Wikipedia talk:Measuring conflict of interest editing on Wikipedia

bluerasberry's initial thoughts
I set this page up. I thought that I would share my initial thoughts. I do not have much idea who might accept or reject this perspective. I expect that most Wikipedians have not thought about any of this one way or the other.


 * I do the initial survey for this project
 * 1) What is your best guess for how often the volunteer team in this project oversees conflict of interest editing events?
 * Among all English Wikipedia projects, I guess that English Wikipedians address 500 COI editing events every day.
 * 1) Imagine all the conflict of interest editing events you have seen in this project. Divide them into two groups, "acceptable or productive contributions" and "unacceptable or unproductive contributions". What is your guess for the percentage of conflict of interest editing events which are "acceptable or productive contributions"?
 * My guess is 30% acceptable.
 * 1) What is your best guess for how many hours Wikipedia community volunteers spend in this project addressing conflict of interest editing events?
 * Throughout all of English Wikipedia, my guess is 125 hours every day, or 15 minutes on average for each of the 500 events.
 * 1) Imagine that Wikipedians in this project with your level of experience provide 10 hours of support to other volunteer Wikipedians. What is your best guess of how many hours of "acceptable or productive contributions" Wikipedia is likely to get in return for that 10 hour investment?
 * 3x, so 30 hours
 * 1) Imagine that Wikipedians in this project with your level of experience provide 10 hours of support to conflict of interest editors. What is your best guess of how many hours of "acceptable or productive contributions" Wikipedia is likely to get in return for that 10 hour investment?
 * 0.1, or 1 hour
 * 1) In 2017 English Wikipedia has about 40,000 active editors. In terms of Wikipedia experience, where do you think you rank among editors? Bottom 20%, second 20%, third 20%, fourth 20%, and or top 20%?
 * top 20%
 * 1) Please sign your username.
 *  Blue Rasberry  (talk)  19:36, 2 November 2017 (UTC)

I think that Wikipedians at OTRS, AFC review, and the COI boards on wiki collectively interact with about 100 COI editing events every day. About 50 people are in those projects in any given day. I am guessing that these processes detect only 20% of the COI editing events, so there are 500 total every day. The other guesses that I made above derive from my perspectives on this information.
 * Other thoughts

The 100/day goes back some time, so maybe there are 30,000 events a year. I feel like these events were less common in the past, so maybe over the past 10 years, there have been 200,000 of these events. Maybe these on average consume 15 minutes each for Wikipedians to think about, stress over, actually take action, or whatever else, so my guess is that there has been 200,000 * 0.25 hours = 50,000 hours or 2080 days of volunteer labor directed to this effort. If Wikipedia volunteer time is worth US$20/hour, then there is a $1,000,000 value investment in this effort.

I could be wildly off about any of this. Making small changes to these numbers could change the outcome and the implications. It is really challenging for me to know how to think or feel about this issue, its scope, or its implications.  Blue Rasberry  (talk)  19:36, 2 November 2017 (UTC)

First cut at arithmetic
From this essay: "anyone can find a place which discusses conflict of interest editing, count all relevant discussions in the past 3 months or year, and classify them in an insightful way. Doing this and sharing the result could provide starting data for determining the scope of the issue." Please see User:Bri/COIN workload analysis for a first cut at such an analysis of COIN performed over a shorter interval of 30 days.

I did some more math at WT:Paid-contribution disclosure to conclude this:
 * We can extrapolate to ~2,250 editors per annum assuming October 2015 was representative. This was a look at all COI matters, whether or not paid, but I'd estimate 85 to 95% are direct paid editing issues involving enterprises' contractors or employees anyway. Factor in the rate at which we actually notice these editors versus how many remain underwater and you're looking at tens of thousands of COI accounts per year, minimum.

Another takeaway from that analysis was that one editor (me) was introducing about 90% of the new cases. It's smoothed out a little since then, but it's apparent that with the loss of one or two volunteers in this area, the whole process could go of the rails.

Other editors contributed their thoughts too, now archived in the "size of problem" thread begun by 21 September 2017. ☆ Bri (talk) 05:38, 12 January 2018 (UTC)
 * Thanks User:Bri. Whenever I look at cases, I am always amazed at how extensive the problems are.
 * User:Bluerasberry was looking for these results aswell. Doc James  (talk · contribs · email) 20:55, 12 January 2018 (UTC)

Presentation in comments of an editorial in The Signpost
I linked to this essay with a comment in this article
 * Reaching six million articles is great, but we need a moratorium

 Blue Rasberry  (talk)  16:00, 27 January 2020 (UTC)