Wikipedia:Mediation Cabal/Cases/27 February 2012/Wikipedia:Verifiability/Archive 2

Step five
Welcome to step five! In this step we are going to begin the process of choosing a final draft for each work group. I have seen a lot of good drafts, and some good attempts at compromise. Thank you all for your hard work. Right now I want to stop focusing on creating new drafts, and see if we can find consensus to use any of the drafts we have made so far. This is a little different from my original plan for step five; the varied approaches we have taken with step three and step four have meant that I have needed to improvise slightly. But if the mediation went 100% according to plan, it wouldn't be any fun, now would it? ;)

You might have noticed that the work group pages are now full-protected. This is to ensure that everyone is commenting on the same number of drafts, and that none of the contents of any of the drafts changes while we are discussing step five.

My plan for this step is to have a kind of mini-RfC to judge what everyone thinks about the different drafts. The aim of this mini-RfC is to find a final draft for each work group, or at least to point us in the right direction for deciding which drafts to put in the RfC. We have had quite a bit of discussion about instant-runoff voting in the last few days, and that has given me the idea to use it in this step. As well as testing out the voting technique, I want to have a more traditional discussion to find out everyone's views on the drafts and on the different work groups. For instance, if you like everything about a draft apart from some specific wording, and that wording wasn't made into a separate draft, I would like you to bring that up.

Also, I would like to know whether you think a certain draft doesn't belong in a certain group. A few of the drafts in group four, for example, talk about the difference between verifiability and truth. Maybe you think some of these should be in group three instead, or maybe you think they are different enough from the group three drafts that they can be in group four, or maybe you would like to create a new group for them. Maybe you would like to see some of the groups merged - whatever it is you think about this, I want you to write it down. Finally, if you have any other thoughts about the drafts or the work groups, I want to hear those too.

I realise, though, that many of these things have already been discussed at length in this mediation. Don't worry - there's no need to write things down again. I will factor in the discussions that we have already had when judging whether we have found consensus. For example, there is no need to repeat the discussion we had about the possibility of making a new group using group 2 draft 6, group 3 draft 6, etc. If you have anything new to add about previous topics we have discussed, though, it might be a good idea to mention it.

So, I would like everyone to prepare a statement, preferably within 500 words, including the following four things:


 * 1) A list of each group's drafts, ranked in order of preference. (Start with the draft you like the best, and end with the one you like the least.)
 * 2) If there are any drafts whose wording you would like to change slightly, but there isn't a new draft that includes those changes, please include what you would like to see changed.
 * 3) Your thoughts on the work group structure, and whether there could be better criteria for deciding which drafts are included in the RfC. Maybe you would like to increase or decrease the number of drafts, or maybe some drafts belong in another work group. The groups are not set in stone, and this is your chance to change things if you want to.
 * 4) Any other thoughts you have on the drafts and the work groups. This could be about a general principle you think we are overlooking in the current drafts, or something that you think some of the drafts do really well. Let us know what's on your mind.

The following is an example of what a statement might look like. (Please note that the drafts in the example were chosen at random, and that the comments are fictional.)

Of course, your statements probably shouldn't talk about Martians. :)

Judging from the time the other steps have taken, I think my estimate of two days for this step was too optimistic. So I'm just going to double it straight away, and give you four days to leave your statements. The deadline for this step is 10.00 am on Friday 6th April (UTC). Also, it hasn't mattered all that much if people missed the previous deadlines, but this time I am going to be strict. If you don't leave a statement by the deadline, then you won't be able to participate in steps six or seven. This is mostly because if people turn in their statements late, the result could change when we factor in their preferences. It will be much simpler to have a clear cut-off point than to keep backtracking for people who weren't on time. If you think you are going to be late turning in your statement, please let me know as soon as possible, and I will see what I can do.

If you have questions, or if you want to raise discussion points about this step itself, let's do it on the talk page. I'm looking forward to seeing all of your statements. —  Mr. Stradivarius ♫ 17:15, 2 April 2012 (UTC)

Statement by LeadSongDog

 * Gr 1: no opinion, it's silly to debate history that's in the logfiles
 * Gr 2: 4 3 6 2 5 0 1 8
 * Gr 3: 7 8 10 9 1 6 5 3 4 11 12 0 2
 * Gr 4: 0 4 1 2 5

I am pleased that most of the draft changes use simplified sentence structures than had been the case. I am less happy that several drafts restate other policies unnecessarily. LeadSongDog come howl!  19:33, 2 April 2012 (UTC)

S Marshall

 * Group 1: 2 1 0
 * Group 2: 4 6
 * Group 3: 2 12 8
 * Group 4: 0 1 4

I want to restate my view that the drafts are a distraction from the important and useful part of this RFC when we ask the community for steerage, and I therefore don't care very much which drafts are used. What matters to me is getting some unstructured discussion that's about principles rather than specific wording.— S Marshall T/C 20:33, 2 April 2012 (UTC)

Statement by Tryptofish

 * Group 1: 0 as a strong first choice, followed by 1 (distant 2nd), and 2 (very distant 3rd).
 * I see 4 and 3 (in that order) as options for something in addition to the Group 1 draft, but would rank them at the bottom here.


 * Group 2: 5 as a strong first choice, followed by 1, 0, 3, and 2.
 * I see 8, 6, and 4 (in that order) as options for something in addition to the Group 2 draft, but would rank them at the bottom here.
 * In draft 5, I would revise "nothing, such as perceived truth or personal experience" to "nothing, such as your personal experience or what you know to be true".
 * Kalidasa 777, below, suggests some additional revisions that could apply to draft 5. I believe that I could support most or all of those, depending on what, exactly, they would look like.


 * Group 3: 8, 5, 2, 3, 4, 12, 11, 10, 9, 6, 1, 0, 7. (I don't feel strongly here.)


 * Group 4: 4, 1, 2, 0, 5. (Again, I don't feel strongly here.)

Something I don't like about the format of this discussion is that I'd rather see discussion on the talk page of each group establish a consensus, whereas this numerical system is likely to lead to artifactual results. --Tryptofish (talk) 21:49, 2 April 2012 (UTC)
 * Agree with Tryptofish on the talk page suggestion. I think that that would be very useful.  So far there has been no collaborative development work as a part of the process. Sincerely, North8000 (talk) 22:41, 2 April 2012 (UTC)

Statement by North8000

 * Group 1: 2, 0, 3, 4, 1  Liked 0 the LEAST, but rated it high because it should be included
 * Group 2: 5 (strong first choice) 1, 2, 3, 4, 6, 8.  (#7 is not a choice)
 * Group 3: 8, 12, 2, 3, 4, 6, 9, 11, 1, 0, 10, 5 (the last 4 distant simply due to one fatal phrase)
 * Group 4: 4, 0, 5, 1, 2

Statement by Unscintillating

 * Group 1: 4 3 2 1 0. 1 and 0 are not RfC drafts, but 1 is slightly shorter.
 * Group 2:
 * 8 6
 * 3 4. both talk about the mysterious "other policies, guidelines and considerations...apply."
 * 2. adds more "thresholds" unnecessarily
 * {0 1 5} As an unordered set, each draft defines, depending on how you read it, either "not truth" as a requirement, or "VnT" as a requirement.
 * 7. not an RfC draft


 * For Group 3 I identified three unordered sets:
 * {3, 4, 6, 7, 9} best group
 * {2} This draft contains the wording, "It is not enough that the information is true."
 * {0, 1, 5, 8, 10, 11, 12} Each draft in this group contains the phrase, "whether editors personally believe information to be true or false should never determine Wikipedia content." IMO drafts in this set would undermine parts of WP:IRS, WP:NPOV, WP:EP, and WP:ELNO, as well as WP:Inaccuracy.


 * Group 4: 5 3 0 2 1 4. Draft 4 contains the wording, "It is not enough that the information is true."

Comments: From my viewpoint the schedule continues to be overly aggressive. I don't currently see how the viewpoint groups need to break out. And where is the point when the mediator reminds us that building consensus takes work, that we are not all at the same skill level&mdash;but then, I'm an engineer, not a people person. As an engineer, I know that systems require feedback, or they would not be systems. Unscintillating (talk) 06:20, 3 April 2012 (UTC)
 * Well, open loop is not as bad as an unstable control loop, but almost. :-) North8000 (talk) 13:04, 3 April 2012 (UTC)
 * Re the "mysterious" other policies, guidelines and considerations, we are only discussing a lede here, not the whole policy. I thought it was obvious without saying that such details would be fleshed out in the body of the page.LeadSongDog come howl!  20:28, 5 April 2012 (UTC)

Statement by ThatPeskyCommoner

 * Group 1: 3, 2, 4, 1, 0
 * Group 2: 5, 1, 3, 8, 6, 4, 2, 0
 * Group 3: 12, 10, 9, 11, 3, 8, 5, 4
 * Group 4:  5, 4, 2, 1, 0

No particular other comments, really. Pesky (talk ) 16:16, 3 April 2012 (UTC)

Aahh, just going to add: I agree with S Marshall's point, above about getting some steering guidelines from the community, but I kinda expect that may happen as part of the RfC process? Pesky (talk ) 09:09, 5 April 2012 (UTC)

Statement by BruceGrubb
I agree with the misgivings of Tryptofish and North8000 and go one further--we have too many drafts scattered over too many groups. I also agree with Unscintillating that this schedule is overly aggressive for something that has been kicked like a football for months.--BruceGrubb (talk) 17:48, 3 April 2012 (UTC)

Statement by Olive
Group one: 3 and worked 4 together as is being discussed on the Gr. 1 talk page. I see the others as versions useful as historical references to the policy, although they could be named as drafts as well.

Group 2: 2, then 5. 0,1,3,4,6,8 equal.

Group 3:

Group 4: 4,1,2,0,5

Statement by Blueboar

 * Group 1: 0 (strongly urge that this should be presented in the final RFC, no matter what... as it was the version that started the entire debate almost a year ago) then: 1, 3, 2, 4
 * Group 2: 2, 5, 3, 4, 1, 0, 6 (might swap 2 and 5 if recent comments to 5 are followed up on)
 * Group 3: 11, 2, 3, 1, 0... (no need to list the others... they would definitely not be my preference).
 * Group 4: 4, 5, 2, 1, 0

Statement by Kalidasa

 * Group 1: 0, 1, 2


 * An RfC with a list of drafts or versions must include a traditional version which includes the words "verifiability, not truth" and "threshold", without recent afterthoughts. If not, the RfC would naturally be perceived as unfair. North's proposal (on the main talk page) that there should be two "status quo" drafts is not a bad idea. One could be Group 1 Draft 0, and the other could be Group 1 Draft 2 (the current locked version).


 * Group 2: 8, 6, 0
 * Group 3: 11, 10, 12
 * Group 4: 5, 0, 4


 * One reason I have given top rating to Draft 8 in Group 2, Draft 11 in Group 3, and Draft 5 in Group 4... Each of these mentions the accuracy/inaccuracy question without using a phrase like "verifiable but inaccurate". That phrase reminds me of "verifiability, not truth" or VNT -- I can't resist abbreviating "verifiable but inaccurate" to VBI. Just like VNT, VBI is punchy, yes, but also open to diverse interpretations, some helpful, some not...


 * Saying "other policies apply" is right in principle but vague. And therefore unhelpful to the reader. So I prefer drafts that are more specific about the need for verifiable material to comply with NPOV, WP:DUE, NOR and copyright policy.


 * If we are going to present four different drafts (or five different drafts, with the two "status quo" ones), then they need to be just that: different. The idea of convergence (reducing the difference between drafts) is fine within each group, but I don't think it should mean putting drafts into group 1 which would fit better into Group 2 (Group 1 Drafts 3, Group 1 Draft 4). Nor should it mean putting drafts into group 4 that really belong in group 3 (e.g. Group 4 Draft 1, Group 4 Draft 2). Kalidasa 777 (talk) 23:11, 5 April 2012 (UTC)

Statement by Becritical

 * Group 1: 4, 0, 1, 2, 3,


 * Group 2: 0, 1, 2, 3, 4, 5, 6, 7, 8


 * Really nothing to choose from in this group: no change is better than any draft.


 * Group 3: 8, 1, 0, 2, 6, 4, 5, 7


 * 7 is kind of nice but does not make the policy imperative.


 * Group 4: 0, 2, 1, 4, 5


 * 0 is beautiful

Statement by Mangoe
A number of versions, particularly in Group 3, contain the statement "whether editors personally believe information to be true or false should never determine Wikipedia content" or its near equivalent. This expresses the position that us in the accuracy camp find most objectionable: that source errors are not to figure in the exclusion of material. Even as I dislike the original formula, it is better than this. Therefore I would prefer to rule out versions which contain this.

My most preferred versions are Group 4 drafts 4 and 0, in that order. I also like Group 3 drafts 2 and 7; draft 3 is a starting point though the last sentence would need further examination. In the Group 1 the only draft I can see working with is #1, and in group 2, draft 0. The others are either too wordy or awkward, or contain the statement to which I object; I am not moved to rank within those subgroups. Mangoe (talk) 17:15, 6 April 2012 (UTC)
 * Ah, I think that's an ambiguity thing! I'm pretty sure that what we meant was that the "unverifiable truth" and the "unverifiable rebuttal of verifiable truth" can't be inserted, if you know what I mean. Not that we can include verifiable stuff even if we have (verifiable) reasons for believing it to be false, just that we can;t delete it if we we just "know" it's false, but can't source the "it's false" statement.  Pesky  (talk ) 09:15, 7 April 2012 (UTC)

Statement by Bob K31416
Policy should be quotable. I just had an occasion to use the second sentence from group 4 draft 0 on an article talk page. Which of the other drafts had something quotable to use in this case? For example, to a user's suggestion of including in an article that Obama is racist, would you say "verifiability, not truth"? --Bob K31416 (talk) 16:10, 7 April 2012 (UTC)

Step five results and discussion
Sorry for being late in getting to this. It turns out that finding the results of instant runoff votes can be quite time-consuming! Before getting to the individual groups, perhaps the clearest result of this process is that instant runoff votes lose their effectiveness when the number of voters is very small. You can judge this by the fact that I had to resort to using a random number generator six separate times to eliminate the last-place drafts. (I used the list randomizer at random.org for my randomization, if anyone is interested.) And of course there is the fact that instant-runoff can't account for "distant last" or "strong first" votes, and that it ignores any conversation. Because of this I have attempted to use it as a tool for finding the few most popular drafts, and not as a tie-breaker. I have also tried to bear in mind the viewpoints expressed by everyone in my comments following the vote results.

Group one

 * SMarshall: 2 1 0
 * Tryptofish: 0 1 2
 * North8000: 2 0 3 4 1
 * Unscintillating: 4 3 2 1 0
 * Pesky: 3 2 4 1 0
 * Olive: 4 3 (determined by random-number generator)
 * Blueboar: 0 1 3 2 4
 * Kalidasa: 0 1 2
 * Becritical: 4 0 1 2 3
 * Mangoe: 1


 * Round 1


 * Draft 0: 3 votes
 * Draft 1: 1 vote
 * Draft 2: 2 votes
 * Draft 3: 1 vote
 * Draft 4: 3 votes

Votes needed for a majority: 6. Eliminating draft 1 (chosen at random).


 * Round 2


 * Draft 0: 3 votes
 * Draft 2: 2 votes


 * Draft 3: 1 vote
 * Draft 4: 3 votes

Votes needed for a majority: 5. Eliminating draft 3.


 * Round 3


 * Draft 0: 3 votes
 * Draft 2: 3 votes
 * Draft 4: 3 votes

Votes needed for a majority: 5. Eliminating draft 2 (chosen at random).


 * Round 4


 * Draft 0: 5 votes
 * Draft 4: 4 votes

Votes needed for a majority: 5. Draft 0 wins.


 * Comments

Draft 0 won, but it was a very close call with draft 4. Draft 2 also saw considerable support, and if a different draft was eliminated in round 3, it would have changed everything. Of course, I cannot ignore the advice from several editors, including a non-participant who chimed in on the talk page, that if we do not include draft 0 we will open ourselves up to objections on procedural grounds. This would be the kiss of death for having a coherent RfC, and so I think it is clear that draft 0 should be our "status quo" draft. Although draft 2 was popular, I think we can set it aside, as it is (I think) identical to group 2 draft 0, and compromises on this general wording can be worked out on the group two talk page. The only real issue that remains in this group is whether draft four should be presented as a separate draft in the RfC, and if we choose to include it, whether it can compromise somehow with draft 3. This can be sorted out through further discussion.

Group two

 * LeadSongDog: 4 3 6 2 5 0 1 8
 * S Marshall: 4 6
 * Tryptofish: 5 1 0 3 2
 * North8000: 5 1 2 3 4 6 8
 * Unscintillating: 8 6 3 4 2 1 0 5 7 (0, 1, and 5 randomized)
 * Pesky: 5 1 3 8 6 4 2 0
 * Olive: 2 5 3 0 8 1 6 4 (last 6 randomized)
 * Blueboar: 2 5 3 4 1 0 6
 * Kalidasa: 8 6 0
 * Becritical: Taking these as all random per his comment, so discounting
 * Mangoe: 0


 * Round 1


 * Draft 0: 1 vote
 * Draft 2: 2 votes
 * Draft 4: 2 votes
 * Draft 5: 3 votes
 * Draft 8: 2 votes

Votes needed for a majority: 6. Eliminating draft 0.


 * Round 2


 * Draft 2: 2 votes
 * Draft 4: 2 votes
 * Draft 5: 3 votes
 * Draft 8: 2 votes

Votes needed for a majority: 5. Eliminating draft 2 (chosen at random).


 * Round 3


 * Draft 4: 2 votes
 * Draft 5: 5 votes
 * Draft 8: 2 votes

Votes needed for a majority: 5. Draft 5 wins.


 * Comments

This is the only group that didn't actually go to the last possible round, so by our standards it probably counts as a large majority. This tells me that we should probably work on refining the language of draft five to make it as good as possible, rather than try and find a way to incorporate the other drafts into it. This refinement process has kept going on the work group 2 talk page, so let's continue that work and submit it as one of our drafts. Draft 4 was popular, but it doesn't really belong in this group as it doesn't use the phrase "verifiability, not truth". However, it was similar to group 3 draft 7, which was also a popular choice, so perhaps we could consider combining them in some way. It strikes me that draft 8 is very similar to the idea put forward in group 1 draft 4, so I think the best way to proceed here would be to include it in the discussion about whether this should be a separate draft from group 1 draft 0.

Group three

 * LeadSongDog: 7 8 10 9 1 6 5 3 4 11 12 0 2
 * S Marshall: 2 12 8
 * Tryptofish: 8 5 2 3 4 12 11 10 9 6 1 0 7
 * North8000: 8 12 2 3 4 6 9 11 1 0 10 5
 * Unscintillating: 4 3 9 6 7 2 11 5 10 1 0 8 12 (unordered sets randomized)
 * Pesky: 12 10 9 11 3 8 5 4
 * Blueboar: 11 2 3 1 0
 * Kalidasa: 11 10 12
 * Becritical: 8 1 0 2 6 4 5 7
 * Mangoe: 2 7 3 (2 and 7 randomized)


 * Round 1


 * Draft 2: 2 votes
 * Draft 4: 1 vote
 * Draft 7: 1 vote
 * Draft 8: 3 votes
 * Draft 11: 2 votes
 * Draft 12: 1 vote

Votes needed for a majority: 6. Eliminating draft 4 (chosen at random from 4, 7 and 12).


 * Round 2


 * Draft 2: 2 votes
 * Draft 7: 2 votes
 * Draft 8: 3 votes
 * Draft 11: 2 votes
 * Draft 12: 1 vote

Votes needed for a majority: 6. Eliminating draft 12.


 * Round 3


 * Draft 2: 2 votes
 * Draft 7: 2 votes
 * Draft 8: 3 votes
 * Draft 11: 3 votes

Votes needed for a majority: 6. Eliminating draft 2 (chosen at random from 2 and 7).


 * Round 4


 * Draft 7: 3 votes
 * Draft 8: 4 votes
 * Draft 11: 3 votes

Votes needed for a majority: 6. Eliminating draft 11 (chosen at random from 7 and 11).


 * Round 5


 * Draft 7: 3 votes
 * Draft 8: 4 votes

Votes needed for a majority: 4. Draft 8 wins.


 * Comments

I hate to admit it, but the instant runoff process has been pretty much useless here. We have so many drafts, and so many people with different opinions, that voting doesn't do the thing justice at all. The top four drafts that our process has given us, drafts 2, 7, 8, and 11, are all quite different, and we don't seem to have come close to a consensus on how we could combine them. We should also consider if/how to include the drafts from group four that really belong in this group. Finding an answer to what we should include here might not be easy.

I think we need to try a new strategy. As has already been suggested on the talk page, instead of dealing at the level of drafts, let's deal at the level of design concepts. Once we can agree on how we should design the lede, or at least agree on what we disagree about, then it should be easier to translate these design concepts into one or two drafts. So, here are a few common threads that people have brought up which I think we should discuss:


 * Whether we should differentiate between the viewpoint of the editor and the viewpoint of the reader.
 * Whether the question of what to do with inaccurate material falls outside the scope of the verifiability policy or not.
 * Whether we should include text that says objective truth should have no influence on what goes into Wikipedia.
 * Whether we should specifically mention other policies in the lede, whether we should just note that "other policies apply", or whether we should not mention anything about other policies at all.
 * Whether we should add/change sections other than the lede.

I think this covers the main differences in the drafts, but if anyone notices something else that we need to add to the discussion, we can do that.

Group four

 * LeadSongDog: 0 4 1 2 5
 * S Marshall: 0 1 4
 * Tryptofish: 4 1 2 0 5
 * North8000: 4 0 5 1 2
 * Unscintillating: 5 3 0 2 1 4
 * Pesky: 5 4 2 1 0
 * Blueboar: 4 5 2 1 0
 * Kalidasa: 5 0 4
 * Becritical: 0 2 1 4 5
 * Mangoe: 4 0
 * Bob K31416: 0
 * Olive: 4 1 2 0 5


 * Round 1


 * Draft 0: 4 votes
 * Draft 4: 5 votes
 * Draft 5: 3 votes

Votes required for a majority: 7. Draft 5 eliminated.


 * Round 2


 * Draft 0: 6 votes
 * Draft 4: 6 votes

Votes required for a majority: 7. Winner would be determined at random from draft 0 and draft 4.

I thought that it would be difficult to choose a draft from this group, but it has turned out to be very easy. On the face of things, we have three drafts with around the same degree of popularity, drafts 0, 4, and 5. However, drafts four and five both mention truth and how it differs from verifiability, so they really belong in group three, not here. So, draft 0 seems like the logical choice of draft to use from this group. (Plus, Bob K31416 said that he couldn't improve on it, and BeCritical said it was beautiful - that has to count for something.) So let's base further discussion for this group around how/if we should tweak draft 0. We shouldn't ignore drafts 4 and 5 altogether, though - let's incorporate them into the discussion for group three.

By the way, I welcome people checking my calculations with the instant runoff. I tried to be careful, but it's always possible that I made a mistake. —  Mr. Stradivarius ♫ 21:40, 8 April 2012 (UTC)

Step six
Step six is going to be quite different from the description I gave in the agenda. It is clear that we are going to need more discussion before we can decide on what the final drafts will be, particularly in group three. Based on the results of step five above, I have a rough outline of what each group needs to discuss for this step.


 * Group 1. There seems to be a strong consensus for draft 0 to be presented in the RfC, so draft 0 itself no longer needs discussion. What I would like you to discuss is whether draft 4 should be presented as a seperate draft, and whether it should be combined somehow with draft 3 and group 2 draft 8.
 * Group 2. Draft 5 has gained a rough consensus, so let's focus on improving it now, rather than trying to incorporate radically different drafts. This process is already underway on the talk page, so let's just continue that until we can reach a solid consensus on the wording.
 * Group 3. There is an awful lot still to be worked out here. Let's agree on the general design concepts to use in the policy first, and then see if we need to split some versions out into new groups. I will post more instructions on the group 3 talk page.
 * Group 4. There seems to be a rough consensus to use draft 0, so discussions now should focus on how/if we can improve it.

Seeing as the deadlines that I have been setting up to this point have had quite a tenuous link with reality, I won't be setting a formal deadline for step six. Instead, I will monitor the progress of each work group and set deadlines only if necessary. This may include setting up structured discussions with deadlines for individual work groups (sort of like a step 6a, 6b, etc.). But for now, I will just set up the discussions and allow you all some time to work things out. I will be posting instructions at the different work groups soon. As usual, if you have any questions or concerns, please bring them up on the mediation talk or on my talk page. —  Mr. Stradivarius  ♫ 21:28, 8 April 2012 (UTC)