Jump to content

Wikipedia:Village pump (proposals)/Archive 215

From Wikipedia, the free encyclopedia

Wikipedia donation ads

in the school I go to, they have chromebooks for Wikipedia but they are reset every time you log on so every time i go to wikipidia I always get the annoying Wikipidia donation adds, and even though I press x they come back the next time I log on, which can be many times in one day.

I am proposing that the poppup system should ask for donations bassed off of the IP adress and not bassed of of cookies. Instead of adds automaticaly popping up when wikipidia is oppened for the first time on the browser, make them poppup when its oppened for the first time on the IP adress or user. — Preceding unsigned comment added by YisroelB501 (talkcontribs) 09:59, 29 November 2024 (UTC)

I have experienced similar issues when I've not been logged in, and they've definitely been getting more aggressive as of late. IP-based messages would be sensible, in my view. JayCubby 21:50, 29 November 2024 (UTC)
According to the Wikimedia Foundation, their banner fundraising campaign only runs once a year and doesn’t start until December 2nd, so it’s unclear where any ads you or I have seen before that date come from. Perhaps there’s some sort of security issue. 216.147.127.204 (talk) 21:56, 29 November 2024 (UTC)
Every country has a different schedule. Also, they test the system beforehand.
@YisroelB501, I agree that it sounds annoying, but unfortunately, I think fixing it would take more than a few weeks, and the main fundraising campaign only lasts until the first few days of January. So I think you'll keep seeing this problem for a while yet. @JBrungs (WMF) can inquire with the fundraising team about whether this would be feasible. WhatamIdoing (talk) 19:54, 30 November 2024 (UTC)
I’m in the United States, which is one of the countries where the Wikimedia Foundation says its fundraiser hasn’t started yet. The messages I saw looked just like a legitimate fundraiser, not test messages, and I didn’t sign up for any testing. I don’t know who is putting these messages up on Wikipedia or where the money is going. 216.147.127.204 (talk) 21:00, 30 November 2024 (UTC)
These are legitimate fundraising requests. They periodically run tests of the real thing (e.g., to see if they get complaints about the wording being confusing, to make sure that the credit card processing system works, etc.). If they had to change anything recently, they'll definitely be running tests right now, because they want everything to work on "Giving Tuesday". "Giving Tuesday" is the response from the non-profit world to Black Friday, Small Business Saturday, and Cyber Monday.
The money goes to the Wikimedia Foundation, a 501(c)(3) non-profit organization that supports Wikipedia and other projects like Wiktionary and Wikimedia Commons (where almost all of our photos are stored and organized).
If you want to donate, you should go to https://donate.wikimedia.org and do so. And if you don't, you should click the little button to make the banner go away for a week (or until you clear the cookies in your web browser). If you really want to make it go away, then go to Special:CreateAccount and create a free account (pick a username/password; no e-mail address or anything else required). After you're logged in, go to Special:Preferences#mw-prefsection-centralnotice-banners and turn off all fundraising banners. That will suppress all fundraising banners until you get logged out. (And if you do get logged out, just log back in again, and your prefs settings will take over again.) You can use the same account on all your devices, as long as you can remember the username and password. WhatamIdoing (talk) 22:24, 30 November 2024 (UTC)
The wording certainly was confusing, because I didn’t even know that these were tests! I had assumed that the Wikimedia Foundation was running its annual banner fundraising campaign until I saw a statement by a WMF employee that the WMF was not doing that. I couldn’t have complained about the wording’s being confusing because I wasn’t told that there was a test going on or where to make any reports about it.
Looking more closely at the “community collaboration page” and its mentions of testing, I gather that these are psychological tests seeking to determine which banners induce the most donations. Conducting such experimentation on human subjects without obtaining their informed consent is a serious breach of ethical standards. (Obviously there was no informed consent, because I wasn’t even informed!) Even the WMF’s own Code of Conduct calls such “psychological manipulation” an “unacceptable behaviour”.
I’d like to formally report this violation of the Code of Conduct, but I can’t find a way to do that. I see that the Code of Conduct has enforcement guidelines, but they contain vague language like “Reporting of UCoC violations should be possible by the target of the violation”…so, um, is reporting of UCoC violations possible, and if so, how? 216.147.127.204 (talk) 00:37, 1 December 2024 (UTC)
The UCoC says it is an abuse of power to engage in:
  • Psychological manipulation: Maliciously causing someone to doubt their own perceptions, senses, or understanding with the objective to win an argument or force someone to behave the way you want.
Did you actually doubt your own perceptions, senses, or understanding? Was producing those doubts (if, indeed, they were produced) in aid of someone winning an argument against you or forcing you to behave the way the other person wanted?
If you can't answer yes to both of those questions, then you've got no complaint to make under the UCoC.
I suggest reading up about the words you're using, like Informed consent, and maybe the difference between Psychological testing of a person and A/B testing an advertisement. WhatamIdoing (talk) 00:00, 3 December 2024 (UTC)
Tracking page customization information by IP address would require storing that information, and would slow down serving pages to non-logged in users, as they can't just be served a pre-composed page right from the cache servers. The only cache-friendly approach is for the client to hide the messages, which means using cookies. Plus, some network configurations have IP addresses shared between users that need to be supported. isaacl (talk) 22:49, 30 November 2024 (UTC)
Perhaps simply waiting a bit to show donation prompts at all would solve the issue (and prevent it from annoying people who have their cookies cleared all the time). JayCubby 03:01, 1 December 2024 (UTC)
If the delay is long enough that people won't see the donation banner, then the purpose of the banner is defeated. A slight pause will either leave a blank space for a period of time, or shift the page layout after the pause. Both of these are arguably more annoying than displaying the banner without a pause. isaacl (talk) 04:09, 2 December 2024 (UTC)
What I mean is a delay of days before the banner is shown to users, rather than bugging them on the first time a user goes to Wikipedia. JayCubby 17:19, 2 December 2024 (UTC)
The problem is that if IP addresses are tracked, then that information on how recently and how often an IP address has been used to access WP will have to be stored somewhere and retrieved every time that IP is used to access WP. There is also the point that it may not be the same IP being used the next time WP is accessed from that device. And if cookies are used to track the device, some devices are in schools or libraries, and not only is there no guarantee that the same person will be accessing WP two times in a row on the same device, it is very likely it will not be the same person, while other users may use more than one device to access WP. All of that will give inconsistent results on when a user see an add. Donald Albury 17:55, 2 December 2024 (UTC)
When is the start of this delay determined? The shared IP addresses in question from the original post aren't never-been-used-before IP addresses. I think for most readers this would just amount to shifting the start date of the campaign. Also note this won't solve either the caching problem or the problem of storing info for every IP address. isaacl (talk) 18:27, 2 December 2024 (UTC)
I have little technical expertise, but my thoughts are as follows:
User visits wp, a cookie is set.
Period of time, let's say a day or two elapses, and next time user visits wp, banner is displayed. No IPs needed. JayCubby 02:03, 3 December 2024 (UTC)
That defeats the purpose of the banner (which I appreciate suits those who don't want to see it, but doesn't help the campaign). isaacl (talk) 05:33, 3 December 2024 (UTC)
I don't see much value in displaying the banner on the first time, I myself wouldn't appreciate being begged for donations the first time I visited a webpage (or cleared my cookies). I don't know who is likely to donate, maybe it's effective to ask the first time a reader visits. JayCubby 16:09, 3 December 2024 (UTC)
I imagine the vast majority of readers is repeat traffic, in any case. isaacl (talk) 17:40, 3 December 2024 (UTC)
I have to wonder what the point of showing the banner to (likely) new visitors is. Maybe someone at the Foundation can chine in on their strategy. JayCubby 18:49, 3 December 2024 (UTC)
I think Wikipedia has far more repeat visitors than first-time visitors. These days, the only criterion I can think of to try to identify a group of likely first-time visitors is by age, and there isn't a good way to filter for that. isaacl (talk) 18:11, 4 December 2024 (UTC)
It's likely that there are some, but relatively few, first-time visitors. https://stats.wikimedia.org/ says the English Wikipedia sees about a billion unique devices each month, and there are only 1.5 billion English speakers in the world.
There will always be some first-time visitors. If we assume, to a first approximation, that everyone in the US will eventually visit the English Wikipedia, then purely due to about 10,000 babies being born each day, we'll (when they've gotten a little older) see about 10,000 first-time users each day. We get about 100,000,000 page views a day from the US, so 10,000 of those will be the first visit to the site.
That's roughly a one-in-10,000 risk that the person seeing the banner today is at the site for the first time. The other 9,999 times, it's at least the person's second page on the site. Since banners aren't run all the time, then it's only about 10% of each year's true first-timers who see any fundraising banners.
In other works, >99.9% of the time, this isn't something we should be worrying about. WhatamIdoing (talk) 00:20, 5 December 2024 (UTC)
Are cookies not now being created with the partial rollout of masked temporary accounts? CMD (talk) 04:31, 2 December 2024 (UTC)
Cookies are currently used, regardless of the IP masking initiative. The original post proposed hiding the donation banner based on IP address, instead of using client cookies. isaacl (talk) 04:44, 2 December 2024 (UTC)

Hi everyone,

Thank you very much for engaging with the fundraiser and I wanted to briefly take the opportunity to reply to you here.

I will take the IP address idea to the fundraising team to think about for future fundraisers, though I am not sure this is possible.

A note on our testing. From July to November, we test different versions of banners to small numbers of readers on English Wikipedia before the primary campaign begins, which this year runs from Dec 2nd. We need to test our technical infrastructure, trial new payment methods, and new features (such as annual recurring). We also conduct a/b testing on the content in order to determine on aggregate what readers are responsive to. This allows us to run overall less fundraising messages on the site while ensuring that we raise the money needed to support the movement.

Ethical considerations around A/B testing are valid and actively discussed across industries. At the Foundation, we approach testing with care. While we analyze donation rates to understand effectiveness, we also rely on ethical guardrails, informed by reader research and discussions with volunteers about the content, for example on our community collaboration page. The banners are displayed publicly without collecting personal data or requiring user participation, and no identifying information is gathered. Additionally, our privacy policy states that some browser information is collected automatically to improve the experience of the website.

Testing in fundraising and marketing is a standard practice but we agree it must be conducted thoughtfully to avoid harm. We appreciate your feedback, as it helps us continue to refine our approach.

I hope this addresses your concerns. Sheetal Puri (WMF) (talk) 21:38, 2 December 2024 (UTC)