Breakdown of a wormhole vote part 2

Today's post is in response to Gevlon's post about the CSM; specifically exploring the exclusion rounds of the wormhole candidates : Cipreh and Ayeson.  This is also a follow up on yesterday's post.

Tldr;

• Random leakage normally doesn't matter, unless someone is already close to a quota
• Preferences clearly flowing in a direction do matter.
• Leakage happens when voting 'below the line'.  
• The wormhole 5 got the seats their vote allowed.

The analysis below pays no attention to where the votes originated from, merely looking at the result of these exclusions.  I also have not spent any time pulling apart the numbers on the null bloc candidates.


First of all I grabbed the votes from Trebor's analysis, grabbing the end of rounds (before and after) any two candidates are excluded.

I then look at the change in votes for all candidates as the candidate is excluded, both in raw terms and then in % of votes terms.

Numbers are available as either (crudely formatted) HTML below, or a google doc.

Exclusion of Cipreh

After this exclusion, 74% of Cipreh's votes remained within the wormhole group.

If you were to treat the WH group as a political party, they would prefer to retain more votes; but groups are leaking votes all over the place for many candidates.

The good news for the wormhole candidate is that the votes leak without strong preference flows to any set of candidates, with the 'preferred' non-wormhole candidate (Roc Weiler) only getting 0.9% benefit. 

I note that Ayeson,  got the least benefit (2.9%) of  the wormhole candidates from Cipreh's exclusion, and James Arget received the most benefit (10.2%)

Exclusion of Ayeson

After this exclusion, 58% of Ayeson's votes remained within the wormhole group.

This shows that Ayeson's supporters have a majority preference for other wormhole candidates, but only just a majority.

However, again the leakage is largely random with the preferred non wormhole candidate (Ali Aras) picking up 1.4%.   This compares to Nathan's 2.9% boost, and James 13.6% boost.

Common items from both exclusions

There is a high correlation between the strength of each candidate and the benefit they get from random leakage.  If you had a high vote before the exclusion, then you were likely to get more votes of the leakage.

When would this have mattered?

For this election, the random leakage would have only mattered if there was already a 'rival' candidate close to a quota. For example, if the quota was 2260 at the exclusion of Ayeson, then Ali would have been elected at that round.  However if a rival candidate was this close, they would have most likely been elected anyway.

In real life, I have however once seen this kind of leakage deliver a seat to a 'hostile' party.  It can happen, but not normally.

Below the line voting

This election was what we would call in Australia, 'below the line'. Read http://www.aec.gov.au/voting/How_to_vote/Voting_Senate.htm for a description of 'above the line' or 'below the line'.

In CSM 8, voters may have been given instructions on how to vote, but they had to select their their own candidates.  There will be leakage under these circumstances, but in the two excluded candidates I have looked at, there was no advantage to any candidate outside the wormhole group.

Now, if CSM bring in a 'simpler' form of voting, where, while notionally, votes are still STV, but you delegate the allocation of preferences to a group ticket ('above the line'), the leakage outside the ticket will be almost negligible.  In Australia, where the choice is given, the more candidates that someone has to vote for, the more likely they are to vote above the line, or not simply not cast a valid vote.

Leakage applies in both directions, both in and out.  Reduce one, and the other is generally reduced as well.

Real life comparison.

Even in real life, voters are notorious for not following ticket instructions.  Parties in Australia often will advise on who to send preferences to.  For 'below the line' elections, these are nearly as often disregarded as observed.  Party instructions matter, but voters will do their own thing.

Gevlon's claim that the WH group could get 3 candidates.

The wormhole group had less than 2 quota's of first preference votes. For them to have received a third seat would have taken an incredible flow from non-wh candidates, requiring over a quota of leaked votes from elsewhere.  While observations on previous elections are not laws to be followed, it would have been truly remarkable for that to occur.  Getting as close as they did was remarkable enough.

Initially, the wormhole group had 1.9 quotas; when Nathan Jameson was eliminated (the last eliminated candidate)  they had 2.4 quotas (or 1482 votes past the quota to elect the 2 seats they got)  The next candidate (Mike Azaria) had 0.9 quotas (or 2885 votes).  For the wormhole candidates to get the third spot, they have to get their third vote in front of Mike.  That is, pick up another 1404 votes.

There was a leakage from Cipreh of 154 votes, and a leakage for Ayeson of 311 votes; totalling 465 votes.  If the wormhole block had kept every single vote on their ticket, they still would have missed out on their third seat by 939 votes.

In my opinion, the wormhole candidates simply did not have the first preference votes to support a third seat, this time around, and had more than enough first preference votes to get the second seat.

[edit: see Gevlon's clarification below.  I may have misrepresented his position]

Table of results. (also available as a google doc)

	candidate	With Cipreh	Exclude Cipreh	Change	%change		With Ayeson	Exclude Ayeson	change	%change
	Ali Aras	2171	2178.64	7.64	0.35%		2211.83	2242.12	30.29	1.37%
	Artctura	709.85	712.9	3.05	0.43%		0	0	0
	Awol Aurix	670.11	670.19	0.08	0.01%		0	0	0
	Ayeson	716.41	737.46	21.05	2.94%		744.104	0	-744.1
	Banlish	1228.24	1229.32	1.08	0.09%		1626.97	1638.72	11.75	0.72%
	Chitsa Jason	1690.17	1847.8	157.63	9.33%		1862.99	1956.9	93.91	5.04%
	Cipreh	595.27	0	-595.27			0	0	0
	Corebloodbrothers	1498.03	1500.13	2.1	0.14%		1509.01	1510.42	1.41	0.09%
	Greene Lee	1688.69	1691.96	3.27	0.19%		1706.04	1718.71	12.67	0.74%
	James Arget	1667.35	1837.76	170.41	10.22%		1856.72	2108.31	251.59	13.55%
	Kaleb Rysode	608.22	610.03	1.81	0.3%		0	0	0
	Kesper North	2955.75	2960.49	4.74	0.16%		3305	3300	-5
	Korvin	2079.91	2084.01	4.1	0.2%		2096.63	2107.26	10.63	0.51%
	Malcanis	1788.24	1800.79	12.55	0.7%		1833.78	1852.72	18.94	1.03%
	Mangala Solaris	2787.69	2802.88	15.19	0.54%		2876.05	2905.38	29.33	1.02%
	Mike Azariah	1811.84	1814.77	2.93	0.16%		1840.66	1854.17	13.51	0.73%
	mynnna	3313	3309	-4			3305	3300	-5
	Nathan Jameson	1652.24	1744.18	91.94	5.56%		1770.37	1857.7	87.33	4.93%
	progodlegend	2357.46	2365.77	8.31	0.35%		2393.28	2401.29	8.01	0.33%
	PsychoBitch	713.9	718.96	5.06	0.71%		751.917	758.533	6.62	0.88%
	Psychotic Monk	1560.43	1566.7	6.27	0.4%		1599.92	1617.52	17.6	1.1%
	Ripard Teg	3313	3309	-4			3305	3300	-5
	riverini	900.26	904.75	4.49	0.5%		932.135	943.285	11.15	1.2%
	Roc Wieler	833.54	840.96	7.42	0.89%		863.783	874.074	10.29	1.19%
	Sala Cameron	1563.76	1564.84	1.08	0.07%		1892.33	1913.65	21.32	1.13%
	Sort Dragon	3299.07	3301.21	2.14	0.06%		3305	3300	-5
	Steve Ronuken	1307.12	1308.31	1.19	0.09%		1326.06	1337.05	10.99	0.83%
	Travis Musgrat	966.43	966.68	0.26	0.03%		979.716	980.886	1.17	0.12%
	Trebor Daehdoow	2187.45	2204.57	17.12	0.78%		2242.12	2258.39	16.27	0.73%
	Unforgiven Storm	970.69	973.11	2.42	0.25%		1298.45	1311.15	12.7	0.98%
	wormhole total	6321.45	6167.2	-154.24			6234.18	5922.91	-311.27
			retained	74.09%				retained	58.17%

PS.  Shameless ads for Eve posts are welcome, especially when on topic.  See tomorrow's post.

PPS.  I will not be writing about the CSM election for at least 2 days.  I have time to write posts such as these or time to play eve, but apparently not both.  Tomorrow's (non CSM) post is written and scheduled.  I also appear to fail terribly at writing short posts.