Re: Wikipedia at 25: A Wake-Up Call (essay)

Christophe, I think you are really hitting the nail on the head with your
principles.

And regarding the things to start, they are certainly thought-provoking and
worth discussing. However, there is one of our central values I want us to
keep held front of mind in this moment, and that is to focus on open source
and not fall for the lure of the proprietary just because it is AI. And
here I would like us to follow the principles of the Digital Public Goods
Alliance (who made us so proud when they awarded Wikipedia and Wikidata
with their certification of being Digital Public Goods) and go even further
than the definition from the Open Source Initiative definition for open
source AI. Their extension means that beyond the free license on the model
and the code, also the dataset used for training should be freely
licensed.[1]

This would not only be the ethically right thing to do, it would also
ensure we aren't dependent on Big Tech when doing our adaptation to the new
landscape.

[1] https://new.digitalpublicgoods.net/blog/ai-systems-as-dpgs

Jan Ainali


Den fre 1 maj 2026 kl 00:10 skrev Christophe Henner via Wikimedia-l <
[email protected]>:

> If I may, I think we need to be much more aggressive in how we address the
> situation. And I'm not saying we don't need interactive graphs, we should
> have had them years ago, but we need to go further than that.
>
> The way we grew is by having one product for each "vertical"
> (encyclopedia, pictures, data, thesaurus/dictionary, quotes, books,
> lessons). And that product is at the same time the content production
> platform for editors and the distribution system.
>
> I believe our mission is twofold: curating/maintaining content AND making
> it available. For years, one product was good enough-ish to do both, but
> not any longer.
>
> Distribution is being eaten by AI summaries, featured snippets, and social
> platforms. The production side works but is far from being optimal and we
> always need to be mindful of both.
>
> But, if you start considering it first as a production platform, it
> unlocks many things and one that becomes magic, the authorization to break
> it. The ability to lower our expectations on "perfection" (and tech teams
> are mostly delivering quite stable code for such a platform). But that
> comes at a cost, speed. It also comes at the cost of piling "problems"
> like... mobile editing.
>
> I now believe that pretending one product can keep doing both is how we
> sleepwalk into irrelevance.
>
> Anyway, I, and many others!, have some very long-term, very theoretical
> ideas on the evolution of distribution, but change needs to happen fast and
> we do have available quick wins. Not saying those are perfect, but it's to
> share what it could look like.
>
> *Three principles:*
>
>    1. *Human running the loop is our number one value.* Wikimedia
>    projects are great, but we have also successfully created something bigger:
>    communities that across topics, languages, and geographies share a vision
>    on what quality is. Even with the differences across projects, this is our
>    biggest achievement. No AI system can replicate community-validated
>    knowledge with editorial accountability. That's our moat. Not the content
>    itself, but the human process behind it.
>    2. *We are techy nerds and we need to act like it again.* We leveraged
>    that identity first with wikis, then with Wikidata, then with ORES (we
>    started working on machine learning in 2015 if I remember correctly!). We
>    need to stop protecting ourselves from technology and make it work for us.
>    Invest in innovation truly. Not as gadget but as our edge... like we did in
>    2001.
>    3. *Invest in ourselves.* I know this is "controversial" but if we
>    want to be relevant in the future, we need to stop chasing our own tails
>    and frowning every time we talk about investment. As a movement, we have
>    resources available or obtainable. We need to get back into a shape where
>    we trust ourselves to invest in ourselves. This is easier said than done,
>    and a lot of things need to happen. But they are needed.
>
> *Three things that can be started tomorrow:*
>
>    1. *Decide that our projects are production platforms.* The core of
>    our investment and focus should be to make them as efficient and useful to
>    all communities as possible. Distribution is a separate problem that needs
>    separate solutions.
>    2. *Put Wikidata and Commons (and Wikifunctions) front and center as
>    core resources.* I know not all projects are enthusiastic about them
>    but we do need to work that out because they are each in their own way the
>    best way to mutualize resources and work across projects. This needs to be
>    a clear goal and to find out how we make it happen properly together for
>    each projects.
>    3. *Unleash capacities.* Two clear areas of investment:
>       - *The production platform.* I know we've been focused on using
>       MediaWiki everywhere. But maybe (or maybe not I'm not the expert) it is
>       time to consider that it may not be the case anymore and build capacity to
>       develop and maintain other product lines. And yes it will mean investment
>       and let go, but sometimes, and, I may be wrong, I feel like we're dragging
>       Mediawiki a bit at the sweat of some insanely dedicated engineers (perhaps
>       Mediawiki is the perfect software but I doubt it).
>       - *AI for editors.* We need to embrace AI, and I know this is a
>       complex decision. AI-assisted translation across linguistic versions,
>       AI-powered content gap detection, AI drafts for stub articles reviewed by
>       humans, AI assistants to help newcomers to navigate rules, CI/CD for our
>       content to make content reviewing able to absorb high volume of AI slopes
>       without exhausting users.. The boundaries are to be defined, but we can't
>       define them without the capacity to experiment. The technical teams and
>       community have built a lot of very cool tools and infrastructure but we
>       need to double down. Make it friendlier, easier, more well known, AND with
>       real AI capabilities made available. Without that, we won't be able to
>       experiment and finding out how to use AI in ways that work for us will take
>       ages.
>
> Those would be a start. With potential quick impact and results. If
> anything, the sheer clarity of the road ahead may be enough to move the
> needle and open up possibilities for the bigger (more exciting?) challenges
> ahead on the future of open knowledge.
>
> But this stub of a plan starts with a clear, strong decision, and the
> first one is to say we consider our projects to be production platforms.
> That needs to be made by the Board of Trustees.
>
> Perhaps it's not the best path. That is fine, but we need a plan and a
> path.
>
> Any plan with any remote chance to keep Wikimedia relevant starts with
> clear, strong, and bold decisions by the Board of Trustees. And the "new"
> Annual Plan is a continuation of the past. It's not opinionated. It's not
> laying a clear path, a clear direction. It feels like a non-decision
> relative to a -20% traffic drop YoY. It reads slow and safe. But no
> decision is also a decision. The worst one, but still a decision.
>
> The best time to move was four years ago. The second best time is right
> now.
> --
> Christophe
>
>
> On Thu, 30 Apr 2026 at 15:41, James Heilman via Wikimedia-l <
> [email protected]> wrote:
>
>> One way we can make Wikipedia better is by adding interactive graphs. We
>> have built software to allow about 2,000 of these on all sorts of topics to
>> be integrated into Wikipedia. A bunch can be seen on Commons:
>>
>>
>> https://commons.wikimedia.org/wiki/Commons:List_of_interactive_data_graphics
>>
>> Today we have rolled out code to add initial machine translation support
>> to the SVG translation tool. For example:
>>
>>
>> https://svgtranslate.toolforge.org/File:Per_capita_energy_use,_World,_1965.svg
>>
>> By translating one SVG an interactive graph can be fully translated into
>> any language. The source of these graphs, Our World in Data, is only in
>> English. We are also developing tools to make keeping these graphs updated
>> much simpler.
>>
>> Many folks within our movement are innovating :-) Best
>> James
>>
>>
>> On Thu, Apr 30, 2026 at 11:43 AM Michael Snow via Wikimedia-l <
>> [email protected]> wrote:
>>
>>> On 4/29/2026 10:07 PM, Luis Villa via Wikimedia-l wrote:
>>>
>>> I would go a much different direction. After the 1st quarter, *our
>>> strategic process has to start by asking the hardest possible question:
>>> what if "reading an encyclopedia" is mostly over, like reading a print
>>> newspaper?* In other words, what if our long-term on-wiki readership
>>> graph looks like this one?
>>> https://www.pewresearch.org/journalism/fact-sheet/newspapers/
>>>
>>> A "print newspaper" is not a valid apples-to-apples comparison, because
>>> that phrase combines both the medium and the message and means the data can
>>> be dragged down by the limitations of either, or both. The graphs after the
>>> first, which leave some hope for the industry in digital subscriptions
>>> (hence the proliferation of paywalls in recent years), suggest that the
>>> medium is the greater factor in that instance. Having myself been part of a
>>> "newspaper" that has never seen the printed page, I would agree. However,
>>> as a free knowledge project there are also limits to what we can learn from
>>> commercial media, and I'm not suggesting subscriptions are the answer.
>>>
>>> We seem to have two leading theories to explain our trend. One is more
>>> about the medium, that there is a transition from text to audiovisuals. The
>>> other is more about the message, that information backed by human
>>> authorship is losing ground to synthetically generated equivalents. Either
>>> way, the premise is that we have fallen behind somehow. Both may even have
>>> a certain truth to them, but if we ask thoughtful questions and have the
>>> appropriate data, it should be possible to tease out how much explanatory
>>> power they really have.
>>>
>>> Ultimately, focusing on traffic means that the underlying issue is
>>> information discovery online. If you analyze in the direction of the
>>> medium, the challenge is the rise of social media as represented by TikTok
>>> or Instagram. If you analyze in the direction of the message, the challenge
>>> is the rise of artificial intelligence as represented by ChatGPT or Claude.
>>> What they share in common, with particular relevance to us, is a bias away
>>> from links - they rarely provide any but also rely on them less for their
>>> own discovery, utilizing apps/embedding/autoplay instead. The wiki
>>> meanwhile has long relied on the link above all else, both as the core of
>>> its internal structure and the entry point for its audience. Given that, I
>>> think there are difficult questions we need to ask. Like, what does a path
>>> to a different core technology look like for us? And, should we get there,
>>> what would we still retain of what we started with?
>>>
>>> The internet is at least middle-aged by now, so it has a decently long
>>> history of disintermediation. We are clearly in a significant period of it
>>> at present. I believe for a long time, our leading source of discovery has
>>> been referrals from search, fittingly enough for link technology. If search
>>> is being disintermediated, we should think about what other sources of
>>> discovery for the wiki that we can create or promote. But I do wonder,
>>> supposing we need some response to the impact of generative AI for us in
>>> discovery - how would using generative AI on the content creation side
>>> change the discovery equation for us at all?
>>>
>>> (The medium and the message can also cross-pollinate, but I don't
>>> believe that is a particularly critical factor here. If anyone disagrees on
>>> that point, they're welcome to demonstrate their case, but I suspect it
>>> would simply illustrate that there exists an opposing trend to "print
>>> newspaper".)
>>>
>>> --Michael Snow
>>> _______________________________________________
>>> Wikimedia-l mailing list -- [email protected], guidelines
>>> at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and
>>> https://meta.wikimedia.org/wiki/Wikimedia-l
>>> Public archives at
>>> https://lists.wikimedia.org/hyperkitty/list/[email protected]/message/BQOYBFCEZ4HLOOT3MHHFDDJZFHQDRYOB/
>>> To unsubscribe send an email to wikimedia-l-leave-RusutVdil2icGmH+5r0DM0B+6BGkLq7r@public.gmane.org
>>
>>
>>
>> --
>> James Heilman
>> MD, CCFP-EM, Wikipedian
>> _______________________________________________
>> Wikimedia-l mailing list -- [email protected], guidelines
>> at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and
>> https://meta.wikimedia.org/wiki/Wikimedia-l
>> Public archives at
>> https://lists.wikimedia.org/hyperkitty/list/[email protected]/message/7PGRAEJ7WJ2K5PLNNA3FIWFMFVAIIMUU/
>> To unsubscribe send an email to wikimedia-l-leave-RusutVdil2icGmH+5r0DM0B+6BGkLq7r@public.gmane.org
>
> _______________________________________________
> Wikimedia-l mailing list -- [email protected], guidelines
> at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and
> https://meta.wikimedia.org/wiki/Wikimedia-l
> Public archives at
> https://lists.wikimedia.org/hyperkitty/list/[email protected]/message/LBS54CSCHXR3GJQFX2WRATWWJLJ3EYOY/
> To unsubscribe send an email to wikimedia-l-leave-RusutVdil2icGmH+5r0DM0B+6BGkLq7r@public.gmane.org

_______________________________________________
Wikimedia-l mailing list -- [email protected], guidelines at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and https://meta.wikimedia.org/wiki/Wikimedia-l
Public archives at https://lists.wikimedia.org/hyperkitty/list/[email protected]/message/V5GZVBTBF4TMOHA46MEBPJC2UEOQWDLL/
To unsubscribe send an email to wikimedia-l-leave-RusutVdil2icGmH+5r0DM0B+6BGkLq7r@public.gmane.org