User talk:Ocaasi/coiquestions

Comments

Question 2

Not all paid editors have a conflict of interest (for example, an university professor writing about their area of research) slightly tweaked from the original

This seems kind of confusing -- a university professor writing about their research might have a conflict of interest, but is probably not a paid editor. Should there be a different example or should the first part be switched? a13ean (talk) 19:21, 8 March 2013 (UTC)

I changed the example to Wikipedian in residence. They wouldn't be regarded as paid advocates or editors with a COI. SlimVirgin ^(talk) 19:27, 8 March 2013 (UTC)

Thanks, that's better, especially the museum part as some of our newer WIRs are in more gray area. Ocaasi ^{t | c} 20:08, 8 March 2013 (UTC)

"Paid Advocacy"

This language risks confusion and I preferred the original phrasing. There are paid editors with and without COI. There are paid editors with a COI who engage in advocacy and those who do not. To use the term "paid advocacy" implies all of them they are actually behaving or editing non-neutrally. Thus I prefer we describe what their position, not its category. Ocaasi ^{t | c} 19:37, 8 March 2013 (UTC)

A Venn diagram of people who are paid to edit, people who have a COI, and people who engage in advocacy with examples and relevant wikilinks in each might be handy. Edit: perhaps not in the RFC, but somewhere in WP:COI or similar. I'll think about it. a13ean (talk) 20:13, 8 March 2013 (UTC)

I do think that would be useful. Paid editors, financial conflict of interest, and advocacy are variously interconnected. I'd like to see that diagram to help clarify terminology. Ocaasi ^{t | c} 20:27, 8 March 2013 (UTC)

Definition of advocate

Would you please define the term advocate? This set of questions is currently using it. Blue Rasberry (talk) 18:37, 10 March 2013 (UTC)

Argument to moderation

The way this is drafted is some form of the argument to moderation fallacy. The inevitable result will be a strong bias toward the middle options. It forces the debate into a single dimension, and frames certain positions as extreme. As an example, we could make the top options of Q3: "take legal action against paid advocates, getting injunctions against further editing" and maybe below that we could have "send cease and desist letters to paid advocates' employers" followed by the "site ban" option. Adding those two options would skew the results toward harsher options, which is why this is poll is set up in a way that invokes an appeal to moderation. Gigs (talk) 19:44, 8 March 2013 (UTC)

How can you elicit a range of opinions without suffering that result? Would it be better if we mixed up the order so that it was not always sequential? Also, what if the community's view is in fact moderate? How can we differentiate that from confounding phrasing? Suggestions welcome. BTW, my original approach was to provide more than 3 options, ideally more like 5, so that editors could choose some 2's and 4's not just all 3's. That seems to have not been well received either. Polling is always a challenge. Ocaasi ^{t | c} 19:55, 8 March 2013 (UTC)

Policy is more or less set by practice, not by polling. Our current practice is to block some advocate editors on sight, (usually ORGNAME blocks, but sometimes hard spam blocks), but to tolerate the ones like Corporate Minion who disclose their COI and work within the system. That kind of subtlety will never be represented in a linear scale of options. It's still framing the issue in a fallacious way to ask "What if the community's view is moderate", because that assumes there is a linear spectrum of options. Gigs (talk) 19:56, 8 March 2013 (UTC)

I'm trying to firm up our conception of what the current practice is and should be, if possible. I fully accept that we might have a bimodal or other irregular distribution of opinions on this matter. Hopefully the questions can gauge that as well, even if they don't come to singular conclusions in all cases. Ocaasi ^{t | c} 20:10, 8 March 2013 (UTC)

I feel confident in Gigs' assessment that we tolerate COIs that act reasonably within the system and shame/discourage those that don't.

What WP:NOPAY should do is clearly define what "operating within the system" means, how to do it, the repercussions of not doing it, and how volunteers should behave when working with editors who either are or aren't working within the system.

But "the system" is tough to define clearly, because the range of COI circumstances are diverse and it requires good judgement that few COIs possess. We also have IAR, NOTBUREAU, etc.; in a sense, Wikipedia has no "system" really, so there are competing philosphies between the COI's need for clear instructions and Wikipedia's desire to operate differently. CorporateM (Talk) 20:52, 8 March 2013 (UTC)

Question

User:Northamerica1000 had a similar effort a while back to boil down COI into a series of questions in order to get us out of cyclical arguments and attempt to discern consensus. He would be a good one to notify.

Question: Ocaasi, how bold do you want me to be here? CorporateM (Talk) 19:50, 8 March 2013 (UTC)

Thanks, I'll ping him/her. I suppose...be as bold as is useful while maintaining the scope of the questions. If you're unsure, post to talk first. Ocaasi ^{t | c} 19:53, 8 March 2013 (UTC)

Ok, I just gave it a close and thorough look-through top-down and made edits. I'm sure it will get completely re-written several times over, but that's my once-through. I'll check-in later on on the discussion. Thanks for inviting me. One issue is that most of these don't have clear (a), (b) answers, becuase they are complex and nuanced, which is part of why it's so difficult to provide concrete guidance. A lot of the questions/answers need more culling through and polishing, but I think it will continue to improve. CorporateM (Talk) 20:59, 8 March 2013 (UTC)

How to conduct a poll

The first requirement of any poll like this is setting out the mechanics so that folks can be sure their input is properly reported, anonymous, not overwhelmed by anons or people who just got a username, etc. But if POTY can have a fair polling/voting system, I'm sure we can as well. The previous RfCs on this matter have been totally useless, because the format and results were poorly organized, and a few editors can dominate the discussion. So formatting needs to be very carefully done.

Unreported results can be a major problem, e.g. I seem to remember that there was a poll - pretty biased on its face - about uncivil behavior where results were never reported. Not reporting those results could be viewed as a serious breach of civility. I also remember somebody who was going to survey all Fortune 500 company articles and count the mistakes - never reported. I certainly don't want to be involved in anything along those lines. Mechanics matter.

The poll should be well advertised, e.g. by a banner and have something like "open to all editors who have made 100 edits before March 1, 2013." With the date and method of reporting the results clearly stated. Keep it under 10 questions - the more questions given, the more interpretations will be given. 5-7 questions looks right to me.

I don't like the "Other, explain below" answers. How are we going to interpret those? Perhaps we give the option on every question to further explain or comment below the question. The likely interpretation of those will be that not enough people commented to affect the overall results, but for a particular question it might reveal bad wording, an obvious omitted choice, or a key question that needs to be revisited.

The number of possible answers, should be small and always given in the same order (or perhaps be 50% of the time in the original order and 50% in reverse order). Most questions should be "linear". The ones that suggest "There should be various options depending on the situation" or "if this, then that, but otherwise..." will in the end not tell us much. There certainly should however be at least one "non-linear" question about how much of the policy should be left up to editors' judgement and how much should just be automatic "no exceptions"

4 possible answers for each question will help avoid the common tendency of people to "Just choose the middle one"

I don't think the question of what to call these people is very useful. Call them "kjabhtsei" for all I care.

I'll suggest that question 4 be put at the top, as it is the most important IMHO.

Straightforward questions and good mechanics will be the key. Smallbones_(smalltalk) 03:18, 9 March 2013 (UTC)

I actually wonder if this whole thing isn't too ambitious - to propose such a sweeping RFC all at once covering 5-10 questions, when it is unlikely we would even be able to achieve consensus even just regarding the RfC itself. I noticed Ocaasi uses "corporate representatives," I added "on behalf of..." and Slim is adding "paid advocates." Maybe it would be best to focus on just one question at a time. CorporateM (Talk) 01:40, 10 March 2013 (UTC)

Perhaps I was misled into thinking this was a poll by its format, when an RfC was actually proposed. Polls and RfCs don't mix - you simply can't have a poll inside an RfC, because in RfCs the questions and proposals change all the time with each new editor adding his opinion. A poll needs to be in a fixed format.

I do think a poll can lead to an RfC. So e.g. if 90% of the respondents say that paid editors should be required to disclose their employers, and 91% say that they should be banned from the article pages where they have a COI, then it would likely be pretty easy to get an RfC to decide on a new policy. Smallbones_(smalltalk) 02:22, 10 March 2013 (UTC)

I think this is not going to work as proposed. Ideally, we should just ask one question. Something like: "Should paid advocates – defined as legal, corporate and public-relations representatives, and the writers they hire – be allowed to edit affected articles directly? (Note: this question is about substantive edits; there is no objection to paid advocates correcting unambiguous factual errors and addressing genuine emergencies such as libel). For the sake of clarity, please answer yes or no." SlimVirgin ^(talk) 19:04, 10 March 2013 (UTC)

On the one hand, the definition is arbitrary as Smallbones points out, but it's difficult for the community to even discuss it without agreement on its name. I boldly put in just one question RE definition. I think if there is agreement to just start on this question for now - we could work on three potential candidates for folks to discuss. CorporateM (Talk) 21:00, 10 March 2013 (UTC)

Bias

Those of you in favour of paid advocacy are loading the questions. If you do that, the RfC will be inconclusive and pointless. Therefore, I suggest you stop trying to introduce doubt, and ask questions that will allow decisive answers. Also, please stop adding straw men. No one has said paid advocates are not allowed to fix typos, or make edits in emergencies. SlimVirgin ^(talk) 16:52, 9 March 2013 (UTC)

Support for Wikipedians

Would love to see asked:

1. Do Wikipedians support fellow editors reporting cases of conflict of interest to the media / journals for further follow up? (this would over rule the outing policy)
2. Editors with COI (the head of marketing departments for large corporations for example) who get people paid by the company in question to send emails to the work colleagues of Wikipedians can be indefinitely banned

Doc James (talk · contribs · email) (if I write on your page reply on mine) 23:47, 9 March 2013 (UTC)

It's something that needs to be discussed. If there's to be an increase in paid advocacy with the support of some Wikipedians, what are the rest of us allowed to do to (as we see it) defend the project? Is it acceptable/advisable to go to the media, and if so in what circumstances? Where the advocates have identified themselves, there's no outing issue, but where they haven't, how do we handle it? As things stand, it seems that we're expected to oppose paid advocacy with one arm tied behind our backs because we're volunteers who aren't being paid, and the second arm tied because we're expected to assume good faith, and never risk outing someone. In the meantime, as we dither, articles get turned into ads. SlimVirgin ^(talk) 00:12, 10 March 2013 (UTC)

I'd like to see that question raised as well. Currently the WP:OUTING policy overrules the WP:COI guideline. Exceptions were made through WikiScanner in 2006-2008, but there is a disconnect in our policies. I believe if an editor comes to our talk pages and discloses their conflict of interest then that information is public. On the other hand, if an editor does not reveal their real name or position, then we are in more of a bind regarding the outing policy. We've had very good editors banned for pushing the outing policy past its coi limits and we don't want that to repeat. Ocaasi ^{t | c} 01:43, 10 March 2013 (UTC)

I've always been opposed to using the media to deal with anything; I took the attitude that what happens onwiki stays onwiki. But increasingly I think we may have to use real-world dispute-resolution to protect the project from paid advocates. The Advertising Standards Authority in the UK is one option: when British companies edit their own articles, or persuade Wikipedians to proxy for them, does it count as covert advertising? I'd be very interested to know how they would rule in a case like that. So yes, if we could come up with a question about whether editors support the use of media/journals/off-wiki dispute resolution, that would be worth adding. SlimVirgin ^(talk) 18:40, 10 March 2013 (UTC)

I would ask this differently - more broadly. The question is how to deter bad behavior, without violating Wikipedia's principles (or at least with the minimum sacrifice possible). Media is not the only tool, though it's the one that's been used historically. In my view it hasn't worked well, but the community also hasn't proactively managed it. CorporateM (Talk) 20:55, 10 March 2013 (UTC)

I've added a draft of the question. It's a tricky one to ask, but the nub of it relates to Outing. Feel free to revise. Ocaasi ^{t | c} 17:31, 11 March 2013 (UTC)

I've removed this because it was written very problematically, e.g. (c) was the same as (a). I also removed the advertising bodies, because it would be the companies who were reported to them, not the hired hands. It could read something like this:

(a) It is legitimate to alert the media (or other interested outside bodies) to occurrences of paid advocacy, only if the paid advocates themselves have disclosed the client/employer relationship.

(b) It is legitimate to alert the media (or other interested outside bodies) to occurrences of paid advocacy, even if it has not been voluntarily disclosed but has been discovered by other editors.

SlimVirgin ^(talk) 18:11, 11 March 2013 (UTC)

Writing

There's a problem with the way this is being written. It needs to be either-or, and the options need to be related. For example, should they be allowed to edit if they ask a question and get no response?

• (a) be permitted to edit affected articles directly if no response is forthcoming
• (b) should expect timely responses for certain things (blatant coatracking, BLP, important corrections), but not for others
• (c) there is no site-wide obligation for volunteer editors to respond

Someone could easily agree with all three; and of course there is no site-wide obligation -- how could there be? The distinctions need to be sharp and directly related to the question, with no tangents, no straw men, and no tautologies. SlimVirgin ^(talk) 17:59, 11 March 2013 (UTC)

So we should make every question a binary choice? I support your efforts to make the questions useful, but I also think there's such a thing as oversimplifying. I agree that the questions are best when a single answer 'best represents' the commenter's views without overlap, but that can still be achieved with 2+ options. Go ahead and suggest improvements, that's what this process is for. Ocaasi ^{t | c} 18:42, 11 March 2013 (UTC)

Reading this again, I think the construction is (almost) irredeemably biased, and I see no clear way to remove it without introducing bias in the other direction. If you offer three options, people will drift to (b), so you have to be careful of that (or whichever is seen as the middle ground). But if you have only either-or, you're right, it sometimes over-simplifies. So the key is to have extremely carefully worded options that don't introduce subtexts. Creating that will involve a lot of work. SlimVirgin ^(talk) 19:35, 11 March 2013 (UTC)

Ocaasi, I see you've again restored one of the middle-ground (b) options. If we write it this way, we will undermine it. Look at this example:

Please say which you prefer:

(a) Paid advocates are not allowed to edit affected articles in any circumstances.

(b) Paid advocates are welcome to correct simple errors and remove serious issues such as libel, but should otherwise avoid editing affected articles.

(c) Paid advocates may edit affected articles the same as everyone else.

A lot of people would naturally incline toward (b), because it's the middle ground, and because it contains the nice word "welcome." That would give us a result that would confirm the current NOPAY guideline.

We have to avoid these tricks if we want an honest RfC with results that will stand. SlimVirgin ^(talk) 01:44, 12 March 2013 (UTC)

I would look at it in the reverse. What would be a trick would be to force respondents into a black-white response on a complex and grey topic, where consensus lies somewhere in-between. On the other hand, "welcome" is a bit POV. CorporateM (Talk) 01:54, 12 March 2013 (UTC)

I think you'd be right sometimes, i.e. sometimes either-or would force people into one of two extreme positions. But sometimes the middle-ground option would lure them into that for the wrong reasons – the way supermarkets will have a cheap, middle and expensive option; they expect people not to pick the lowest or highest price. This is why I said that presenting a multi-question RfC would need some careful writing. Maybe we should find a professional pollster who knows all the tricks and how to avoid them. There's bound to be one among all these PR people. :) SlimVirgin ^(talk) 02:12, 12 March 2013 (UTC)

It seems like SV is echoing my concerns above. I compared it to the Argument to moderation fallacy, though I can understand the position that the extreme options are straw men, since most of them haven't been seriously proposed by anyone. The bias this introduces toward the middle options will only generate a lot of nodding about things people already agree on, and little resolution on the controversial details. Gigs (talk) 15:21, 13 March 2013 (UTC)

I feel confident in the assessment that the extreme options are not credible outcomes. If taken literally, a BrightLine philosophy suggests blocking any suspected COI that edits article-space, even if they are correcting errors or grammar. I would defend my privilege to correct my own mistakes, make janitorial edits, make edits specifically asked of me, etc. On the other hand, telling PRs to edit freely is equally extreme.

My suggestion would be to eliminate the extreme options that represent opposing political camps entirely and expand with 4-5 options dedicated exclusively to the grey area. What can plausibly be explored are things like "advised" versus "expected." Proposing we create extreme policies from within a guideline is just a waste of time. CorporateM (Talk) 15:33, 13 March 2013 (UTC)

A semantic approach

To beat my drum from several months ago again, I think we need to focus on vocabulary and definitions. Without common terminology, rational debate on this subject is impossible. I disagree with Smallbones who asserts that the names aren't important, I think they are of crucial importance. Without rehashing too much of the previous debate, this basically falls into 2 main areas:

• Roles and status of editors
  • Definition and universal names for various COI-affected statuses and external roles/obligations

• Behavior of editors
  • Define rules of engagement (or guidelines) for each different type of status (not one size fits all)

As an example of the flaw in the current draft, the final question attempts to come up with one name for everything. I think we need to break it all down. One reason we get people with vastly different opinions when we run RfCs on this is because everyone has a different idea in their head of the type of user we are talking about. Or they advocate a very tolerant universal approach because they feel like we are putting draconian restrictions on a certain type of editor that they don't find offensive or counterproductive. This is a symptom of trying to come up with a one-size-fits-all approach to both classifications and rules/guidelines. Gigs (talk) 15:33, 13 March 2013 (UTC)

My take for what it's worth:

COI definition

Marketing, financial connection and advocacy Editors that have a financial connection with the subject of the article, such as an employer or client relationship, are advised to.... In addition to public relations, search engine marketing, reputation management, and Wikipedia consultants, anyone assigned to participate on behalf of an organization or initiative with a potential conflict of interest should follow this advice. This includes editors contributing on behalf of an organized protest, class-action lawsuit, or as a volunteer at a non-profit advocacy group.

CorporateM (Talk) 16:31, 13 March 2013 (UTC)

Any description of problematic paid COIs has to exclude Wikipedians in residence, because the community is happy for that to continue, including the paid roles. The reason those roles are supported is that they're seen as acting on behalf of Wikipedia, even though they are helping the museums. That is the crucial distinction: Wikipedians should be editing on behalf of Wikipedia (even though we all have our pet issues, peeves, POVs, interests, etc). But paid advocates are editing on behalf of someone else (even though they might argue that what they are doing helps Wikipedia too). Any definitions have to include this concept of primary and secondary roles. SlimVirgin ^(talk) 17:30, 13 March 2013 (UTC)

Yes, the distinction lies here: "on behalf of an organization or initiative with a potential conflict of interest". While GLAM a similar relationship to an organization without a COI, so they are not contained in the definition. CorporateM (Talk) 17:55, 13 March 2013 (UTC)

It's the editors who have the COI, not the company, so it has to be worded differently. As the moment, your definition reduces to tautology: they have a COI if they have a COI. SlimVirgin ^(talk) 18:09, 13 March 2013 (UTC)

Lol - touche. Any suggestions? CorporateM (Talk) 18:20, 13 March 2013 (UTC)

It depends what you want to define, whether it's paid COI or COI in general. There's are some books I keep meaning to read about how COI is defined offwiki. I'm going to force myself to do that so we can apply whatever vocabulary is used elsewhere. SlimVirgin ^(talk) 18:41, 13 March 2013 (UTC)

I've always felt the better comparison to make is with business terms - to look at how PR is done with other mediums, as oppose to COI. A COI puts us in the framework that they are an editor who "just so happens" to have a potential competing interest, as oppose to doing PR. I've never called a journalist and said "I have a COI with your readers and I just want to disclose that" - even if it's technically true... CorporateM (Talk) 19:27, 13 March 2013 (UTC)

I can't figure out would it mean to have a COI with a newspaper's readers. Can you elaborate? SlimVirgin ^(talk) 19:35, 13 March 2013 (UTC)

The best analogy I can think of is as follows. A judge may have a COI with "justice", say if the defendant is an old friend from High School, but the lawyer does not have a COI with justice because he represents his clients' interests. Technically, because he represents his clients interests and not those of justice, he has a COI, but this is not the conceptual framework we use. We accept that the lawyer represents their clients interest, but we also expect them to act ethically, not withold evidence and to let the judge make the decisions.

By the same token, NYT's readers want impartial, independent information, while PRs want content aligned with corporate messaging. But PRs know they can only work their magic by influencing reporters, not write themselves into NYT however they well please. CorporateM (Talk) 16:11, 14 March 2013 (UTC)

PRs and the readers of the NYT have different aims, yes, but it would be odd to think of it in terms of COI.

A COI is not a point of view; you could have a COI and have no point of view about the issue in question. Similarly you could have a point of view and not have a COI. A COI is only concerned with roles and relationships.

A COI is where there is a relationship in your life that makes you not trustworthy for the role in question. For example, a judge would have a conflict of interest if her husband were the accused (not a conflict with an abstract thing, "justice," but a conflict between her two roles: judge and wife). She would have that COI whether she believed him innocent, guilty or didn't know or care; she might even be glad to see the back of him. But her role as judge is undermined by her role as wife, so she can't do both. It has to do with the way things look, not necessarily with the way things really are. It also has to do with our own inability to separate out our roles in our minds. Maybe the judge thinks she can be perfectly neutral when her husband is the accused, but we say no, we would never be able to trust the decision she reached. So where there is a clash, we don't arrogantly assert that we can handle that clash. Instead, we acknowledge that perhaps we can't handle it, and we recuse.

That's what we mean by a COI, where the primary role of Wikipedian (the primary role from our perspective) is undermined by the secondary role of corporate communications officer for a company (which is no doubt the primary role in that person's mind). SlimVirgin ^(talk) 16:33, 14 March 2013 (UTC)

Exactly - you explained it better than I: "PRs and the readers... have different aims... but it would be odd to think of it in terms of COI."

The judge may have a COI, because they have two competing roles as a friend and a judge, but the lawyer cannot have a COI, because they have only one role, to support the interests of their client. The lawyer may have a different interest than the judge's, but that's different from COI. CorporateM (Talk) 22:40, 14 March 2013 (UTC)

Right. The lawyer's role is to defend his client, the corporate communications officer's role is to please his employer, the role of the freelance writer being paid to edit Wikipedia is to satisfy the person who hired him, and the PR person's role is to make his client look good. None of these roles involve putting Wikipedia first. That's why there is a conflict of interest, and that's why it's not good for these people to edit affected articles, or have Wikipedians act as proxies willing to copy-paste their drafts word for word. SlimVirgin ^(talk) 21:28, 15 March 2013 (UTC)

There was a time in my life where my wife and I were meeting with a lawyer every week. He pulled together all the evidence(reliable sources) and helped us be neutral about offering a reasonable settlement (Wikipedia article) based on the likely legal outcome (content policies). Our priority was to avoid lengthy and resource-intensive arguments or court time (dispute resolution) and reach a reasonable resolution ASAP. The fact that he offered the settlement as a representative of our interests didn't make it less valid.

This is a much more compelling analogy than COI. Lawyers have only one role, to serve their client, but if they and their client do all the heavy-lifting and offer an article ready-for-approval off-the-bat, this is the best way for them to serve their clients.

But the real point is that my role is more comparable to the role of a PR person with the media, a lawyer, etc. than it is with COI. We could chase these analogies down a lot of pathways, like how a PR person relies on a journalist to balance their spin, or that a lawyer is ethically required to disclose all available evidence. And every client is different in how neutral they can be and what neutral looks like for the topic. CorporateM (Talk) 22:20, 15 March 2013 (UTC)

I don't see what the analogy is between WP and the example you gave. A lawyer has to act in his client's best interests (within the limits of the law, professional ethics, etc). He would not write a Wikipedia article that put them in a bad light. He would do his best to spin things as positively as possible. That much is clear. And so we have a rule that says, if you have a real-life role that clashes with your role as Wikipedian, please don't edit in affected areas. Lots of organizations have rules like that, and there's no need for us to be any different. SlimVirgin ^(talk) 22:26, 15 March 2013 (UTC)

I'm not contesting the brightline; on the contrary, the conceptual framework we have now: "you're a Wikipedian who happens to have a COI" lends itself to direct editing, while if we look at the role of people who represent their clients in politics, sales, media or law, none of these people can write their own verdicts. In each case, we're required to provide value to an independent party or get ignored. Astroturfing Wikipedia is illegal, but when one follows the law, it is in their best interest to provide content that will get approved, which is synonymous with doing their best to be neutral. It is only due to the approval cycle that such an interest can be aligned, as it is with any of the other paradigms. 00:01, 16 March 2013 (UTC)