Wikipedia talk:WikiProject Ireland Collaboration/Poll on Ireland (xxx)

Missing option - "Ireland"?

Just curious, though it would be my first preference, why isn't Ireland an option? There's a precedent for this (e.g. Samoa (cf. Samoan Islands, American Samoa)). My ideal would be Ireland (modern state), Ireland (island), Ireland (historic state), Republic of Ireland (descriptive term from the 1948 Act).

Cheers, This flag once was red^propaganda_deeds 18:25, 24 June 2009 (UTC)

Please see the main WP:IECOLL talk page. This is a preliminary round to select what options go into the main poll alongside the ones already proposed (which include 'Ireland'). Basically, there are several proposals to move Republic of Ireland to an article titled 'Ireland (xxxx)' where 'xxxx' is what you're voting on here. The most popular options from this poll will be added as options in the main, final poll. Bastun^{Ėġáḍβáś₮ŭŃ!} 18:31, 24 June 2009 (UTC)

Phew! Got it, thanks. Cheers, This flag once was red^propaganda_deeds 18:36, 24 June 2009 (UTC)

Why is Ireland not a preference? BigDunc^Talk 18:50, 24 June 2009 (UTC)

See my question (and the response) above ;-) It's already through; we're !voting for the remaining finalists. This flag once was red^propaganda_deeds 18:53, 24 June 2009 (UTC)

D'oh. BigDunc^Talk 18:55, 24 June 2009 (UTC)

Yeah - think of it as the semi-final of the Eurovision... (or maybe not...) Bastun^{Ėġáḍβáś₮ŭŃ!} 23:20, 24 June 2009 (UTC)

I hope it's not like that, as that event has history of not bringing the most talented musicians to the main show. I hope this vote won't miss out on the best option!!MITH 00:29, 25 June 2009 (UTC)

Ah now I get it! Like Eurovision, the votes will be discarded and the "experts" will decide which ones go through to the final? Skinsmoke (talk) 00:23, 25 June 2009 (UTC)

Well have a group acting like the Eastern Block all voting for each other regardless of quality of talent. --Domer48_'fenian' 13:45, 25 June 2009 (UTC)

Nyet. GoodDay (talk) 21:46, 26 June 2009 (UTC)

Talk page

So why is there a discussion going on on the poll page when it should be happening here? Anyone object if I move it? Fmph (talk) 12:02, 25 June 2009 (UTC)

Whatever rows your boat. GoodDay (talk) 21:47, 26 June 2009 (UTC)

Watching the results come in...

This is more exciting than a real election! Fmph recommended a software called OpenSTV to caculate the results. Here's how to use it:

• You will first have to create a "ballot file" (in the File menu) where you fill in all of the ballots one at a time (quite tedious work).
• Save the ballot file then start a "New Election" from the File menu.
• Select the ballot file and an STV formula. I'd recommend "Fractional transfer STV" for the formula, which is more 'precise' in making transfers - it's what is used in Senate election in IRL. The alternative is "Random transfer STV", which is easier for a manual counter - it is what is used in Dáil elections in IRL.
• On the next screen, leave all of the options at their defaults except for the number of "seats". We are going to have 2 or 3 seats depending on what we decide later.
• When you press OK the results screen will pop up showing the details of all transfers/elimations/etc. Scroll down to the bottom, there will be a line saying, "Winners are ...". That's what we're interested in.

... grinners? Daicaregos (talk) 21:28, 24 June 2009 (UTC)

I've done a calculation on the first 11 ballots. The results for the first two two of the options are predictable. The choice of a third one option is *very* interesting though ;-) --rannṗáirtí anaiṫnid (coṁrá) 20:20, 24 June 2009 (UTC)

Please don't obsess over this and please don't give running tallies for a week. 20:28, 24 June 2009 (UTC)

Agree we don't need a running tally, but thanks for the manual on how to use the software. BigDunc^Talk 20:30, 24 June 2009 (UTC)

True. Just being giddy. --rannṗáirtí anaiṫnid (coṁrá) 20:32, 24 June 2009 (UTC)

From a manual inspection, the third place result is close enough that it may possibly depend on the precise version of STV used. So can I suggest that before we get too far through this, we nail down the exact version of STV that we intend to use? I have deliberately not researched the possibilities too much so that I cannot be accused of selecting an STV variant that favours my preferred solution(s). However I would suggest we agree to follow the same procedure that was used to select the three MEPs for the Northern Ireland European constituency in the 2009 elections? So far as I'm aware, this is the only system of PRSTV in current use in Ireland or the UK, although others variants are common elsewhere in the world. It may well be that the result will be sufficiently clear that the particular counting mechanism makes no difference, but it seems prudent to agree to this before the vote gets too far under way, and before it becomes too easy for people to predict what effects the STV variant might have on the final outcome. —ras52 (talk) 21:32, 24 June 2009 (UTC)

Yes, the exact formula can make a difference. I don't know exactly what formula is used in NI but the OpenSTV software has a "N. Ireland STV" formula so we could go with that. (STV is used throughout Ireland, north and south, with different variants in different types of elections so there is no definitive "Irish" STV formula.)

Otherwise the "standard" STV options are "Fractional transfer STV" or "Random transfer STV". The difference between the two lie in how transfers beyond the quota are calculated. "Fractional transfer STV" makes these transfer proportionately to the entire ballots cast for the option. "Random transfer STV" makes transfers beyond the quota by selecting ballots randomly from ballots cast for the option. "Fractional transfer STV" is therefore "better" but harder for a human to calculate. We however have what looks like a pretty nifty piece of software to do the calculating for us so we can go with "Fractional transfer STV".

The default options chosen by the software are to use a "Droop", "static" and "whole" threshold (quota), which is standard. It will only "batch eliminate" options if they have zero votes. It won't "delay" transferring ballots (a short-cut technique used when counting by hand). The precision of the fractions used when transferring ballots (for "Fractional transfer STV") is 6 decimal places. This all sounds good for us here.

For the BIG ballot we could use the built-in "Instant runoff voting" option. We also have the option of using "Random transfer STV" or "Fractional Transfer STV" with only 1 "seat" and using the "Hare" threshold. The advantage of doing this would be that the count would be continued all the way to the end (i.e. all possible transfers will be made). This means that that we can see the proportion of votes for the winning option beyond 50%+1 (whereas the Droop quota would stop counting at that point).

My 2¢ is to use:

• "Fractional transfer STV" with the standard options for the poll being conducted on this page
• Edit: For the "BIG" ballot: "Instant runoff voting" (simplest) or "Fractional transfer STV" with the standard options except to use the Hare threshold for the "BIG" ballot (only useful to see how transfers work out if we continue on transferring past 50%+1)

--rannṗáirtí anaiṫnid (coṁrá) 22:31, 24 June 2009 (UTC)

Edit: change of mind on what to use in the "BIG" ballot. For simplicity, let's use instant-run off.

Not to suggest that it all sounds like witchcraft to me, I think the same method for counting should be used for both ballots, so there is no question of using different rules or anything. Or is your suggestion because this poll seeks more than one winner, and the big poll will seek only one? -- Evertype·✆ 09:44, 25 June 2009 (UTC)

It is because this poll seeks more than one winner - otherwise they are practically the same. For them to be identical, I suggest we do either:

• Use the "N. Ireland STV" option in OpenSTV in both ballots
• Use the "Fractional transfer STV" option in OpenSTV (using the default options) in both ballots

I don't know exactly what formula the "N. Ireland STV" option uses. I suspect it employs some randomness and/or other short-cuts used when manually counting. Not using the "Hare" quota in the "BIG" ballot means that counting will stop when any option gets 50%+1, but I think that there was no support to go beyond that point so it is unnecessary.

Alternatively, if there is no desire to keep counting past 50%+1, the "Instant run off" option is (speaking precisely) the option we want for the "BIG" ballot. It too is practically identically to either of the bullet-pointed options above. --rannṗáirtí anaiṫnid (coṁrá) 10:18, 25 June 2009 (UTC)

(ec) In answer to Evertype, the important difference is that this poll will yield than one winner. When selecting several winners there are quite a lot of variants to the basic mechanism, and the particular variant used can make a difference to the outcome. When only precisely one winner is required, the variants still correspond to different algorithms for calculating the result, but you can prove mathematically that all of the common STV variants will yield precisely the same answer in every single possible case. So in the case of one winner, you may as well use the simplest algorithm which is almost certainly instant-runoff voting (IRV): it won't effect the result. However, IRV is a special case of STV and doesn't uniquely generalise to the case of multiple winners.

As a side issue, it has been suggested elsewhere that we might want the final community poll to have the option of returning a none of the above result if even after considering second, third, etc. choices, a consensus cannot be found. None of the ordinary ways of handling an STV election can result in such an outcome, and we would have to develop a custom variant specifically to do that. I am concerned that we are sigificantly underestimating the complexity in doing that. Also, I'm far from convinced that community wants there to be any possibility that this vote might not result in an answer.

To address some of Rannṗáirtí's comments, I think there are significant problem using any mechanism involving random transfers (as occur, for example, in most US elections using STV). In the real world, random transfers are robust for two reasons: first, the number of votes is large enough that the small percentage of them transferred is still going to be statistically significant; second, with an open, independently monitored election process, it's possible to be fairly sure that votes transferred really are selected at random and not tailored to suit the perceived biases of the person counting the votes. Neither of these apply in this case. (Note: I'm not suggesting for a moment that Masem, or his appointed vote-counter, would manipulate the count in any way — I'm merely saying that it is impossible for anyone to verify that he hasn't, and that can breed suspicion.)

When I suggested we use the mechanism used throughout the UK and Ireland, I deliberately didn't look at the precise details used, other than to verify that it was not a random-based mechanism. As this vote concerns the UK and Ireland, it seemed most relevant to use the precise version used throughout these islands. The specific suggestion of the NI European elections in 2009 was simply so that, in the event of any variation within Ireland and the UK (which I don't believe there to be), we had a specific set of rules on the table. I would sooner use something like this than OpenSTV's idea of "Fractional transfer STV" simply because the term "Fractional tranfer STV" isn't well-defined -- several subtly different ways of doing fractional transfer exist. I think it's a good idea that (in principle) any Wikipedian can verify the result of the poll without downloading and running OpenSTV.

Hopefully this conversation will turn out to be unnecessary because it may well be that the votes will be clear enough that all methods will yield the same result. Nevertheless, I do think we should get agreement before rather than afterwards in order that we can avoid any unnecessary unpleasantness if the results are not unambiguous.

—ras52 (talk) 10:53, 25 June 2009 (UTC)

I couldn't agree more with regard to random transfers, etc. However, there is no definitive Irish/British system. Even within the Republic of Ireland there are several different formulas used: random transfers for the Dáil (House of Representatives), fractional transfers for the Seanad (Senate) and IVR for the president. The North uses fractional transfers (i.e. not random) for certainly for Assembly elections and EU elections, but it does something freaky between elimination rounds that causes a difference in transfers. I don't know what the Scots do with their "top up votes". That is why I suggest using the default fractional method from an agreed software - there can be no mistakes made, all of the subtle differences are avoided, we simple replicate each other actions.

None of this, though, will affect the "BIG" ballot (because there will be no transfers). If there is a desire to just "end this" then an instant run-off will give us an answer that will be proportionate. If we say, "end this only if the result has more than 50%/60%/etc. support" then we need the keep counting past where IVR would normally stop. But should I take it from your comments that that is not what people want i.e. that people want to "end this"? And so, can we assume we will use "instant runoff" in the "BIG" ballot (regardless of whether we use software or count it by hand)?

"I think it's a good idea that (in principle) any Wikipedian can verify the result of the poll without downloading and running OpenSTV." In principle, but a) most people don't know/don't care how to actually do an STV/IVR vote count and b) we would be here forever and a day arguing over who miscounted the vote and who the *real* winner was (either by ignorance, error or poor communication of the subtitles). Software, even if it removes us a little from the process (which is not a bad thing either!), gives us a definite, easily replicable process and an answer to a relatively difficult vote counting process. OpenSTV is a freely available software that everyone will be able to type the ballots into, punch "Instant run-off" and get a single definite answer. No need to worry. No need to argue.

RE: "None of the above" could run as an ordinary "candidate" i.e. could be eliminated/could win. If it won, it would mean that the preferred choice was "none of the above". If it it was eliminated then, like any other option, it could make transfers to remaining "candidates". I just threw it out there because it was mentioned before. --rannṗáirtí anaiṫnid (coṁrá) 14:23, 25 June 2009 (UTC)

From the votes to date (up to Fionnsci's vote) using the Seanad system with 2 seats-

Total Poll:39 Total Valid Poll:39 "Seats":2 Quota:14
••••1st Count•••••••Result
A•••8
B•••0
C•••8
D•••0
E•••0
F•••21••••••••••••••ELECTED
G•••2
Non-Transferable:
--------------------------------------------------------------
••••Change•••••••••2nd Count••••Result
A•••2.666666667••••10.66666667
B•••0•••••••••••••••0••••••••••••ELIMINATED
C•••0.666666667••••8.666666667
D•••2•••••••••••••••2••••••••••••ELIMINATED
E•••0•••••••••••••••0••••••••••••ELIMINATED
F•••-7••••••••••••••14
G•••0.666666667••••2.666666667
Non-Transferable: 1
--------------------------------------------------------------
••••Change•••••••••3rd Count••••Result
A•••1•••••••••••••••11.66666667
B
C•••0•••••••••••••••8.666666667
D•••-2
F•••••••••••••••••••14
G•••1•••••••••••••••3.666666667•••ELIMINATED
Non-Transferable: 0
--------------------------------------------------------------
••••Change•••••••••4th Count••••••Result
A•••1•••••••••••••••12.66666667•••ELECTED WITHOUT REACHING QUOTA
B
C•••1.333333333••••10•••••••••••••ELIMINATED
D
F•••••••••••••••••••14
G•••-3.666666667
Non-Transferable: 1.333333333
--------------------------------------------------------------
A & F are two preferred options, if it is to be 2. Can be redone for 3 seats, if decided. Done using excel.
Bogger (talk) 15:35, 25 June 2009 (UTC)

I've spent a little while playing with the results so far too. It looks like irrespective of which particular method is used, the top three options come out in the same order. I shan't state what those results are, but lets just say they're not surprising. —ras52 (talk) 16:16, 25 June 2009 (UTC)

So much for my suggestion that we be circumspect and let the poll run its course for a week. -- Evertype·✆ 18:23, 25 June 2009 (UTC)

I can't get the same exact transfer values as the above using any method (probably owing to mistakes in taking down the results and differences in transfer methods), but the end result is the same. For the "BIG" ballot, are we agreed to use the "instant run-off" option in whatever software/hand-counting we use (not that it will make any difference, but just so we are all on the same hymn book).

For this poll, can we get agreement on the exact method we will use to calculate the transfers. (Bogger, did you do it by hand or with using an Excel plug in?) The result looks like it will be one easy winner and close run for 2nd and 3rd. At least one editor has expressed that he/she will want one of the 2nd/3rd placed options included in the final ballot, so (depending on people's mood), exact results might matter. For example, none of my calculations had A not reach the quota - but that could be down to discrepancies in ballots. --rannṗáirtí anaiṫnid (coṁrá) 21:54, 25 June 2009 (UTC)

The method used for the "big" ballot should be decided at WP:IECOLL, not here. Bastun^{Ėġáḍβáś₮ŭŃ!} 22:20, 26 June 2009 (UTC)

How many options to go forward?

Can someone explain to me whats the point in calculating what comes second and third. Surely the process is the calculate the option that will fill [[Ireland (something)]] slot? If too many options are given then it will become too fragmented once the final solution is decided.MITH 02:22, 27 June 2009 (UTC)

The plan has been to choose two, or possibly three, options which will end up in the Big Poll. -- Evertype·✆ 08:28, 27 June 2009 (UTC)

The more choices we forward, the better the chances of overturning the status quo. I fear some of the SQites may be on to that! So three it should be. Sarah777 (talk) 10:05, 27 June 2009 (UTC)

I'm just a bit confused how that works though. An example in my head: 4 people vote. I likes the status quo and 3 others like "Ireland (something)" solution instead. 1 votes for Ireland (country), the other Ireland (state) and the other Ireland (Republic) with the 4th voting for ROI. Does that not make it a tie and worse for removing the status quo instead of if Ireland (state) had just gone through. That way it would be more likely that the 3 initial voters would have voted for just Ireland (state) as it was the only option for Ireland (something)?MITH 10:34, 27 June 2009 (UTC)

It could, but not in this case I think. If, like me, you find the SQ objectionable but there are three alternative options you could support with varying degrees of enthusiasm then the less people vote for the SQ the better. While a few folk have made an emotional investment in the SQ, most haven't, and would look at a compromise. You put forward only "Ireland(state)" and they don't like it, their default will be the SQ. But if they see two other options they could live with then the SQ loses a first preference. Looking at the trends in the mini-vote it seems many editors will transfer from their #1 to a few non-SQ alternatives, leaving out the ones they don't like. But if they are not offered an option that they do like - the SQ gets their vote. Simple, eh? Sarah777 (talk) 10:50, 27 June 2009 (UTC)

Of course this would largely "self-correct" if everyone voted the full range. But I notice that many non-RoI folk don't seem to be doing that:) (probably 'cos they haven't figured the implications) Sarah777 (talk) 10:55, 27 June 2009 (UTC)

All this SQ stuff, is tough on my limited IQ. GoodDay (talk) 14:06, 27 June 2009 (UTC)

G'day - SQ is merely shorthand for "status quo". Sarah777 (talk) 15:22, 27 June 2009 (UTC)

Okie Dokie. GoodDay (talk) 15:38, 27 June 2009 (UTC)

"The more choices we forward, the better the chances of overturning the status quo." I'm not at all sure you're right, Sarah. Since there is only one winner, and by definition there is only one SQ option, the more alternatives there are the more the vote will be split. If there is a majority opposed to the SQ, then in principle it will never achieve 50%, and some alternative will. In practice, however, not all of the eliminated options will have lower preferences, so the option with the biggest first preference vote is favoured. In practice, also, options that are not supported in this poll will most likely not be supported in the big poll either. I reckon the "change" lobby (and the community in general) will be better served by limiting the number going forward to two, particularly if there are two that are obviously ahead of the rest. Scolaire (talk) 20:22, 27 June 2009 (UTC)

Yes, a "the-more-options-the-the-better" approach will not lead to the status quo being more likely to be overturned, it will only lead to there being more options. That could have a very undesirous effect for those who favour "(xxx)"-like solutions for the current Republic of Ireland article. Every "(xxx)" options we nominate to the "BIG" poll results in 3 additional options being included on that poll. The more options there are, the lower the rate of transfer between options, particularly for between very similar options.

Example: If we include a single "(xxx)" option, a voter may rank all three variations of that option in the "BIG" poll. If we include two "(xxx)" options, the same voter may say "screw it" and only rank one set of variations and leave the set of variations other on the shelf. And, if we include three "(xxx)" options then do you really imagine that only voter would be really bothered to rank a preference against nine near-identical options? The result would be that the greater the number of "(xxx)" options we include, the greater the "(xxx)" vote would be split. (This why political parties limit the number of candidates they put onto a ballot in a real election.)

I think the best interest of the "(xxx)" community is served by nominating only one "(xxx)" "candidate" to the "BIG" vote. The usual choice is the candidate that is most likely to win. In this case, it would be the "candidate" that tops this poll. --rannṗáirtí anaiṫnid (coṁrá) 22:48, 27 June 2009 (UTC)

Indeed, multiple pro-change options has & will continue (if not reduced) to prevent getting a change. GoodDay (talk) 23:27, 27 June 2009 (UTC)

It is fully agreed that 3 options will be put forward? I really think we should leave it at one or 2 at the most.MITH 23:37, 27 June 2009 (UTC)

Nothing is agreed. Something really ought to be, and soon! Scolaire (talk) 05:53, 28 June 2009 (UTC)

I agree that TWO options should go forward. This is what was generally agreed before the poll. My understanding is that the decision to include a THIRD option will be made (or not) after the poll is closed and counted. I think Sarah is a bit out of line with not-so veiled canvassing like "the more choices we forward the better the chances of overturning the status quo." Moreover, I think the more choices, the more confusing the poll will be. -- Evertype·✆ 07:34, 28 June 2009 (UTC)

As I said, if everyone voted the full slate it would not matter much how many options went forward. My analysis is specific to this real vote. There is a hard-core for SQ; they are supplemented by a non-returning "soft" supporters who can be attracted away if their personal favourite is presented. Key to this is my perception that these "soft" SQ supporters are less committed and more likely to detach from the SQ; and consign their vote to oblivion after one or two preferences. Sarah777 (talk) 10:45, 28 June 2009 (UTC)

Canvassing issues aside, I think you're being unfair to your own side! It has always been considered as axiomatic that apart from the "hard core" nobody agreed with "Republic of Ireland", so that while many people who voted for ROI would give "Ireland (bracket)" as a second preference, the reverse would not happen. You now seem to be saying that people would have to be "attracted" away by presenting a range of alternatives! I said before that two would be better than three, but on reflection I think that if one option gets more than 50% of first preferences in this poll, it would be better to have only that one in the main poll. Remember that it will still make up three out of six options, with ROI→Ireland as a fourth. That way, with six options, it's virtually certain that ROI will be renamed, unless a majority of users actually do want it to be kept. Scolaire (talk) 12:18, 28 June 2009 (UTC)

OK - as we are all only speculating I'll go with just two names if it helps move this along. Sarah777 (talk) 14:01, 28 June 2009 (UTC)

Enfranchisement

I've just checked the earliest date of contribution for every one of the people who voted down to Staighre and the newest two voters are Jack1755 (2009-05-17 though as he says on his page he had been editing previously) and Rannṗáirtí anaiṫnid (2009-05-24). Since the project agreed that enfranchisement was open to those who registered prior to 2009-06-01, all of the votes cast are valid as far as that criterion is concerned. -- Evertype·✆ 10:12, 1 July 2009 (UTC)

Closure

As I opened this poll I have made plans to close it after the appointed time. Just so we don't have edit conflicts in 3 hours. -- Evertype·✆ 17:44, 1 July 2009 (UTC)

Now, what happens next? GoodDay (talk) 21:09, 1 July 2009 (UTC)

Now is as good a time as any - on to the main vote? Sarah777 (talk) 21:28, 1 July 2009 (UTC)

Yeah, I think tomorrow the notice should be put on all the wikiprojects/noticeboards and should begin on Friday. FF3000 · talk 22:37, 1 July 2009 (UTC)

What happens next is a tally of this poll. -- Evertype·✆ 23:55, 1 July 2009 (UTC)

Aye - tally this poll, decide how many make it though to the final, finalise the arrangements (the draft poll page, for example)... and really, User:Masem should decide when it starts (or at least announce that there's consensus for it to start). Bastun^{Ėġáḍβáś₮ŭŃ!} 00:43, 2 July 2009 (UTC)

Who's tallying? Rannṗáirtí Anaiṫnid? -- Evertype·✆ 07:05, 2 July 2009 (UTC)

I'll do it; it is a simple matter of isolating the random factotum (x) and solving successively for each !vote (i);

$f(x)={\tfrac {(x^{2}-1)(x-2-i)^{2}}{x^{2}+2+2i}}$

reiterating then for each option that exceeds the quantum quota by $-{\frac {\hbar ^{2}}{2m}}{\frac {d^{2}\psi }{dx^{2}}}=E\psi .$ (or the ∞ root of Option C).

Squared. Then add 3 and throw some salt over your left shoulder. Bingo! — Preceding unsigned comment added by Sarah777 (talk • contribs)

The salt goes over your right shoulder, Sarah777. Common mistake. You also didn't hop anti-clockwise around the ballot papers seven times holding your left foot in your right hand behind your back, but that's not necessary depending on the method used. I've dusted the salt and randomised the factotums here. --rannṗáirtí anaiṫnid (coṁrá) 08:10, 2 July 2009 (UTC)

Giddy? -- Evertype·✆ 08:19, 2 July 2009 (UTC)

Frequently. Sarah777 (talk) 00:29, 3 July 2009 (UTC)

Please, conduct discussion here. -- Evertype·✆ 08:56, 2 July 2009 (UTC)