| Reroute the security zones. Tell the press to quit
| crying.
|
| What's the other option, fill in the aisles? Given the
| frequency of terrorist attacks in Europe, preserving the
| ability of lawmakers to quickly leave a room seems more
| important than the feelings of journalists.
| chki wrote:
| Search for "Bundesversammlung" (the constitutional body
| that elects the German president) to see some pictures of
| the Bundestag with 1260 seats. It's pretty crowded.. but
| not impossible. Although I doubt that they would have all
| these chairs permanently in the Bundestag.
| ThePadawan wrote:
| Wow, TIL! Thanks!
|
| It's definitely a crowded house, and much more of a
| temporary arrangement (it looks like they had to take out
| most of the desks and swivel chairs)!
| chki wrote:
| That's not correct, the upper limit is much higher. If you
| win 299 districts and get 0.1% of the party vote, the
| Bundestag needs enough seats to properly represent 99.9% of
| the other parties. So you basically (it's somewhat more
| complicated in practice) need to multiply 299 by 999 to get
| the proper amount of seats..
| Dagonfly wrote:
| An mathematical perspective on this problem is if you consider
| the German system as a transformation from a FPTP to a Popular
| Vote system:
|
| Let's say you have a FPTP system with a total of D direct seats.
| To evaluate how distorted the FPTP vote is, you run a PV election
| in parallel. Let's say a party wins 'd' out of the 'D' FPTP
| seats. The party is over-represented if d/D >
| PV_percentage
|
| How do you fix this? Step 1: Introduce a scale factor 's' that
| increases the number of seats. You choose a minimal s > 1, such
| that for all parties: d/(s*D) <= PV_percentage
|
| In particular, for the most over-represented party the less-
| equals will become an equals. Step 2: Award additional seats 'x'
| to all other parties, until the invariant becomes equal for all
| of them: (d+x)/(s*D) == PV_percentage
|
| Perfect! Now, in the German system the 's' is additive (s'+D)
| rather than multiplicative (s*D). But you can always transform:
| s*D == ((s'/D)+1)*D == s'+D
|
| Originally, the system set s' to a fixed s'=D (implying s=2).
| With the rationale that parliament ends up with half FPTP seats
| and half additional (list) mandates. But, of course, then you
| can't scale past s=2 in Step 1 and end up with
| "Uberhangmandanten". Since 2011 the new constraint is s>=2,
| making the size of parliament variable.
|
| This year, the CSU is predicted to get around 43 (out of 299)
| direct seats, but less then 5% of the popular vote. So:
| 43/(s*299) == 0.05 // s=2.88 and sizeof(parliament)=s*299=861
|
| ---
|
| Imo, the solution would be to reduce D (the number of direct
| seats) and increase the number of list seats (s'). For example:
| D=200 and s'=400 (implying s>=3), which leaves more headroom for
| over-represented parties. Another lever would be to change from
| FPTP to approval voting as froh said.
| earthboundkid wrote:
| The US wrote the German constitution. We should just drop any
| clauses that are Germany-specific (the ones about reunification
| and so on) and adopt it wholesale as our constitution. Our
| current constitution is very bad and leading us slowly into
| dysfunction and possible civil conflict...
| chki wrote:
| No, the US did not write the German constitution. It was
| written by German politicians and lawyers. The Allied forces
| definitely had an influence on the constitution and requested
| certain things to be in the constitution (human rights,
| federalism). But the constitution is definitely a properly
| "self-made" document (building on existing constitutions in
| other democratic states and the Weimarer Reichsverfassung).
|
| Edit: also, it would be absolutely impossible to "simply take
| the German constitution and apply it to the US". To somebody
| who knows a bit about constitutional law that's like saying
| "the computer isn't working, let's just use this bicycle
| instead."
| carschno wrote:
| How do you come to that assumption?
| https://en.wikipedia.org/wiki/Parlamentarischer_Rat
| inglor_cz wrote:
| Germans wrote their own _Grundgesetz_ , it only had to be
| approved by the Allies.
|
| Are you possibly confusing Germany and Japan? Japanese
| constitution was mostly written by Americans, though with some
| review and input from the Japanese, and with a lot of respect
| to the previous Meiji constitution.
| froh wrote:
| I don't get why the district seat doesn't simply use approval
| voting. If I look at my local district, Nuremberg North, the
| relative majority with fptp is always with the CSU. However the
| more or less left leaning parties, SPD, Grune, Linke always have
| a combined solid majority over the neolib spectrum FW, CSU, FDP.
|
| With approval voting, my educated bet would be a higher approval
| for one of SPD or Grune than for any in the
| neoliberal/conservative spectrum.
|
| Same goes for many districts which produce paradoxical overhang
| seats, in both directions, like SPD in NRW, and CSU in Bavaria
| districts.
| adsche wrote:
| > the relative majority with fptp is always with the CSU.
|
| I feel like this is part of the reason. CDU/CSU currently seem
| to profit most from this system and they have been governing
| for 16 years. Before that, SPD got the most direct mandates and
| was governing.
|
| Even if the governing _parties_ wanted a change, it would
| probably be hard to convince their MPs to vote for a law that
| would endanger their own re-election.
| jhbbp wrote:
| Interesting system. How are districts boundaries defined?
| adsche wrote:
| They were defined by law. They are supposed to have roughly
| equal populations and must be re-assigned if the difference in
| population exceeds 25% (I don't know what that means in
| practice, e.g., what the reference is.)
|
| Here's a list:
| https://en.wikipedia.org/wiki/List_of_Bundestag_constituenci...
| Dagonfly wrote:
| They start "re-districting" if a district deviates from the
| average population by +-15%.
|
| The new districts are proposed by an independent commission
| and have to be approved by parliament. Here is the proposal
| for 2021 (pdf warning): https://www.bundeswahlleiter.de/dam/j
| cr/b1752d5e-aecd-4cff-9...
|
| And here is the decision by parliament, I guess:
| https://dip.bundestag.de/vorgang/.../261890
___________________________________________________________________
(page generated 2021-09-24 23:02 UTC)