Wikimedia: wprov | Hacker News Reader

Wikimedia: wprov

wikitech.wikimedia.org

71 pointstosh34 comments

I noticed this being appended to Wikipedia URLs when I share from the Wikipedia mobile app. Apparently it cannot be disabled, which I find very distasteful.

Yep it's annoying. Always have to go through the clipboard so that I can remove the parameter manually.

Just wondering why you’re removing it?

I often remove such appendages because I prefer clean, compact, semantic URLs that narrowly reflect my specific intent when sharing an article: "this article is good".

I don't want any other website/software agendas overlaid into my communication in ways that (1) leak info about what tools I'm using, even seemingly-innocuous info; (2) dilute any salient aspects of the URL by layering on excess (or even worse opaque) tracking info.

I've deleted that damn "?s=21" from the end of Twitter URLs hundreds of times.

Makes URLs longer and harder to read at a glance, and the reader cannot easily judge how much info they are leaking when they click it.

I always attempt to remove cruft form links before sharing/publishing/including. Cruft is in bad taste, and sharing it is bad manners. This assertion may well reflect my bad taste, and even bad manners toward people analyzing the provenance of clicks!

Maybe https://docs.clearurls.xyz/ (extension for Firefox and Chromium-based browsers) will strip wprov from URLs at some point https://gitlab.com/KevinRoebert/ClearUrls/-/issues/920

Another thing that ought to be built in for browsers to live up to being agents for users.

Short, clean URLs. I hate all those fbclid&utmz&affiliateid&xyz parameters with a vengeance, often they end up longer than the actually meaningful part of the URL.

While techniques like these are undoubtedly handy for analytics, advertising, affiliates, etc. in my opinion they go against the actual purpose of a URL. Removing such suffixes makes my URLs cleaner and defeats the purpose of the tracking mechanism, a practice which hopefully some day will be so commonplace that this kind of tracking is abandoned. (And there are tools that automate it, so it could some day become a tracking blackhole the way adblock is, if such a feature is ever integrated into a popular piece of software.)

I think it's very rude to share any URLs with tracking parameters, so I do that for all links. Added bonus is that it makes the links much more readable like the other commenter said.

Some of the most annoying ones are Amazon links. The product base URL is actually very compact, they just add an amazing list of tracking params that is pretty unwieldy to truncate on mobile.

For an encyclopedia that holds itself to high standards of integrity and information sourcing, disguising outbound-to-inbound tracking parameters by camouflaging them as "provenance" is pretty on the nose.

It’s hardly disguising. No one likes verbose URL parameters.

Having a reliable way to know where readers come from (in aggregate) is low-stakes stuff.

> is pretty on the nose.

why ?

To be fair, Wikimedia is the for profit arm of Wikipedia. Wikipedia itself is still pretty good in terms of user friendliness.

This is incorrect

This is completely backwards. Wikipedia is the name of a project. That project is operated by the Wikimedia Foundation, which is a nonprofit.

Wikimedia is also used as an umbrella term for everyone involved (e.g. wiktionary, wikidata, etc, some in person orgs, cross wiki management stuff, etc)

Without commenting on any other part, I don't get what you're saying about provenance. They're just using a word correctly.

I don't see any camouflage. The meaning of word is clear to me. Wiki is not in business of advertising or personal data trading, it's volunteer run project you know, so I don't see a problem with analytics there.

Some open source product owners have understood the value of not tracking their users. Others, it seems, still think it's a benign thing to do.

To be more constructive: if you need data, ask your users to enable telemetry (or in this case, provenance parameters). Some will understand what you need, and you'll get a usable sample from those users.

That’s the definition of biased sampling.

Only if you have reason to suspect that those who opt-in are a skewed sample with respect to the things you want to measure.

The problem with this is that nobody will change the defaults, so that approach is utterly useless if you want actionable information.

People label a number of different things, some bad and some good, under the general umbrella of "tracking". I'm not sure that name conveys enough nuance.

Wikipedia wants to evaluate how effective a particular feature is and so they are slightly modifying a URL to do this. It's not tracking you, your family or your dog. It's basically appending "&source=ios".

I'd think that is perfectly fine to any reasonable person.

Maybe I don't want to share a URL like that and tell everyone what OS and app I'm using to browse. I get that a lot of people (yourself included) don't care about that, and I think it is fair to ask what the big deal is, but that's just how I feel. I can't back that up with data or worries about a threat (though I imagine some people might be able to)... I just don't like it, even if it's one of the most benign forms of tracking.

> I'd think that is perfectly fine to any reasonable person.

I reject the assertion that I'm not a reasonable person. Please avoid arguments of this form, as it essentially says "if you don't agree with me, you're an idiot/wrong/defective", which is not productive way to argue your point. (I know I'm guilty of doing this on occasion as well, but I'm trying to get myself to stop.)

This shit is added without even a thought of what the hell you are trying to evaluate. Insecure middle management types that want to have this noise added should be required to write up a study proposal and register a hypothesis.

Instead its everywhere and for what? What is the actionable information?

> This shit is added without even a thought of what the hell you are trying to evaluate. Instead its everywhere and for what? What is the actionable information?

They are trying to evaluate the effectiveness of the social sharing features in their various projects. There are any number of reasons to do this - maybe nobody uses these features and they can remove them, or maybe "New Readers video in Hindi distribution on Facebook" has an insane click through rate that might have an impact on the allocation of resources in the future/feed into other products being developed.

You know, stuff you need to do when you run one of the worlds largest websites.

It's also quite clearly spelled out in the linked page, the mailing list thread[1] and the implementation.

If you can think of a better way to get this kind of data that fits more comfortably within your strong personal feelings about the contents of URLs then I'm sure the "insecure middle management type" that proposed it[2] would love to hear your counter-proposal.

1. https://lists.wikimedia.org/pipermail/analytics/2015-Februar...

2. https://meta.wikimedia.org/wiki/User:ABaso_(WMF)

> The problem with this is that nobody will change the defaults, so that approach is utterly useless if you want actionable information.

This is not just about defaults. When sharing from the Wikipedia Android app, I cannot use the settings to disable this tracking tag from being appended. I have to delete it by hand every time.

Changes like this are normally discussed and voted on by the community, right? Could someone link to the discussion where this decision was made, so we can better understand it?

It was discussed on a mailing list at some point.

https://lists.wikimedia.org/pipermail/analytics/2015-March/0...

> I'd really rather this be either something that's totally not understandable by the user (e.g. ?saf=1), or something that is clearly understandable (e.g. ?appshareafact=1).

I guess they decided to go with "totally not understandable."

Also - is this mailing list something that anyone can participate in, or only wikimedia office staff?

There's a form for subscribing to the mailing list here: https://lists.wikimedia.org/mailman/listinfo/analytics

Context? Explain?

It looks like it serves the same purpose as UTM codes

Should it then be added to block lists in various tracker-removal browser addons?

Yes

Reply on news.ycombinator.com