https://meta.wikimedia.org/wiki/Rebuilt_Wikimedia
Posted by: User:Grandmaster Huon
After years of contributing across Wikimedia platforms, I have come to a hard truth: Wikipedia and Commons are no longer modern, inclusive, or balanced. What began as a groundbreaking knowledge-sharing movement has calcified into something resembling a fandom wiki held together by aging software, outdated priorities, and fragmented editorial cultures.
Key Problems and Proposals
[edit]- 1. Wikimedia is stagnating.
The platform hasn’t aged well. Interfaces are dated, policies are stuck in the past, and Wikimedia's governance remains insular. There has been little meaningful structural reform since the late 2000s.
- 2. Wikipedia now resembles a fandom site disguised as a mainstream one.
Over-detailed anime episode guides, full-length articles on Pokémon species, minor characters, fan mods, and obscure console variants—this is not what the world's encyclopedia should prioritize. The result is a bloated, uneven content landscape dominated by niche interest groups.
- 3. AI moderation is urgently needed.
Commons needs AI tools that pre-screen uploads for copyright violations and quality *before* they go live. The current human-patrolling system is inefficient and demoralizing. We also need AI-assisted edit filtering on Wikipedia to reduce vandalism before it appears.
- 4. The article count must be pared down.
Wikipedia is flooded with stubs, trivia dumps, and fan-written cruft. Every article should meet a baseline of general interest, global relevance, and quality. Inclusion standards must be dramatically raised to prevent content bloat.
- 5. Vital article lists are miscalibrated.
For example, Final Fantasy is rated as Level 4 Vital, while Super Mario—a global icon—is only Level 5. That’s backward. Pokémon and Mario are cultural giants. Final Fantasy, though notable, is genre-specific and should not be elevated above more universally known franchises.
- 6. Niche topics dominate due to editor bias.
Touhou Project is rated "mid-importance" on WP:VG despite being niche even among gamers. Retro gaming and animanga fandoms exert outsized influence simply because they are persistent and loud—not because their topics have mainstream weight.
- 7. Wikimedia needs global unification.
Separate Wikipedias in every language result in massive duplication and inconsistency. We need a single multilingual Wikipedia and Commons with shared templates, unified article structures, and cross-language AI-assisted translation.
- 8. Governance structures discourage reform.
ArbCom, Ombudsmen, and entrenched admin groups often resist change and enforce legacy policy for its own sake. Meanwhile, many blocked or disillusioned editors walk away—or worse, turn into long-term abusers. There’s no effective channel for internal accountability when you're shut out of the system.
- 9. Mascots like Wikipe-tan no longer represent Wikimedia.
Wikipe-tan, a maid-themed loli anime character, belongs to a different era and subculture. She should be retired. Wikimedia must present a mature, inclusive image if it wants credibility with educators, journalists, and institutions.
- 10. Donations should be anonymized.
To prevent bribery or editorial bias disguised as financial support, donations must be anonymized through a secure third-party processor. Public trust requires financial neutrality.
- A rebuilt MediaWiki software stack, modernized for multilingual modular content and AI-first moderation.
- A unified Wikipedia and Commons with one backend, one content tree, and modern editorial tooling.
- A leaner article base—focused on topics of lasting, global significance.
- A new governance structure that empowers contributors, not just defends policy.
- A cultural shift from fandom enthusiasm to professional knowledge stewardship.
I say all this because I still believe in the Wikimedia vision. But to keep it alive, we must be willing to reboot what no longer works.
The proposed Wikimedia overhaul requires significant enhancements to the existing MediaWiki software, as current capabilities are insufficient to support the envisioned modernization. Key new features and extensions needed include:
- AI-Driven Pre-Publication Moderation:
A robust AI-powered system integrated into the editing and upload workflows to automatically screen new edits and media uploads for copyright violations, vandalism, spam, and policy compliance before content goes live. This would involve real-time content analysis with machine learning models specialized in text, images, and metadata.
- Unified Multilingual Backend:
Development of a centralized multilingual content management system that allows articles to exist as a single entity with multiple language versions linked and synchronized. This requires new architecture supporting cross-language templates, consistent data structures, and seamless AI-assisted translation workflows.
- Advanced Identity Verification Integration:
An extension for secure third-party identity verification to tie user accounts to verified real-world identities (while preserving privacy), thereby reducing sockpuppetry and malicious multiple accounts. This system must be modular, respect user privacy, and integrate with Wikimedia’s existing user management.
- Content Quality Enforcement Tools:
New extensions to enforce article quality thresholds automatically—flagging or restricting publication of articles that do not meet minimum standards (e.g., notability, sourcing, length). This may include AI-assisted article scoring and automated suggestions for improvement.
- Streamlined Editorial Interface:
Improvements and simplifications to the editing interface to reduce complexity and lower barriers for new and existing contributors, including context-sensitive help, guided editing workflows, and smarter template management.
- Donation Anonymization Module:
Integration with third-party financial anonymization services that process donations securely and anonymize donor identities before funds reach Wikimedia, ensuring transparency of funds without compromising contributor anonymity.
Implementing these features will likely require developing new MediaWiki extensions, significant backend architectural updates, and close collaboration with AI researchers and privacy/security experts. These enhancements are critical to transforming Wikimedia into a modern, inclusive, and sustainable platform.
Comments welcome.
Grandmaster Huon
Retains same domains as current projects.
In the interest of focusing our resources and attention on the platforms with the greatest global impact and sustainability, I propose that all Wikimedia projects other than Wikipedia and Commons be retired or consolidated.
Projects such as Wikibooks, Wiktionary, Wikiquote, Wikisource, Wikiversity, Wikinews, and others have struggled with fragmentation, inconsistent quality, low active participation, and limited visibility outside niche groups. Maintaining these multiple parallel projects:
Divides community efforts and technical resources.
Creates confusion and duplication across related content.
Hinders unified growth and modernization.
By concentrating on Wikipedia and Commons, we can:
Invest more effectively in core infrastructure and innovation.
Build a more coherent, accessible knowledge ecosystem.
Improve editorial quality and community engagement.
I invite discussion on this proposal with an open mind towards what best serves Wikimedia’s mission and long-term viability.
-Grandmaster Huon (talk) 17:14, 2 July 2025 (UTC)[reply]
- You probably forgot Wikidata – how do you envision existence of this project? How would your proposal handle content of (then) deleted other sister projects? I will follow with some critics of the points you have listed above.
- If policies were struck in the past, why then very actual policies, like enwiki WP:AI exist at all? Not only policies aren't meant to be written in stone, they are meant to be shaped by local communities and it leaves me distasteful to have attempts of rewriting local policies here on Meta with global input – that actually is an attempt to threaten existential right of every project to be independent.
- I agree that Wikipedia is not a fancruft, it is still far from being a Fandom wiki. All these topics you have listed have received coverage in notable gaming media so I don't see why they shouldn't be included in an encyclopaedia that claims it is a sum of all encyclopaedic knowledge.
- How will you achieve AI-driven patrolling systems? They require immense systemic changes and given MediaWiki is mostly volunteer-driven ecosystem we cannot undertake such a gigantic task on ourselves. Especially as such idea requires strict implementations on how models will recognise vandalism and be unbiased.
- This point gets my strongest oppose. I feel like this is a quiet way of trying to make policies of English Wikipedia, which probably has one of the best notability policies out there, of global scope. There is no need to play cards of global importance – afterall, how would you define that? That would only deepen the bias Western (Anglophone) sphere already has on wikiprojects and is detrimental to WMF mission of spreading and providing open access to knowledge.
- Most wikis don't have such hierarchy as the English Wikipedia has. Irrelevant to the global scale, and these grades are only approximations, not an objective truth representation. This is something that can be changed locally and cannot be directed through a RfC like this.
- I think you're unfairly accusing ArbCom and admins of misconduct – any specific examples you can provide? Burden of proof is usually on the accuser. And yes, policies exist for a reason. Wikipedia is not an anarchist server.
- Donations are anonymous and users (=content creators) are not paid so point nr. 10 is equally void.
- If for anything else, I'd like to hear your further answers to topics of AI neutrality and implementation. One last question, was this proposal written by a large linguistic model?
- Thanks, A09|(pogovor) 17:46, 2 July 2025 (UTC)[reply]
- Response from User:Grandmaster Huon
- Thank you for your thoughtful and detailed reply. I’ll address your questions and concerns point-by-point.
- 1. Wikidata
- You're absolutely right to bring this up. I should have clarified: Wikidata is not among the projects I propose retiring. On the contrary, I view it as a crucial infrastructural layer — not a front-facing public project like Wikipedia or Commons, but an essential multilingual data backbone that supports both. I envision it continuing and even expanding its role under any unified or modernized Wikimedia system.
- 2. What happens to sister project content?
- A structured, humane consolidation process would take place:
- Content with clear educational or cultural value could be migrated into Wikipedia (text) or Commons (media, static educational materials).
- Projects like Wikiquote and Wikibooks might have key materials merged as supplementary sections or references within broader Wikipedia topics.
- Other content that doesn’t meet notability, neutrality, or sourcing standards could be archived in a read-only state or deprecated.
- The point is not to erase contributions — it’s to reduce redundancy and improve editorial focus while preserving what has value.
- 3. On policies and local independence
- To be clear, I’m not calling for Meta to override local projects. I respect the autonomy of local communities to shape their own policies. My concern is with long-standing fragmentation and stagnation. I’m proposing that we start conversations about **alignment and modernization** — not imposition.
- There is a legitimate tension between global coherence and local independence. My aim is to explore how to balance that more effectively in 2025 and beyond.
- 4. On fandom content
- You’re right that many topics (e.g., *Touhou*, *Final Fantasy*) are covered in reliable sources and technically meet notability. My critique is not that they exist — it’s that they are disproportionately detailed and overrepresented relative to their global cultural footprint.
- Wikipedia should include all knowledge, yes — but it should still apply proportional editorial weight. When deep fandoms drive the shape of content, the result is not balance, but distortion.
- 5. AI-driven patrolling
- I fully agree: this would be a large undertaking. That’s why I view it as a **strategic direction**, not an immediate mandate.
- Implementation would involve:
- Partnerships with academic institutions or AI nonprofits.
- MediaWiki extensions designed for moderation assist, not replacement.
- Human-in-the-loop architecture and transparent training data.
- It’s ambitious, yes — but feasible if approached incrementally and with external collaboration. Commons and vandalism-heavy areas would benefit first.
- 6. “Global importance” and Western bias
- This is a critical point, and I appreciate your caution.
- To clarify: I do not want to enforce "Anglo-notability" as the global norm. I propose instead that we:
- Elevate content with broad human relevance, regardless of geography.
- Resist overgrowth of hyper-specific, fan-driven, or trivial articles.
- Use editorial discretion to highlight underrepresented regions and topics, not just what’s popular in the Anglosphere.
- This is about global balance — not imposing one region’s standard over others.
- 7. On article grades (e.g., Final Fantasy vs Super Mario)
- You're correct that importance grades are local, and their subjectivity is acknowledged. But they matter — both within the project and externally (e.g., in search results and perception). When something like *Super Mario* ranks below *Final Fantasy* in importance hierarchy, it reflects a misjudgment of cultural weight.
- That said, I agree this is something that must be addressed locally through discussion, not imposed globally.
- 8. On ArbCom/admin structures
- Let me be clear: I’m not accusing any specific ArbCom member or admin of misconduct. I’m critiquing a systemic dynamic — one where:
- Blocked users have no effective internal review path.
- Editorial governance often becomes circular or resistant to reform.
- Long-term contributors burn out while bad-faith actors adapt.
- The solution isn’t to tear it down, but to introduce more transparency, feedback loops, and rotation of responsibility.
- 9. On donation anonymity
- Most donations are already anonymous, yes. However, I propose that anonymity be enforced at the processor level — such that even the Foundation cannot access identity metadata. This further insulates editorial decisions from financial influence (real or perceived) and avoids any conflicts of interest.
- 10. Was this written by an LLM?
- Yes — this proposal was written with the aid of a large language model (LLM), but under extensive human input and supervision. The core ideas, direction, and priorities come from me, a human contributor and editor.
- The LLM was used for structure, clarity, and tone — like a skilled assistant. This mirrors my broader argument: Wikimedia should embrace AI tools when used responsibly and transparently, not fear them.
- ---
- Again, thank you for the critique. I welcome further discussion and refinement. We are all here, ultimately, because we want Wikimedia to succeed.
- — User:Grandmaster Huon Grandmaster Huon (talk) 18:25, 2 July 2025 (UTC)[reply]
- Thanks for clarifying the position of Wikidata. Though there are still baseless arguments in your response that a) are simply result of being mistaken or b) outside of scope of Meta as they should be dealt locally.
- Sister projects exist to a cause. They aren't just something that was made spontaneously, they have their own mission and their own set of rules. These projects are not compatible with Wikipedia, in fact so much that the only similar thing is having the same license and being an open-project. Merging all these projects into Wikipedia in an attempt to gain more popularity is misleading and against the most basic general rule of 5 pillars (WP:5P), which states It is not a dictionary, a newspaper, an instruction manual, nor a collection of source documents or media files, although some of its fellow Wikimedia projects are. A merge wouldn't be just detrimental, it would present a dangerous threat to what Wikipedia was originally designed for. Furthermore, how would you achieve community consensus of all wikis to have them joined into Wikipedia? Obtaining consensus would be near impossible, even if your proposal was ideal.
- You're contradicting yourself. One cannot propose all existing projects merge and then claim striving for project independence. What if some of the projects don't agree? You'd get a mess of some projects being merged and others staying independent – making crosswiki patrolling a true pain in the ass.
- Argument on including all knowledge and giving editorial weight is contradicting as well – also, your proposed model would only worsen the unconscious editorial (user) bias. One couldn't care less for Mario Cart but has a deep love for Mack trucks, while another one has interests the other way around: how would these two come up to an agreement? Would not. Policies exist for reasons, and this is one of them. What the term "trivial" constitutes differs between users. Furthermore, how will you define what "hyper-specific" is and how is this not in conflict with the goal of gathering all knowledge?
- You did not propose any concrete plan about AI implementations apart from very general statements. I'd like to see much more details.
- Statement of having no internal review path is simply false. Since every bit of your proposal is oriented based on the English Wikipedia (and disregards other 300+ Wikipedia projects) I'd just like to point out that even unblocks are dealt with transparently. Please check before you publish such baseless arguments, and I will point you to editorial mass. Every single second there are tens to hundreds of editors watching changes so any argument of having unadapted and unfair ecosystem is, personally, null to me.
- Your proposal about having anonymous donations at the processor level is illegal actually. Nonprofits must track where payments were made from, however they also should not publish such data. Again, I will reinstate: fear is irrational as users are not paid, hence whole payments transparency proposal is equivocally void.
- Sorry to say, but this proposal is not well thought and it mixes local reform proposal with some very controversial global ones. A09|(pogovor) 09:17, 3 July 2025 (UTC)[reply]
- === Response from User:Grandmaster Huon ===
- Thank you, A09, for your detailed feedback. I value this exchange and will respond thoughtfully to each of your points.
- 1. On sister projects and incompatibility with Wikipedia
- You are correct that each sister project was founded for a specific purpose and that the five pillars of Wikipedia explicitly differentiate it from other forms like dictionaries, newspapers, and manuals. My proposal does not intend to violate these core principles. Rather, it raises the question of whether, after 20+ years, we are maintaining too many loosely maintained or underused projects that could benefit from consolidation or clearer boundaries.
- To clarify: I am not suggesting we turn Wikipedia into a dictionary or textbook. I am suggesting that:
- Some sister projects (e.g., Wikibooks, Wikiquote) could have their strongest content summarized or integrated into encyclopedic articles where appropriate.
- Some smaller or underused projects could be placed in long-term archival mode rather than closed abruptly.
- I fully accept that not every project can or should be merged, and I recognize that consensus would be difficult. The intent is to raise a conversation about strategic focus, not to bulldoze existing communities.
- 2. On the contradiction of local independence vs. merger
- Thank you for pointing this out. I see how my language came across as contradictory. To clarify:
- I support local independence where projects are active, healthy, and fulfilling a distinct mission.
- My concern lies with projects that are abandoned, minimally active, or redundant.
- I agree: a partial merge would create inconsistency and complexity. That’s why any such action would need to be accompanied by a comprehensive **global consultation and binding consensus**, not unilateral action. If consensus isn’t possible, the idea would be dropped.
- 3. On defining editorial weight and “triviality”
- This is a fair challenge. It is true that what one editor deems “trivial” another finds essential. I do not propose that we **ban** niche topics or impose personal preferences.
- What I suggest is that:
- Wikipedia (and Commons) should work to better enforce existing standards around **notability, reliable sourcing, and proportionate weight**, which are already embedded in policies like WP:DUE and WP:NOTE.
- AI and human editors could be used to **flag** (not delete) articles that lean too heavily into trivia or in-universe focus, leaving final judgment to humans.
- No system will ever be perfect. I propose that we openly recognize this imbalance and begin to address it through community discussion, not dictate outcomes.
- 4. On AI implementation specifics
- You are right that my original AI proposals were broad. To be more concrete:
- AI-assisted moderation could first target **Commons uploads**, specifically flagging:
- - Suspected copyright violations via reverse image matching.
- - Low-quality images (blurry, unsourced, misnamed).
- On Wikipedia, initial deployment could focus on:
- - **Vandalism detection** using language models trained on prior vandal patterns.
- - **Content flagging** for neutrality violations, policy breaches, or unverifiable sources.
- These systems would require:
- **Open-source AI models** with transparency about training data and bias mitigation.
- **Human override at every stage** — no AI decision would be final without review.
- This would be iterative, gradual, and entirely opt-in for projects.
- 5. On internal review and unblock fairness
- Thank you for this correction. I accept that unblock requests are available and transparent in most wikis. My statement was too broad. What I meant to highlight is that for **indefinitely blocked users**, particularly those caught in complex conduct disputes, there can be **very limited paths back**—often with inconsistent standards between projects.
- That said, I withdraw the claim that no review path exists. A better way to frame it is: we could explore **broader or more compassionate unblock pathways** for users who demonstrate genuine rehabilitation.
- 6. On donations and legal compliance
- I appreciate this legal clarification. If processor-level anonymity violates nonprofit law, then I retract that suggestion in its current form. The spirit of the point was to **increase perceived financial neutrality**, not to break transparency regulations. If current practice already achieves that, I accept the correction.
- 7. On global vs. local reform mixing
- I agree this proposal spans both local (English Wikipedia-specific) and global issues, and this may have muddied the waters. Going forward, I am open to:
- Splitting this into a **global movement-wide discussion** (AI, sister project strategy) vs.
- Separate **local proposals** for English Wikipedia or Commons-specific reforms.
- I appreciate your fairness in pointing this out.
- ---
- In closing, I fully acknowledge that some of my early framing was too blunt and some points require deeper development. My goal is not to dictate but to invite reform-minded discussion. I remain open to withdrawing or significantly revising this proposal based on community feedback.
- Thank you again for the respectful and rigorous critique.
- — User:Grandmaster Huon Grandmaster Huon (talk) 14:44, 3 July 2025 (UTC)[reply]
- You're still contradicting yourself (or well, the AI does) so I will not loose any time more on this IMHO pointless proposal. A09|(pogovor) 18:36, 4 July 2025 (UTC)[reply]
Oppose shocking: someone who was blocked on enwiki and frwiki goes to claim that Wikipedia isn't "no longer modern, inclusive, or balanced". Going on to then write comments about sister projects lacking when it's evident you don't know the scope of those projects is the cherry on top (everything else has already been mentioned in A09's comment). Also, really? //shb (t • c) 09:27, 3 July 2025 (UTC)[reply]- === Response from User:Grandmaster Huon ===
- Thank you for your input. I would like to address your points directly:
- 1. On my block history:
- I understand that my previous blocks on enwiki and frwiki may raise concerns for some. I want to be clear that:
- I am not hiding my history, nor do I believe that past sanctions invalidate constructive ideas for the future.
- Many longstanding contributors — including respected ones — have experienced conflict or blocks. This is part of any large volunteer project.
- My focus here is on the ideas presented, not personal histories. I hope others will judge the proposal based on its merit, not on past mistakes.
- 2. On the use of AI in drafting:
- Yes, I have been transparent that this proposal was written with extensive AI assistance but with full human direction, review, and authorship of its ideas. The AI was used to improve structure and tone, not to create the core content. I have never misrepresented this.
- The use of AI in drafting should not be disqualifying — in fact, the responsible use of such tools is central to the very reforms under discussion.
- 3. On understanding sister projects:
- I acknowledge that I may not have the same depth of experience in every sister project. That’s precisely why I brought this forward for discussion — not to dictate outcomes but to invite perspectives from those more deeply involved.
- If some of my characterizations of sister project scope were incomplete, I welcome correction. My aim is to raise concerns about long-term sustainability and impact, not to dismiss the value of existing communities.
- ---
- I recognize that some of my proposals are ambitious and controversial. I am open to being corrected, to revising ideas, or to withdrawing specific points if the consensus deems them unhelpful.
- What I am asking for is to keep the focus on the substance of the ideas rather than on personal attacks or past history.
- — User:Grandmaster Huon Grandmaster Huon (talk) 14:52, 3 July 2025 (UTC)[reply]
- To be fair, every WM project other than WP, Commons or Wikidata is a zombie project now, kill them off. Grandmaster Huon (talk) 18:10, 3 July 2025 (UTC)[reply]
- Even the WMF official website knows this and only talks about WP and Commons now. Grandmaster Huon (talk) 18:10, 3 July 2025 (UTC)[reply]
- Even ChatGPT knows. Grandmaster Huon (talk) 23:06, 3 July 2025 (UTC)[reply]
- I'm not wasting my time responding to AI slop, thank you very much. //shb (t • c) 07:41, 4 July 2025 (UTC)[reply]
- AI reflects complete internet consensus, not my opinion. Grandmaster Huon (talk) 13:56, 5 July 2025 (UTC)[reply]
- That is irrelevant. Wikimedia community should decide on these matters, not your preferred internet club/forum. A09|(pogovor) 09:02, 6 July 2025 (UTC)[reply]
- Such comment is simply a nullification and sterns hardly into WP:JUSTDONTLIKEIT. I have yet to see official WMF comment that these projects are dead or that they are planning to close them. And you know very well that gaslighting an AI model is not a valid argument either. A09|(pogovor) 18:34, 4 July 2025 (UTC)[reply]
- WMF has no official plans to close such projects, but they are de-facto dead in my view. Grandmaster Huon (talk) 21:12, 4 July 2025 (UTC)[reply]
- Then do not misrepresent what WMF had said. A09|(pogovor) 09:00, 6 July 2025 (UTC)[reply]
- There is a reason why the website isn't as popular as it used to be. Grandmaster Huon (talk) 21:31, 4 July 2025 (UTC)[reply]
- ChatGPT is very hard to gaslight now, ChatGPT's responses reflect a general consensus from the internet, given that it was trained from it, along with a multitude of other sources. Grandmaster Huon (talk) 22:06, 4 July 2025 (UTC)[reply]
- Your chatGPT thread says otherwise. Furthermore, Meta doesn't care what Twitter or Reddit say so importing external discussions is pointless and against established editing patterns.A09|(pogovor) 09:00, 6 July 2025 (UTC)
Discussion open until at least 14 July 2025 (UTC)
Baseless proposal, mostly written with the help of an LLM. Proposed by a user blocked on enwiki and frwiki for disruption. Not a single proposal suggested on this page aligns with what Wikimedia projects stand for, so it should be deleted to save editors' time. – DreamRimmer ■ 12:07, 6 July 2025 (UTC)[reply]
- Speedy delete A not useful proposal created by an user with a bad history. Per DR, it should be deleted to save editors' time. @SHB2000 and A09: Pinging users who left comments at that page. – Phương Linh (T · C · CA · L · B) 12:14, 6 July 2025 (UTC)[reply]
- Weak delete (and not speedy) – extremely unserious proposal largely AI-generated; user history extremely uncompelling as well. //shb (t • c) 12:36, 6 July 2025 (UTC)[reply]
- Weak oppose Deleting would IMHO just open a way to bad proposals and it'd be harder in the future for us to point what was already discussed and what arguments not to make. Do we have precedence in deleting such pages, is this even covered by any specific policy?--A09|(pogovor) 12:52, 6 July 2025 (UTC)[reply]
- Weak keep I don't want to throw the baby out with the bathwater. This page could have some good ideas for future reference. Grandmaster Huon (talk) 20:37, 6 July 2025 (UTC)[reply]
- Keep I don't see how this goes against the meta inclusion policy and deleting it seems like it could set a dangerous precedent as A09 was eluding to. If an editor is blocked on other projects, I don't think we can use that as a reason for treating them any differently here since this is a separate project. Ternera (talk) 23:02, 6 July 2025 (UTC)
