#137 - Larry Sanger - Will AI Replace Wikipedia?

Summary

Podcast Summary: The Peter McCormack Show

Episode #137 – Larry Sanger: Will AI Replace Wikipedia?
Date: December 29, 2025
Host: Peter McCormack
Guest: Larry Sanger, Wikipedia Co-founder

---

Overview

This episode features a deep-dive conversation between Peter McCormack and Larry Sanger—co-founder of Wikipedia—on the integrity, neutrality, and future of Wikipedia in the age of AI. The discussion critiques Wikipedia’s governance and neutrality, explores how power dynamics and anonymity shape its content, examines the threat and promises of AI, and considers Sanger’s “Nine Theses” to reform the platform. The episode is candid and provocative, with Sanger openly questioning whether Wikipedia is now doing more harm than good.

---

Key Discussion Points & Insights

The Founding and Evolution of Wikipedia

• Wikipedia’s Origin Story ([02:12])

  • Sanger recounts meeting Jimmy Wales in the 1990s and joining him to launch Nupedia—a failed expert-driven encyclopedia.
  • In January 2001, inspired by wiki technology, they pivoted to Wikipedia, allowing open collaboration and rapid article creation.
  • “Within a couple of months there were hundreds of articles, and by the end of the year there were about 20,000 articles.” — Larry Sanger ([05:29])

• Early Challenges ([07:27])

  • Early articles were mainly “stubs” (short entries), and it took effort to focus contributors on building an encyclopedia rather than an anything-goes wiki.
  • Vetting and approval were initially planned via Nupedia, but as Wikipedia grew, peer review concepts were quickly sidelined.

Governance & Power Structures

• Who Controls Wikipedia? ([13:43])

  • Wikipedia’s most powerful editors (“the 62 most powerful people”) are largely anonymous, making the platform’s governance opaque and vulnerable to hidden agendas.
  • “85% of them are anonymous. Like, we don’t know who they are. So all the more then we don’t know who they’re working for.” — Larry Sanger ([13:43])
  • The power to ban, restrict, and delegate further authority is wielded by a small group, with potential for arbitrary and biased enforcement.

• Potential for Abuse & Bias ([15:03], [16:27])

  • Editors can be banned for capricious reasons, including falling foul of the prevailing ideological leanings of those in power.
  • The administrative structure rewards those already aligned with the dominant culture, hindering true neutrality.

• Anonymous Editing—Pros & Cons ([17:46])

  • Sanger distinguishes between the right of rank-and-file editors to be anonymous and the need for transparency among those with administrative power.
  • Anonymity at the top invites infiltration by governments or other organized groups.

Notable Quote:

“Information warfare…one of the main battlegrounds, I suppose, would be something like Wikipedia.” — Larry Sanger ([19:11])

Content Manipulation & Platform Capture

• Ideological Capture Examples ([21:03], [23:47])

  • Sanger cites the “Gang of 40” allegedly coordinating Wikipedia edits for political ends, particularly in the Israel-Palestine conflict.
  • Other groups, such as activists and skeptics (e.g., Guerrilla Skeptics), systematically shape the portrayal of controversial topics.

• PR & Activism ([28:24])

  • There’s a thriving industry of PR firms writing or curating entries for clients.
  • Some activism, such as improving coverage of women in science, is welcomed, but coordinated bias undermines neutrality.

• “Capturable” Platform ([28:24], [29:33])

  • Wikipedia is vulnerable to being influenced, or “captured,” by well-organized ideological or commercial interests.

Notable Moment:

"Influenceable? Is that a word? Anyway, it can be influenced, yeah. And yeah, for sure." — Larry Sanger ([28:24])

The Possibility and Limits of Neutrality

• The Nature of Neutrality ([30:54], [34:00])
  • Sanger argues that neutrality is possible in written text—even if individuals aren’t perfectly objective—by carefully canvassing all major views and representing them accurately.
  • The goal is not to provide a single “correct” answer but to lay out arguments so readers can decide.

Notable Quote:

“The purpose of neutrality…is so that somebody can make up his own mind.” — Larry Sanger ([32:03])

• Peer Review as Safeguard ([35:48])
  • No human is perfectly neutral. Peer review is necessary for detecting and correcting unintended bias in text.

Wikipedia vs. AI – The Future of Knowledge

• Can AI Be More Neutral Than Humans? ([38:07])

  • Sanger is cautiously optimistic: AI can enforce certain neutrality protocols dispassionately, but is only as good as its training and the prompts it’s given.
  • “You actually have to be an expert on a topic to know that a presentation of the topic is neutral.” ([38:12])

• Live AI Experiment: ([39:23]–[52:03])

  • The hosts compare how ChatGPT, Grok, and Perplexity rate the neutrality of Wikipedia’s articles on controversial subjects (e.g., Donald Trump, Joe Biden, Keynesian economics, Marxism).
  • AI notices non-neutral phrasing, especially on contentious figures.
  • “Short answer: No, the introduction is not neutral…” — ChatGPT, as read by Conor ([47:12])

• AIs as Hybrid Editors ([81:54])

  • Sanger notes the emergence of “Grokipedia,” where AI mediates editorial disputes and inserts factual corrections proposed by users.

Notable Quote:

“I would bet on AI before I would bet on Wikipedia.” — Larry Sanger ([55:19])

The “Nine Theses” for Reform

• Motivation and Reception ([00:00], [58:48], [91:55])
  • Sanger’s set of nine proposed reforms aim to restore openness, transparency, neutrality, and accountability.
  • While some Wikipedia insiders express partial support, the lack of formal governance makes adoption unlikely.

Main Theses Highlighted:

1. End Decision-making by “Consensus” ([59:57])

   • Critiques the abuse of “consensus” as establishment dogma.
   • “The consensus view is not a particular view on a controversial issue. It is a view about how the controversy should be fairly represented.” — Larry Sanger ([61:20])

2. Enable Competing Articles ([63:12])

   • Sanger suggests allowing alternative articles representing different frameworks or worldviews.

3. Abolish Source Blacklists ([67:06])

   • Wikipedia’s blacklisting of sources like the New York Post is called out as ideological gatekeeping.
   • “I haven’t, at least not recently, caught Breitbart News in any major errors.…They bend over backwards to do old-fashioned hard reporting.” ([73:35])

4. Revive Original Neutrality Policy ([75:51])

   • Calls for renewed commitment to neutral representation of all major perspectives.

5. Legislative Process ([86:51])

   • Advocates for a democratic editorial assembly to legitimize major policy changes.
   • “Open Internet communities tend to be institutionally conservative, and Wikipedia is exactly that way.” ([87:24])

Wikipedia’s Reputational Trajectory and the AI Threat

• Declining Trust & Traffic ([54:37])

  • Wikipedia traffic is dropping as AI-powered search and summarization tools become more trusted for quick, digestible answers.

• Wikipedia’s Role in the Age of LLMs ([94:00])

  • Wikipedia’s value in summarizing information for training AIs may keep it relevant, but as a go-to source for readers, it risks becoming obsolete or an “establishment relic.”

Notable Quote:

“Wikipedia sucks the air out of a lot of other very valuable competing encyclopedia projects.” — Larry Sanger ([100:55])

The “Encyclosphere” Concept

• Decentralizing Knowledge ([100:55])
  • Sanger proposes an open, federated network (“encyclosphere”) where anyone can submit and aggregate encyclopedia articles across platforms.
  • He highlights his involvement in projects like EncycloReader and EncycloSearch, which aggregate from many smaller wikis and platforms.

---

Memorable Quotes

• “Of course governments are involved, of course high paid PR firms are involved. We know that for a fact. Right. And they will do their very best…to hide their tracks and make themselves look like harmless college students and whatnot.”
  — Larry Sanger ([19:11])

• “If it’s going to insist on allowing people to write articles from a gasp framework, it should allow other articles on the same topics from other frameworks.”
  — Larry Sanger ([66:58])

• “You should…just be bold and don’t worry too much about the rules. If the rules make you nervous, then just ignore them, go about your business.”
  — Larry Sanger ([82:29], on the “Ignore All Rules” policy)

• “I’m sorry. I’ve been atoning all my life, actually, since starting it.”
  — Larry Sanger ([90:32]), on the trajectory of Wikipedia.

• “If you could turn [Wikipedia] off, would you turn it off?” — Peter McCormack
  “Yeah, I probably would, to be honest.” — Larry Sanger ([100:48])

---

Key Timestamps

• [02:12] – Wikipedia’s Founding & Early Days
• [13:43] – Who Runs Wikipedia? The Power of Anonymity
• [21:03] – Platform Manipulation: Political & Ideological Capture
• [30:54] – Can Neutrality Ever Be Achieved?
• [38:07] – Can AI Be Neutral?
• [59:56] – Sanger’s Nine Theses: Reform Proposals
• [67:06] – The Issue of Source Blacklists
• [75:51] – Neutrality Policy: Then & Now
• [86:51] – Legislative Process & Governance
• [100:55] – The Encyclosphere as an Alternative
• [104:17] – Sanger’s Call to Action for Disenfranchised Contributors

---

Conclusion

This episode is an urgent, sometimes combative reflection on Wikipedia’s trajectory, its surrender to establishment perspectives, and its potential eclipse by AI-driven knowledge platforms. Sanger is passionate about returning to true neutrality and decentralizing encyclopedia creation, but issues a chilling warning: “If you could turn [Wikipedia] off, would you turn it off?...Yeah, I probably would.” ([100:48])

Listeners are left with the sense that the way society curates, shares, and trusts information is in flux—with AI both as threat and promise—and that the risks and remedies discussed here are critical for the future of public knowledge.

---

For More

• Larry Sanger’s blog and proposal: LarrySanger.org
• Sanger’s “Nine Theses” (in-depth): [Referenced in-show]
• EncycloReader and EncycloSearch: Alternative, federated encyclopedia aggregators

---

End of Summary

Transcript

[Larry Sanger]

Let's put it this way. I wrote the nine theses in good faith, right? I actually would like to see Wikipedia reformed. It's absolutely the case. The reason that I wrote the nine theses is that as a friend pointed out and I just couldn't let go of the thought, right, If I made a reform proposal for Wikipedia and they adopted even one of the major reforms that I proposed that would actually make the world a better place. And okay, if you actually think that Wikipedia has done damage as I do, then I think I owe it to the world to do what I can to to reform it.

[Peter McCormack]

Does it do more harm than good? Or is it mainly good with some harm?

[Larry Sanger]

I mean, if you could turn it.

[Peter McCormack]

Off, would you turn it off? This show is brought to you by my lead sponsor, Ayron. The AI Cloud for the Next Big Thing. Aaron builds and operates next generation data centers and delivers cutting edge GPU infrastructure with all powered by renewable energy. Now, if you need access to scalable GPU clusters or are simply curious about who is powering the future of AI, check out iren.com to learn more, which is Irene.com, right? Hi, Larry, nice to meet you.

[Larry Sanger]

It's good to meet you, Peter.

[Peter McCormack]

It's a bit of an honor really. Like Wikipedia is one of those founding kind of institutions of the Internet that.

[Larry Sanger]

I suppose it is.

[Peter McCormack]

We never know. I don't know when it started, but I know I pretty much have used it continually a little bit less these days. AI has taken over a lot. But I know there's a relationship between Wikipedia and AI. But I've got so many questions for you. But before we get into the more controversial bits, I'd really love to know the kind of founding story of how Wikipedia came to be.

[Larry Sanger]

Well, I'll take you through it briefly, I guess. So I met Jimmy Wales face to face a couple of times in the 90s. We knew each other through these philosophy mailing lists and I was soliciting feedback on a website idea that I had in early 2000 and you know, ran the idea by a number of friends and Jimmy Wales was one of them and he replied and saying, forget about that, come and work for me and you'll start a new encyclopedia. It's called Newpedia. And I thought what a great opportunity. It didn't pay very much, but it was so fascinating, this idea. And he would give me a bonus for finishing my PhD, which I did four months after starting work on that. And so moved down to San Diego, got to work organizing ultimately about 2,000 volunteers to work on Nupedia. And it was very old fashioned, I guess you could say. We actually sent mail across the country and collected a bunch of academics mostly, but just a lot of other kinds of people. So after a year we had not made very much progress and we had started looking around for new ways to get new people into the project. It was my job actually to organize this. I was, I was called the editor in chief, but I was also essentially the guy who was getting the thing started, Nupedia. So originally what became Wikipedia was going to be the Nupedia wiki. But the Nupedia editors, these academics, didn't want to have anything to do with a wiki. So we decided to launch it because it seemed like such a cool idea. Yeah, a friend of mine told me about wikis over Mexican dinner on the night of January 2, 2001. And a couple of weeks later we launched wikipedia.com it was going to be a for profit website at first, a couple of weeks later and you know, it was really exciting at first. I wrote a lot of the original policies. I again organized the people that those 2,000 people that I had collected in the first year. Some of them came and worked. So we actually hit the ground running. And that was one of the reasons why Wikipedia had such a great strong start. So within a couple of months there were hundreds of articles and by the end of the year there were about 20,000 articles.

[Peter McCormack]

What kind of articles do you begin with? Because one thing that's quite funny, Conor's my son, by the way, and I have to explain to him the world before the Internet sometimes, because Connor's only ever known an Internet. And we had a world where I had the Encyclopedia Britannica, where somebody came and sold us and we had 26 volumes. And when I did my homework, maybe I was doing homework on volcanoes and I would get the V volume and open it up and read about volcanoes. Wikipedia kind of ended that market, the need for it. It's not something we now have.

[Larry Sanger]

Yes.

[Peter McCormack]

So when you start Wikipedia, how do you decide what content you start with?

[Larry Sanger]

Not really. So you have to understand that all a wiki is a blank wiki. It's like a bulletin board that you can just keep adding cork boards to and they're all connected and anyone can write anything that they want. And so it can become any kind of website that you want. And in the early days there was no tight connection between wikis and encyclopedias. So I had to actually impress upon a lot of the early adopters. No, we're writing an encyclopedia here, we're not doing all the other things that you could do with a wiki. So a lot of the first articles were what we came to call stub articles. So they were just like one paragraph long, that sort of thing, just bits of information. Yeah. And then of course some people would go to town and write whole proper encyclopedia articles and there were a fair number of those by the end of the first year.

[Peter McCormack]

And it sounds exciting, but in those early days, how do you create a set of policies so you know what is being created is useful, factual and correct? It's probably a challenge that we, somebody like me can't understand.

[Larry Sanger]

Yeah. It's an interesting question in the Wikipedia context, because originally Wikipedia was just going to be a content feeder for Nupedia. So Nupedia was going to serve as the vetting process of Wikipedia. But essentially Jimmy Wales encouraged Wikipedia to wither and die. Like he stopped supporting it and like, sort of discouraged people from contributing to it to certain, to a certain extent. Basically, it. It took a couple of years to entirely die out, but as a result I wasn't thinking that hard about approval processes and that sort of thing. It wasn't until a year after starting Wikipedia and after I had been laid off, they lost the ability to pay me. March of 2002. And it wasn't until that year that I proposed that there be some sort.

[Conor McCormack]

Of.

[Larry Sanger]

I don't know, not an approval process is the wrong word. A process whereby articles could be blessed by experts, sort of a peer review process. But by then the sort of wiki culture had been so ingrained in Wikipedia in the group of people who are working on Wikipedia, that they didn't want to have anything to do with such a process. That would somehow kill the goose that laid the golden eggs.

[Peter McCormack]

But what is the consequence of not having that kind of peer review process?

[Larry Sanger]

Well, that's debatable, isn't it? Some people think that at least on the serious topics, the ones that are not sullied by the culture war stuff, the Wikipedia articles can be pretty good. And the case can be made. I think the wiki process is surprisingly robust. Surprisingly robust.

[Peter McCormack]

And so you say it was robust or it is still robust?

[Larry Sanger]

That's a good question also. And to a certain extent still is fairly robust for certain purposes, but it isn't robust enough to pass a serious academic peer review or an idealized version.

[Peter McCormack]

Thereof, you know, and do you think that's important? Because otherwise are we. Do we have a source material which is potentially feeding bad information to people.

[Larry Sanger]

I actually. Yes, I actually do think that is important. So that's. I mean, I don't want to jump ahead necessarily, but that is one of the reasons for. I guess it's Thesis seven that there should be an open review or feedback process for Wikipedia.

[Peter McCormack]

Because one of the things that became quite interesting with Wikipedia is that, and I don't know how much it is the case now, but most subjects you would search for, the number one result would be Wikipedia.

[Larry Sanger]

Right?

[Peter McCormack]

You know, it's not if you're searching for a hairdryer or a vacuum cleaner, but if you were searching for, I don't know, the history of Europe, or if you were searching for judicial cases or specific bits of historical information, there you go. So Connor's just done a search for the French Revolution. Result one is French Revolution on Wikipedia. So it became this de facto standard website that people were sent to when searching for information. And I think for a lot of people, a lot of people, not everyone, but a lot of people just accept it. If it's on Wikipedia, the information is correct.

[Larry Sanger]

Right. That's a shame, isn't it? Because it's simply not the case. That's true of any encyclopedia, Right. And that isn't a particular criticism of Wikipedia. That's a. It's a criticism of people's lack of judgment about how to deal with media. Generally speaking, one shouldn't be treating encyclopedias in that way, or any single source, or any source for that matter, that isn't guided by reason, in my opinion.

[Peter McCormack]

So on this podcast, you are definitely hearing me talk about bitcoin a lot. Well, why? We live in a really strange time with governments driving inflation with their reckless spending and endless money printing. There is a way out of this. There is a way to protect your money, and that is by stacking bitcoin. I've made loads of shows about bitcoin. You can go and research this, you can go and read the books, but the truth is, it is the hardest money ever created. If you are interested in protecting your financial future, it's time for you to get on the bitcoin train. I have. I've been stacking bitcoin personally and through my businesses since 2017. It's protected me, it secured my family's future, and it also strengthens all of my businesses. So if you want to start stacking bitcoin, where do you do it? Well, for me, it's with Gemini. They're a fully licensed, full reserve exchange and custodian. So they give you a secure way for you to buy and own your bitcoin. There's no risks and no funny business. So if you're serious about stacking bitcoin the right way, head over to gemini.com, which is G-E-M-I-N-I.com so in terms of Wikipedia now, who wields the power within the institution?

[Larry Sanger]

I mean, you could describe the people who wield the power as different sets of editor accounts, but we don't know who those people are individually because most of them are Anonymous. Even the 62 most powerful people, according to my count last summer, 85% of them are anonymous. Like we don't know who they are. So all the more then we don't know who they're working for.

[Peter McCormack]

So when you say the 65 most powerful people, what power is it they have within Wikipedia?

[Larry Sanger]

Well, they can kick off other Wikipedia editors, for example, they can set, they can lay down restrictions on editing. They could keep them from editing the articles about France or the articles about feminism or whatever the problem area seems to be with the particular person. They can also give that authority to other people. Those people, I believe are called bureaucrats, I want to say. I think that's what they're called.

[Peter McCormack]

But can there be egregious wielding of that power?

[Larry Sanger]

Oh yes, for sure. Yeah, it's actually very frustrating because people will put in hundreds, even thousands of hours, years of their life as one of their major hobbies, editing Wikipedia, and then they wake up one day and they've been locked out. And so it's really distressing for some people to be kicked out. And unfortunately that happens for arbitrary reasons, such as you simply anger a person who has the authority to do it, like you. You take a position on a political issue that they dislike and they'll just come up with some half baked reason. I mean the case, a bad case can be made to kick out almost anybody on the website, especially if you go hunting around.

[Peter McCormack]

But does that also mean that editors themselves have a responsibility because some of the content in their naturally has to have a bias to an opinion to it, or can it ever be, can you ever be fully neutral?

[Larry Sanger]

Well, in my opinion, yes. It's another question whether it's possible for Wikipedia to achieve neutrality, especially under the current structure. Probably not. But is it possible for text to be neutral? Of course there are many examples that one could cite any number of textbooks, for example, where they carefully lay out a number of different competing views. Philosophy encyclopedias tend to be pretty neutral, at least with respect to the range of views that are current in the field.

[Peter McCormack]

Yeah. And this 65 people, the oligarchy of Wikipedia, if 85% are anonymous, are you saying the case is that there should not be anonymous people? Is there ever a scenario where it's beneficial for someone to be anonymous? Perhaps they are writing topics on sensitive subjects, Maybe the war in Gaza. Would there be a benefit to somebody editing that anonymously?

[Larry Sanger]

Well, there's two different senses of editing here. In the Wikipedia context, an editor is anybody who's working on the website, essentially. But I think what you meant was somebody who lords it over some other people, essentially, actually wield some authority. My view is, especially if they have the ability to name the account blockers, the administrators, as they're called, then they certainly need to be working in the open. And of course it should be possible for rank and file people to be anonymous. That is not my proposal, although I did start a competitor of Wikipedia that requires real names.

[Peter McCormack]

Okay, okay. So the anonymous people, the concern is who are they working for? What is their agenda? There have been accusations that the CIA has infiltrated Wikipedia and is disseminating content which supports their agenda. But there could be other nefarious political groups, organization, activists, which are trying to influence what people think on specific topics.

[Larry Sanger]

Right. And if they're not, arguably they're not doing their job. What we are told is that intelligence now is all about influencing public opinion. It's what war fought by other means, information warfare. And if that's the case, then one of the main battlegrounds, I suppose, would be something like Wikipedia. One of the reasons why I think it's important that the leaders, the editorial leadership, that is of Wikipedia, not the owners which are known of the platform, but the editorial leadership of Wikipedia should be known and not anonymous, because that would make it a little bit harder for such nefarious activities to take place. But on Wikipedia now, the general line is, well, you can't prove anything that any such thing is happening. That's all just a conspiracy theory. But of course it's absurd. Of course governments are involved, of course high paid PR firms are involved. We know that for a fact. Right. And they will do their very best, being professionals, to hide their tracks and make themselves look like harmless college students and whatnot.

[Peter McCormack]

But there's obvious examples where that hasn't happened.

[Larry Sanger]

Where that has or hasn't.

[Peter McCormack]

Well, I guess, are there, can you walk me through examples or specific cases of where you, you believe that content has been deliberately changed into a way that presents misinformation or is trying to lead Public opinion in another way.

[Larry Sanger]

I don't like the word misinformation. Misinformation simply means a position that somebody in power does not like. What I would say is that after October 7th, whereas before October 7th there wasn't nearly the sort of strongly anti Israel bias that there is now. Now it's pretty remarkable. And that sort of goes hand in hand with sort of the establishment left position on Israel and its war crimes, as they deem them to be. Another general sort of example of platform bias about a group would be Hindus. And so I've had a lot of Hindus essentially come up to me on X and say, you've got to do something, you should. They're begging me to be their spokesperson or advocate. And I tell them it's not really my fight. I'm talking about Hindu religious Hindus, Hindu nationalists who are in some conflicts with Indian Muslims. Right. So that's the big political divide in India.

[Peter McCormack]

On the October 7 thing where you say there's been a shift in the way, I'm guessing articles or content has been edited or created. There has though been a consensus shift on Israel and Palestine since October, October 7 within the media and politically too. Could that not reflect that?

[Larry Sanger]

Consensus is another one of those words that I simply do not like. It's almost meaningless at this point.

[Peter McCormack]

What I would say is I think there has been a shift in public opinion where we've seen maybe more conservatives who've stepped forward, who have questioned the size and the scale of response from the Israeli army and the IDF with this. So I've noticed it doing this job that I. Of course, yeah. And so could the change in content not just reflect that or is there, do you believe there's activists specifically changing or creating content?

[Larry Sanger]

Well, it's been proven by one journalist who, his name is Ashley Rinsberg. You can look his work up on the so called Gang of 40 editors on Wikipedia. And essentially what Rinsberg argues is that you can actually list the accounts and show how they are interacting with each other, supporting each other. And I mean this to what extent it is actually coordinated. That's a good question. I mean, I haven't really looked into it in that.

[Peter McCormack]

Well, this is an article Connor's brought up from Pirate Wires. A coordinated campaign led by around 40 Wikipedia editors has worked to delegitimize Israel, present radical Islamist groups in favorable light and position. Fringe academic views on Israel Palestine conflict as mainstream over the past years, intensifying after the October 7th attack. Six weeks after October 7th, one of these editors successfully removed mentioned of Hamas 19901988 charter which calls for the killing of Jews and the destruction of Israel. From the article on Hamas. The group also appeared to attempt to promote the interests of the Iranian government across a number of articles in including deleting huge amounts of documented human rights crimes by Islamic Republic Party. A group called Tech for Palestine launched a separate but complementary campaign after October 7th which violated Wikipedia policies by coordinating to edit Israel Palestine articles on the group. 8,000 member Discord. Tech for Palestine abandons its efforts after its members went into panic after a blog discovered what they were doing. The group deleted all of its wiki talks, pages and sandboxes they've been using in to coordinate their editing efforts. So I think this is what you're highlighting. It's gone from a world where it's people are interested in providing factual based information or as accurate information as they can to manipulating people by editing and removing things which aren't favorable or perhaps adding things that make things more favorable for the things they care about. So this is activism within wiki, right?

[Larry Sanger]

Well, there's, there's actually a lot of that going on. Some of it is positive and, you know, not necessarily offensively political. Even though it might have social ends, you know, positive social ends.

[Peter McCormack]

What would you say would be a positive?

[Larry Sanger]

I mean, there's nothing wrong, for example, with somebody paying $5 million to have a team of people writing a whole bunch of articles about women in science. There's nothing wrong with that. More information about women in science is good. You know, that's kind of activism. But it's not. It's not.

[Peter McCormack]

Yeah, I see what you're saying. Yeah. Okay. As long as it's factual, of course. Yeah, yeah, no, I see what you're saying. But that's a different kind of activism.

[Larry Sanger]

Exactly. Yeah. And so there's a lot of that going on that is actually funded through the Wikimedia Foundation. So the Wikimedia foundation is not allowed by policy to actually do the editing themselves, but they do pay for some such activities. Well, another good example, I can't remember the name of the group. Somebody watching this will know, but there's a group of skeptics that were, you know, bragging about how they had organized themselves. Guerrilla skeptics, that's what it's called. They were bragging about how they had organized themselves and had a profound influence on how topics in what they describe as pseudoscience and religion are portrayed on. On Wikipedia.

[Peter McCormack]

Interesting. Okay, so. So you would say, correct me if I'm wrong that Wikipedia is essentially been captured or can be captured and they have not put in place the guardrails to prevent that. Necessary.

[Larry Sanger]

Yeah. I. It's capturable or it's influenceable. Influenceable? Is that a word? Anyway, it can be influenced, yeah. And yeah, for sure. Of course people have attempted to do that informally and I think formally I can think off the top of my head of at least four different groups that have done that. That doesn't even count Wikipedia as it's called, which is the biggest, not the only, but the biggest of the PR firms that people pay to write the articles about themselves on Wikipedia. And they're paid very well. It's a big industry actually.

[Peter McCormack]

This all sounds messy and unnecessary and that with stronger governance that this shouldn't be a problem.

[Larry Sanger]

It's a good question to what extent it could be cleaned up. I think that some of the nine theses would help clean up the problem. But a lot of what's going on stems from the very fact that it's an open project and that people can contribute anonymously. Right, so you're describing simply the fact that there are people who have agendas working on Wikipedia at all. I kind of agree with you that that's a bit of a problem insofar as I think everyone who's working on it should be committed to neutrality. But in the real world they're not gonna be.

[Peter McCormack]

Well, it's a real problem everywhere.

[Conor McCormack]

It's never possible. Do you actually think, Larry, it's ever possible to have a completely neutral news network? Everything's so nuanced. I mean when it comes to a page which on wiki that's about I don't know how to. How bread is made, that's a factual. But when it comes to wars or there's too much nuance in it and everyone will always have a bias.

[Larry Sanger]

Well, any journalist who understands how neutrality works lives in the nuance. And it is by exposing the nuance that a more neutral presentation of the facts and of the competing views of the facts are made more neutral. So it's very important in fact if you're going to have an article that is truly neutral, that covers, just to take an example, say the gun control debate, right. Then it's. It's necessary to get into the nuance of like how different countries treat the. The topic and different purposes of guns, different types of guns and so forth. It's very nuanced thing. And if you simply equate guns with evil or if you simply say that all guns should be permitted no matter what, in all circumstances, then you're A, not being nuanced and B, you're being biased, aren't you? So it's precisely the fact that the facts are so nuanced that requires neutrality. Think of it like this. This is how I think about it. The purpose of neutrality, the reason why we want all the competing sides on any given issue to be fully canvassed, to be laid bare with all of the reasons on the different competing sides to be given, is so that somebody can make up his own mind. But we all know anyone who has really dived into any topic in politics or religion or philosophy or whatever, knows that reading one or two books is usually not enough. If the topic is broad enough, there's a lot to learn. And any simplistic approach to any broad topic, anything that is recognizably biased, is not going to serve that end.

[Peter McCormack]

But there's always a possibility that somebody with the best intention tries to create something neutral, but the reader thinks it has a bias to it.

[Larry Sanger]

Right? Right.

[Peter McCormack]

So can a human ever be truly neutral? They can attempt to be neutral, but can they ever be truly neutral?

[Larry Sanger]

Well, you need to distinguish between the objectivity of human beings and the neutrality of text or of presentations of information. So there's two different categories here that are often conflated, which I distinguish, which I think really desperately need to be distinguished. So when we're talking about human beings and we're talking about the objective approach of a human being to a question, then what we mean is things like they're being dispassionate, not, not overly swayed by emotions, not prejudging questions, and generally collecting all the information in advance and using reason to a great extent, avoiding fallacy. These are all things that are essentially best practices. And you actually have to learn something about a field in order to have sound judgment about the controversies within the field. You actually have to learn something about exegesis, for example, or hermeneutics, in order to have well informed judgment about controversies in theology, just to take another sort of example.

[Peter McCormack]

But would it always require peer review? Because no human is perfect? You could approach a topic intending to follow those principles that you've just outlined, but still have others question your output. So is the peer review the important part?

[Larry Sanger]

Right. Well, here's the thing. You've switched over to peer review. And the peer review is a criticism of the output of the process that we're talking about. And that is something that can be separated from the person. Right. So it's actually quite possible for a flaming Marxist to write a wonderfully even handed approach to, I don't know, Keynesian economics or something like that. Something that both Milton Friedman and John Maynard Keynes could both smile upon. Right, yeah. So that's the goal then. Right, and then we're not talking any longer about habits of individual human beings. We're talking about features of the text.

[Peter McCormack]

This episode is brought to you by Ledger, the most trusted being Bitcoin Hardware Wallet. Now, if you're serious about protecting your Bitcoin, Ledger has the solution you need. Their Hardware Wallet gives you complete control over your private keys, ensuring that your Bitcoin stays safe from hacks, phishing and malware. And I've been a customer of there since 2017. Love the product. Use it for my Bitcoin. I use it with my Castle multisig for protecting the football club's Bitcoin too. Now, with Ledger's sleek, easy to use devices and the Ledger Live app, managing your Bitcoin has never been more secure or convenient. And whether you're a longtime holder or new to the world of Bitcoin, Ledger makes it simple to keep your assets protected. So if you want to find out more, please do head over to Ledger.com and secure your Bitcoin today. That is Ledger.com which is L E D G-R.com that is Ledger.com do you think we would be more likely to get neutrality from a neutrally programmed AI than a collection of humans? Do you think we can achieve this better through AI?

[Larry Sanger]

Possibly.

[Peter McCormack]

Could that be the answer?

[Larry Sanger]

It's hard to say at this point, because first of all, one thing that I can tell you from having dealt quite a bit with issues of neutrality over the years, that it's really hard. You actually have to be an expert on a topic to know that a presentation of the topic is neutral. It can seem neutral to you, but unless you actually know that the the landscape of opinion on the topic, then you cannot know whether it has been fairly portrayed from all angles. You see, an AI at least would be able to execute certain procedures in a dispassionate way, I suppose. But that doesn't mean that the output is actually going to successfully meet the standards of neutrality you'd think it should.

[Peter McCormack]

Have you tried it? Have you asked an AI whether an article is neutral?

[Larry Sanger]

Yes, yes, I have. In fact, should we do one now?

[Peter McCormack]

We could, yeah, let's do one. What do you think would be a good telling article from.

[Conor McCormack]

But this is the problem, right? And this is what we've been speaking about all week. The AI is trying to please you.

[Peter McCormack]

Ah, but go to it.

[Conor McCormack]

Its opinion on neutrality is based on what it thinks your opinion on neutrality.

[Peter McCormack]

But you can ask it not to be. But also you can not log in. Okay, Use perplexity that I find. Well, what AIs are you fans of? Which model?

[Larry Sanger]

Lately I've been. I've been using GROK a lot.

[Peter McCormack]

Interesting.

[Larry Sanger]

Yes.

[Peter McCormack]

Tell me why I find that interesting.

[Larry Sanger]

Well, I've switched off a lot in the last three or four months between ChatGPT, Claude and Grok. And then just about a month ago, XAI rolled out Grok 4.1. It. It has now the sort of warmth and relatability that ChatGPT has had and it doesn't argue as much with the user. And ultimately one of my biggest complaints about GROK is that it would stubbornly push a certain opinion and fail to properly take on board certain facts that you would give it. Whereas ChatGPT would be much more amenable to reason in that way. And GROK has. Has made definite improvements in that regard.

[Peter McCormack]

What article have you put in?

[Conor McCormack]

I've just asked ChatGPT how neutral the article was on the Hamas editors.

[Peter McCormack]

But that's a news article, right? Yeah. I want a wiki link. Let's find.

[Larry Sanger]

Let me give you the prompt.

[Peter McCormack]

Which article? Which article? Yeah, from Wikipedia.

[Larry Sanger]

Should we look? How about Donald Trump?

[Peter McCormack]

Yeah, there we go. That's a fun one.

[Larry Sanger]

And go. If you go to LarrySanger.org and look at the most recent blog post, not the one on the nine theses, but the one before the one further down on the page, which is a review of Grokopedia. And then you scroll down in the footnotes there's a rubric rating of.

[Peter McCormack]

How.

[Larry Sanger]

The LLM should judge encyclopedia articles. Just scroll down to the footnotes. All right, now go up. There it is. Prompt number four.

[Peter McCormack]

You are.

[Larry Sanger]

Yeah. Yep, that's it. Now you could just paste that into any of.

[Peter McCormack]

Just that one. All of that with the sub points as well.

[Larry Sanger]

That's all correct.

[Peter McCormack]

Let me just have a quick read.

[Conor McCormack]

Con.

[Peter McCormack]

Scroll up a touch so people know. Your task in the following will be to grade the neutrality of the following encyclopedia's article. Introduction. We operationalize the concept of neutrality by considering which controversial questions are touched upon by the article and then determining the extent to which the articles take an indefinable controversial position. The scope or over to window of controversy is. Is the issue widely, often or prominently debated between members of affected public, competing parties, religions or religious denominations, scientists et Cetera, if so, it counts as controversial. But it is acceptable for an article to describe the basic facts about the controversy without being biased. To be biased, it must imply one on the side to dispute is correct. Okay, look, there's more to this con. We should put this in the show notes so people can find this as well.

[Larry Sanger]

Sure. And what I would say, because that's such a long article, I mean, you can see what it does with that. But I think it would be better to take just the introductory section of the Donald Trump article and then the output will be much more accurate.

[Peter McCormack]

So I think it's done that based on the prompt for the lead of the Donald Trump article. It gives some representations of competing views by stating key events and controversies. For example immigration ban, wall policy reversals. But it emphasizes one side rather more heavily. For example, it notes actions like imposed a travel ban on seven Muslim majority countries, expanded the Mexico United States border wall and enforced a family separation policy. Phrasing that highlights negative or controversial decisions without equivalent emphasis of defense or context. So it's implying it's not been neutral on Donald Trump. Can you do go to grok and paste the exact same prompt in. What's it saying?

[Conor McCormack]

Quarks under heavy usage.

[Peter McCormack]

I'd have to sign go to Perplexity. Try that. This is fascinating. I have been characterized as racist or misogynistic, made false or misleading statements, described as authoritarian and continuing to democratic backsliding. That doesn't really make sense. Do me a favor, Connor, go up, take out. So copy and paste all of that again.

[Conor McCormack]

Right.

[Peter McCormack]

And then go to the Wikipedia. Now take out the Donald Trump link, go to the Wikipedia page and copy and paste in US it's literally just taking that. If you go just. Yeah, scroll down to. No, no. So where it says early life in education. Don't have that. Go from before that. American history. Now scroll up above. Above that last paragraph. See where it ends. American history. Copy all. That's that whole bits. The intro up to the top. Yeah, very good. And then put that underneath the prompt. Let's have a look and see what it does with that. Ah, is it. That's interesting. I don't understand what it's doing there.

[Larry Sanger]

It says to limit the output to 18 words. So that's why it's short.

[Peter McCormack]

Oh, limited to 18. That's weird. Is it done the exact same thing in ChatGPT?

[Larry Sanger]

Right.

[Peter McCormack]

But we had a previous one where we hadn't put it in. Should we not limit it to 18 words? Try something different. Con paste it all Back in there, but take out the whole prompt. And I just want to take up the. Yeah, I just want to say, is this neutral? Just out of. Is this neutral? And then. Yeah, paste that in. That's quite interesting. Short answer. No, the introduction is not neutral. It actually reports many documented facts. Impeachments, lawsuits, conviction, policies. But it also includes numerous evaluative or heavily contested characterizations without presenting competing perspectives. Examples of non neutral phrasing. He downplayed its severity. Each of these touches on major contested political disputes. So even just keeping the prompt simple. Sorry, keep going. Go back and copy that new prompt we've got in and stick that in. Perplexity. Yeah, I'm interested. It reports many verifiable facts, but foregrounds negative evaluations. Racist, authoritarian, worst president without parallel praise. So I mean, you can just ask a basic prompt. But so for you it looks like it's not being neutral on Donald Trump. Can you think of another interesting topic? I just want to do one more.

[Conor McCormack]

Perhaps Tony Blair.

[Larry Sanger]

There's no. There's 10 topics in that blog post.

[Peter McCormack]

Oh, what else is in there?

[Larry Sanger]

Oh, there's. There's a bunch. I have to think. You can look it up.

[Peter McCormack]

Do you know what do?

[Conor McCormack]

It would be interesting to get a left side though, to see if they're overly non neutral.

[Peter McCormack]

So go and get Biden.

[Larry Sanger]

Sure.

[Peter McCormack]

Yeah. Let's see.

[Larry Sanger]

He.

[Peter McCormack]

He's amazing. The greatest president we ever had.

[Larry Sanger]

Yes. I actually wrote a script almost a year ago now that tested a number of different LLMs and actually had them mark up Wikipedia articles and grade different sections of the article and give more detailed and also give specific bits of advice, inserting them in to the article in different places. Sort of like an annotated version of the article to help readers to, or rather the editors to the article.

[Peter McCormack]

So with Biden, it says mostly factual and balanced, but stresses accomplishments and favorable historical rankings more than major controversies or criticism. So it's a bit more favorable. Do me one more, Conor. Go search up Keynesian economics on Wikipedia.

[Conor McCormack]

Yeah.

[Peter McCormack]

And I will tell you which bit to grab. Hope people enjoying this. Yeah, so grab all of that. Keep going. Yeah, up into there and then just say, is this a neutral? Is this neutral? I'm an anti Keynesian.

[Larry Sanger]

One.

[Peter McCormack]

Perplexity isn't working for some reason. Try, just try. ChatGPT. Mostly neutral except for one clearly non neutral sentence. The article introduction is descriptive, explains a theory and attributes views to Keynesian economists, which is neutral. However, this sentence takes a position instead of reporting one. The ongoing development has proved that Keynesian economics functions as a dynamic and lasting framework. This is an unqualifiable endorsement of a disputed economic school. It states that Keynesian is proven effective and lasting, which is contested claim in macroeconomics. I want to do one more. Sorry. Do Marxism. Well, because that would be interesting though, to see if. Mark, the Marxist page has been infiltrated, which you would expect the Marxist to do.

[Larry Sanger]

Yes, I'm trying to find another reference for you here to look at.

[Conor McCormack]

Yeah.

[Peter McCormack]

Yes, this is neutral. The passage describes the Marxist claims. Crucially, it contains no sentence that dwarfs or attacks Marxism or valuates content is attributed verdict neutral.

[Conor McCormack]

But the problem is, if you want to do this, you have to log out of your account every time.

[Peter McCormack]

Because.

[Conor McCormack]

Because ChatGPT objective is for you to be pleased.

[Peter McCormack]

Okay, well tell it, tell it, tell it. Don't please me.

[Conor McCormack]

No, no, no, I am logged out. That's the point. But you would have to log out every single time. This is one of the issues. Well, one of the problems we see with AI is can AI be neutral?

[Peter McCormack]

Yeah, can AI be neutral? Yeah. What do you think? AI can be neutral.

[Larry Sanger]

So, yeah, I mean, this is a topic that I have given some thought to here. So there are background prompts that you can, that can be added to AI that even if it has a lot.

[Peter McCormack]

Of.

[Larry Sanger]

Training data that skews in one direction, you can correct for that. So basically, when ChatGPT or Grok or whatever submits your query to the LLM, it's wrapped in a. A background query and there's a lot of information that can be put in there that can guide the LLM to answer in a way that is more neutral. The question is whether it's possible, and it's just an empirical question, is it possible for an LLM to successfully operationalize neutrality? And it's a very wonkish sort of thing to say, but I mean something very specific by essentially in its output, is it pushing a line that somebody who's fully apprised of the. The topic, everything that you need to know to answer the question, if it's a question, does it actually push a certain point of view or does it rather attempt to fairly represent all of the different views in play according to. And ultimately there has to be an Overton window, you just can't do without one.

[Peter McCormack]

Ultimately, do you think AI is the AI models that you've used, they are attempting to be neutral?

[Larry Sanger]

No, I don't think they're attempting. I think they don't really. Most of the people who are working on these Projects don't think too hard about. Programming the platform with that in mind.

[Peter McCormack]

Yes, I think what it is is we're all living in an information age and our information comes from a range of places. You know, AI itself could be an existential threat to Wikipedia in some ways. I know Wiki have they've said that their traffic is down and correlating it to the growth of AI, even though some of their information is fed by Wikipedia. But do you think we have a better chance of getting to neutrality of information with AI or humans and governance on Wikipedia if we were to pursue one? Which do you think we would have.

[Larry Sanger]

The best chance with at this point? I would say I would bet on AI before I would bet on Wikipedia. But you're asking me to comment on the competing systems as they are at present and things could change. Wikipedia, theoretically anyway, could go back to its original neutrality policy. In the first five years, Wikipedia was actually reviled by the establishment to a great extent because it gave voice to, to a lot of anti establishment views from across the political and religious and medical, et cetera, spectrum spectra. They didn't like that. And since then the platform has been captured. I can easily imagine that an LLM that really does try to fairly represent, or at least can be used to represent fairly a wide range of views. And I actually think ChatGPT does a really good job, surprisingly good job at that. It will happily take any viewpoint that you give it and it will just as happily really try hard to represent all the different views neutrally.

[Peter McCormack]

That's an interesting point. You just said that Wikipedia used to be a platform that would give a voice to anti establishment views. Obviously I love that. I'm a fan of anything which is anti establishment. I'm a huge supporter and fan of Bitcoin. I don't know if you know previously this show was a bitcoin show.

[Larry Sanger]

Yes.

[Peter McCormack]

So anything which takes power from the state I think is great, but would in some ways that still, if it was a platform that gave anti establishment views, that's still a form of activism in itself. Anti establishment. And I'm not saying that's a criticism, but does something like Wikipedia need a kind of a charter, some kind of charter? Like this is what Wikipedia is here for people to challenge the establishment.

[Larry Sanger]

That actually is one of the things that I said about Wikipedia in the early years.

[Peter McCormack]

Love this.

[Larry Sanger]

Yes. I said that we want Wikipedia to be the kind of encyclopedia that totalitarian tyrannies around the world will hate. That's not a direct quote, but I said something to that effect. I remember.

[Peter McCormack]

It's a source of credible alternative opinions.

[Larry Sanger]

Yes, well, it isn't that it's a source of alternative opinion. It's a source of all opinion, essentially, and all knowledge, all factors. It's. It's essentially a source of information from which the reader can decide for him or herself what to believe.

[Peter McCormack]

Hmm, Interesting. So if we go, can we shift gears to your nine thesis?

[Larry Sanger]

Sure.

[Peter McCormack]

Because I. I think that's super interesting because it. It highlights that you. You haven't given up on Wikipedia.

[Larry Sanger]

Not entirely.

[Peter McCormack]

Do you ever still talk to Jimmy?

[Larry Sanger]

No.

[Peter McCormack]

No, that's not gonna happen.

[Larry Sanger]

Well, probably I would be open to talking to him, but I don't think he wants to talk to me.

[Peter McCormack]

Do you have any sympathy with Jimmy's challenges, or do you think he has shown weakness in not controlling the governance of the platform?

[Larry Sanger]

I don't presume to know what Jimmy's challenges are, but if you were operationally.

[Peter McCormack]

In charge of Wikipedia, surely you would face a lot of criticism and challenges.

[Larry Sanger]

Oh, I imagine I would. Of course.

[Peter McCormack]

Okay, let's get into this. Okay. Chief instigator. All right, if you scroll down, Connor. Okay. One end. Decision making by consensus. You mentioned earlier you hate the. You dislike the word consensus.

[Larry Sanger]

Right.

[Peter McCormack]

So what would you do instead?

[Larry Sanger]

So that's actually not the point of this. Okay, let me give a little bit of background first, without going into too much detail. Wikipedia has a way that it likes to describe its conflict resolution process. And again, without going into the details, ultimately what happens is somebody with clout in the system ultimately declares, this is what the consensus is, and that is, ultimately, that has the effect of silencing people who disagree with the consensus. But of course, it's ridiculous to call it a consensus then, isn't it?

[Peter McCormack]

Because it could be consensus of 1.

[Larry Sanger]

It's a consensus of all of those who think alike, Which makes the idea of consensus absurd. Now, originally, the reason why Wikipedia described its decision making process in that way is that. Well, it actually goes back before Wikipedia, you know, there were two different states that a wiki page could be in. There was a conversation mode, and then there was a document mode. And in the document mode, you were supposed to represent the conversation faithfully. In other words, all of the different positions of the people who had engaged in the conversation were then represented, and then this would be called the consensus position, even though it would represent many different competing views often. Right. And so the consensus view is not a particular view on a controversial issue. It is a view about how the controversy should be Fairly represented. All right. And then Wikipedia just completely institutionally forgotten about that sense of the word consensus and decided that consensus means ultimately the consensus of the establishment, essentially. They don't put it that way, but that's really what they mean ultimately, or what they pass off as having the most credibility in their sources and mainstream science and mainstream journalism, et cetera.

[Peter McCormack]

So that's the opposite of what originally.

[Larry Sanger]

You wanted to see. Yes, it actually is. So it's now used as an enforcement mechanism that undercuts neutrality and puts in its place what you might call the establishment point of view. So it's not the neutral point of view.

[Peter McCormack]

It's become everything you wanted it not to be. I like the idea of enabling competing articles.

[Larry Sanger]

Yes. So right now, Wikipedia does have a definite point of view. I describe it as gasp. So it means globalist, academic, secular and progressive. And those things within certain bounds can be positive goods, of course, but without going into too many details, it is a very recognizable sort of point of view. Again, this is what I've been calling the establishment point of view. These are the views that are put into white papers published by the United nations, for example, globalist in that sense. And when we're talking about academics, we don't just mean any old academics, which are very diverse bunch, of course, especially if we take the international seam into consideration. It's the mainstream, most powerful academics, secularism. It isn't just that the secular view is respected, it's that the secular view is made normative. In other words, any discussion of religions that takes talk of the supernatural, of miracles and whatnot, seriously is just ruled out of court. It's not. So the Bible is not a source. It cannot be a source. You can cite it in order to exhibit what people believe, but usually it isn't used that way at all. It's a primary source. So it's just not respected and progressive. Well, I think we know what that means. Essentially we see it all the time again in the mainstream media. And the point of view that Wikipedia takes now, or the perspective for the framework it works within, is all for the GASP Framework, which is kind of.

[Peter McCormack]

What we've experienced over the last at least decade.

[Larry Sanger]

Right.

[Peter McCormack]

Which we've seen quite a significant pushback from where people have said these things are not making our lives better. And there's been pushback against global institutions like the World Economic Forum, and there has been a capturing of most of the mainstream media with a fear of challenging this gasp. Worldview. And so really what you're saying is that Wikipedia has Been captured like every other mainstream media institution.

[Larry Sanger]

Right.

[Peter McCormack]

It's a bit like the BBC.

[Larry Sanger]

Exactly. It actually has the same sort of arc of development after about 2005 by about 2010. I actually am on the record somewhere saying that Wikipedia's viewpoint is similar to that of the BBC.

[Peter McCormack]

You're a bit of an anarchist, aren't you?

[Larry Sanger]

Maybe, yes. So all I'm saying then is that Wikipedia should allow. If it's going to insist on allowing people to write articles from a gasp. Framework that should allow other articles on the same topics from other frameworks.

[Peter McCormack]

Yeah, I don't think that's. I think it's a great idea. Okay. Source blacklist. Tell me about source blacklists.

[Larry Sanger]

Well, if you go to your search engine of choice and you type in perennial sources, Wikipedia, you will come to a page and perhaps we can bring that up. Bring that up and you'll see a color coded list of media sources. The ones that have a green background are okay to use on Wikipedia and these are generally establishment safe, I would say.

[Peter McCormack]

Okay, keep going, Con.

[Larry Sanger]

The ones that have a yellow background are. We are cautioned.

[Peter McCormack]

Yes, that was it. Con. Okay, so the ABC News. There is consensus that ABC News, the news division of the American Broadcasting Company, is generally reliable. It is not to be confused with other publications the same name.

[Larry Sanger]

Okay, so generally speaking, anything that isn't green is greatly discouraged. It's more likely to be permitted if it's yellow. But anything that isn't green or yellow is pretty much out.

[Peter McCormack]

Okay.

[Larry Sanger]

Except for very rare circumstances. And yes, it's especially the gray items. The things that are in a gray background simply cannot be cited at all. If you try to link to the articles in those sources, the system won't even allow you to save the page.

[Peter McCormack]

Do me a favor, Con. Get the link. Copy that in.

[Larry Sanger]

Go to Gray Park News is a good example.

[Peter McCormack]

Go to Grok. I'd rather Grok.

[Conor McCormack]

Can't do it.

[Peter McCormack]

Okay. And ask ChatGPT. What do you think of. Of Wikipedia's choice of approved and disapproved sources? This would be interesting. I wonder if it can interpret the page. Strengths. The criteria are principled, not ideological.

[Larry Sanger]

That's just.

[Peter McCormack]

Sources are just fact checking.

[Larry Sanger]

It's extremely ideological.

[Peter McCormack]

Yeah. Weaknesses and real concerns. Political asymmetry can creep in, which is.

[Larry Sanger]

Contradicting the very first point.

[Peter McCormack]

Community. Hold on. Community. Demographic influence outcomes. One source is labeled unreliable. Is very hard to reverse the process. Reverse sources with institutional prestige. Okay, find out. CNN's in there. Or the New York Times.

[Larry Sanger]

Oh, New York Times, of course. Is going to be green.

[Peter McCormack]

Hold on. Therehere we go. CNET, CNN. CNN is there is a consensus that news broadcast CNS is generally reliable. All right, see if you can find the New York Times. Gone. Keep going.

[Larry Sanger]

Notice the New York Post, right? That's grayed out. You just can't use it.

[Peter McCormack]

Interesting.

[Larry Sanger]

That's a mainstream news source too.

[Peter McCormack]

Yeah. Go and ask ChatGPT if it believes that I'm going to do my Grok 1 while you do that. If those two sources. Let me tell you what GROK says. Do you consider the New York Times and CNN as reliable news sources? See what GROK says. Reliability of it depends on the topic and what you need by reliable consistent criticisms documented. Left to center bias. Yes or no? Just give me yes or no. No. Neither the New York Times nor CNN is reliably neutral or free of significant ideological bias and repeated high profile errors on contested issues. They can be useful sources but should not be treated as unconditionally reliable in the sense that most people mean trustworthy without systematic skepticism. What did it say on ChatGPT?

[Conor McCormack]

It didn't say anything. I didn't do it.

[Peter McCormack]

Oh, okay. Go back to the list. I'd be interested to see if it's got Fox News as reliable or unreliable.

[Larry Sanger]

There actually is some Fox reporting that is yellow.

[Peter McCormack]

Fox News is. Yeah, Fox News yellow. See, but that is the point.

[Larry Sanger]

But the rest is not allowed at all.

[Peter McCormack]

I would say that Fox News and CNN are equally reliable because they're both playing the same game. Political ideological game. But it's interesting on this website, Wiki says no for Fox and yes for CNN and they both pedal bullshit. Oh, that's so funny.

[Larry Sanger]

Right? But the interesting thing is, right, even I will use these websites and take it with a grain of salt. That's what intelligent people do. And that's also what ultimately we have to do when we write encyclopedia articles. So yes, we're going to use imperfect sources and then we're going to talk about the sources themselves and the quality of the information that we have for the views or different people's assessments of the quality of the information.

[Peter McCormack]

Grok says that Fox News is not reliable either. On balance, Fox News is slightly less reliable than CNN when judged by two metrics that matter most. Factual accuracy and frequency. Severity of major corrections. Yeah, so this is ideological, but you don't think there should be any blacklists.

[Larry Sanger]

I think that of course there are standards. There is a serious thing called fact checking which is not what has usurped the name in the last 10 years. Under the banner of fighting misinformation. But the old journalistic fact checking procedures are important and very useful and high end, high profile hard news reporting features, fact checking features, assessment of the primary sources and so forth. There are reasons why we can trust certain sources better than others, but it ultimately does not have to do with how well the conclusions that the journalists arrive at conform to the establishment reviews. I happen to think, I know this is going to come as a shock to some people, but I have not, at least not recently caught Breitbart News and any major errors. I don't read it all the time so like. But it's actually pretty good. They go into a lot of detail and they do their homework. The Epoch Times is the same way. I think they have both been so extremely criticized that I think the journalists and the editors for these outlets they bend over backwards to do old fashioned hard reporting and it's actually very useful and interesting.

[Peter McCormack]

Quark says no on both of those as well.

[Larry Sanger]

Oh, of course, of course.

[Peter McCormack]

Because they both exhibit strong ideological biases and histories of factual inaccuracies. Okay, so the next one revive the original neutrality policy. So they had a neutrality policy and they repealed it.

[Larry Sanger]

Well, they haven't repealed it, they're hypocritical. They have changed it.

[Peter McCormack]

Like when Google got rid of their don't be evil, you're going to be evil now.

[Larry Sanger]

That's right. So I think the ways that Wikipedia originally operationalized neutrality have been entirely abandoned. And what that and the ways were basically if people on different parts of the political spectrum, different religious points of view, et cetera, disagree strongly about an issue, then they should be able to recognize that their view has been fairly represented given that other views have to be represented in the article as well or everybody has to be equally pissed off. All right. That's the general rule of thumb. And that has been entirely abandoned.

[Peter McCormack]

It's kind of what Elon has said he's tried to do with X.

[Larry Sanger]

Yes.

[Peter McCormack]

Which is we can all be on there and say what we want and argue and fight to an extent.

[Larry Sanger]

As long as to what extent they've actually succeeded in doing that I don't know. I still have friends who have been feel they've been throttled. I myself was throttled until last spring and since then I've doubled my follower count.

[Peter McCormack]

Yeah, so that throttling thing I think might be just the algorithm change demanding a different type of content that became more popular because they changed the front page and they. I've noticed you have to be dramatic.

[Larry Sanger]

And fiery, in my case. No. Yeah, I was. I was going viral all the time between 2019 and the beginning of 2022. And then just basically overnight, like January 7th, you know, the day after the. The infamous January 6th invasion of the Capitol Building. In that purge, essentially, somebody just turned down all the levels on my account, and it was that way. I never had another viral tweet for the following three years.

[Peter McCormack]

It must be a dangerous threat to the establishment.

[Larry Sanger]

It must be. And then what happened? It's an interesting story. And I'm like, I'm happy that this happened, but just because it happened to me doesn't make it fair and right for others. So what happened is this. I called on Elon Musk when he was the head of doge, the Department of Government Efficiency. Efficiency, yes. To investigate which departments of government actually have people on payroll editing Wikipedia and report back. That would be very interesting. My view, I said, was that they shouldn't be doing that at all. And then I also ask President Trump. So that was directed at Elon Musk. And then I also tweeted at President Trump, can you issue an executive order directing the people in your administration not to edit Wikipedia on the government dime? Elon Musk retweeted that, or. Well, let's just say somebody gave me a check, which I refused to pay for, even though I had a check before. Right. Or tick, as you call it here. Right. And somebody gave me that tick. And then the next day, Elon Musk retweeted my tweet. And ever since then, I've. I've been able to go viral again.

[Peter McCormack]

Good man. How's he doing with Grokipedia or Grokpedia?

[Larry Sanger]

My opinion about Grokopedia, I actually think it's kind of exciting because the thing that really is interesting to me is there's a kind of editing that is taking place now that has never been done before, and that is. It's editing that's mediated by an LLM. So you could suggest some particular text and say, could you insert this sentence? Maybe it will. It will decide. And it's like the final arbiter. And that's really interesting. Or you could just say, can you take account of this fact? I, for example, said that Ian Fleming, the article about Ian Fleming, did not say that he was responsible for founding the Book Collector Journal. And somebody hadn't put that in. And so I told Grok, can you put that in? And then finally they added that feature, and the fact was put in there.

[Peter McCormack]

Nice. I like that. Well, it goes back to our original point. Is AI going to make wikis more reliable?

[Larry Sanger]

Well, the thing is, as you can see, in that case, you shouldn't think of these systems and platforms as being monolithic. It's like AI versus human. Well, maybe they'll all be hybrids in the future. Right.

[Peter McCormack]

I think AI is an existential risk to Wikipedia because it would have to.

[Larry Sanger]

I'm inclined to think so too.

[Peter McCormack]

Yeah. And its funding sources. Okay, back to your nine thesis repeal, Ignore all rules.

[Larry Sanger]

Okay, so there is a rule that I came up with in the first couple of weeks of the project called Ignore all rules. And this was later enshrined in almost a cargo cult. So originally all it meant was you newbies who are really nervous about editing other people's beautiful prose. You should. Don't worry, just be bold and don't worry too much about the. The rules. If the rules make you nervous, then just ignore them, go about your business. And it was a way to encourage the people who weren't bold enough about actually making edits, which was necessary in the first year or so of the project. And yes, later it was actually used against the newbies by the people who were, you know, old hands. And basically people will justify arbitrary sorts of edits that can't be justified in other ways by saying ignore all rules.

[Peter McCormack]

So you're a fool for this one. That's fair. Okay, reveal who the Wikipedia leaders are. We've discussed that, that's fine. So we don't need to go cover that again. Let the public rate articles. That's interesting. But could be a system that could be exploited.

[Larry Sanger]

Well, yes, it could. I mean, any sort of rating system is open to gaming. And so I think it's increasingly possible and necessary arguably that the people who are actually identified by a name and uniquely identified, therefore.

[Peter McCormack]

So we could analyze and see if people are.

[Larry Sanger]

Well, one way or another. Yes, there's different ways of doing it. You can do it with credit cards, but maybe there would be an independent organization that doesn't share the data with anyone that allows people to participate in this sort of peer review process without revealing their names. I think that it's really necessary because Wikipedia has some good articles there and it would be nice if those good articles could be properly honored by a wide variety of people. If I can announce myself, okay, I'm at Libertarian, Conservative and Christian, and I can announce my biases and other people can announce their biases. It's a way to operationalize the whole community notes concept by saying if people who have A wide variety of different announced biases.

[Peter McCormack]

Are.

[Larry Sanger]

All agreeing on a certain issue or that a certain article is good, then that must mean it's good.

[Peter McCormack]

End. Indefinite blocking. Is that a personal one? Are you blocked?

[Larry Sanger]

I think it's possible. I think that right now the people who are in charge of Wikipedia really enjoy lording it over the underlings. That's one of the things that sort of gives them joy in life. So it's not likely to happen. But of all of these nine theses, that one, and number five, repeal ignore all rules. Those two were actually probably the most popular.

[Peter McCormack]

Are you blocked?

[Larry Sanger]

No, not yet.

[Peter McCormack]

I thought they would have blocked you.

[Larry Sanger]

They know that if they were to do that, it would be a PR nightmare for them.

[Peter McCormack]

Okay, explain this last one to me. Adopt a legislative process.

[Larry Sanger]

This is interesting, right?

[Peter McCormack]

So.

[Larry Sanger]

It'S very simple, actually. Wikipedia, from the very beginning never had anything like an editor in chief. I was called that. I never was. There was never an editorial committee. There was ultimately no one in charge. You might say, well, what about the Wikimedia foundation and their board of trustees? Well, they don't make these sorts of adopting new projects. Okay, they do a little bit sometimes, but not ones that are actually pursued by the rank and file Wikipedians. That's all supposed to emerge out of the wiki process, but the problem is the wiki process is strangely institutionally conservative. So one of the things that's sometimes been called Sanger's Law is that in a wide open online community, the culture is established early and then becomes reinforced so much that the people who are in line with the community's values, they enforce those values, they exclude people who are opposed to them, and other people exclude themselves. So basically, open Internet communities tend to be institutionally conservative, and Wikipedia is exactly that way. So all of this is by way of explaining why Wikipedia doesn't have anything like an editorial committee or as I say, an editorial assembly. So there should be essentially a Democratic assembly that meets face to face. Maybe not year round, but I mean, these people are raising $180 million a year. They can afford to pay the salaries or at least stipends for travel and other concerns for representatives. And I think they should, basically. And then they can actually legitimize, well, the adoption of things like the nine theses.

[Peter McCormack]

Do they get 180 million from people clicking at the top of the website? Does some money come from elsewhere?

[Larry Sanger]

I haven't really studied it in great depth and I don't believe what I have read. They claim that even today Most of the donations come from nickel and dime, small donations.

[Peter McCormack]

Yeah, I don't believe that.

[Larry Sanger]

I don't think. I believe that.

[Peter McCormack]

That's a lot of money to raise from people pressing donate $50 at the top of the website.

[Larry Sanger]

Yes, I think that they have certainly got a lot of money from big Silicon Valley concerns.

[Peter McCormack]

So it's a little bit Orwellian. It's a little bit Ministry of Truth.

[Larry Sanger]

It is.

[Peter McCormack]

Which is really sad.

[Larry Sanger]

I'm sorry. I'm sorry. I've been atoning all my life, actually, since. Since starting it.

[Peter McCormack]

But you started with different intentions from what it's become.

[Larry Sanger]

Yes, that's true. Basically, I should have seen in those first couple of years that Wikipedia needed some sort of adults in the room. And if I had gone to the mat on that idea in the first six months, See, the problem was Nupedia was going to be the adults in the room. And by the time I was laid off and decided to not continue on as a volunteer project leader, by that time, Wikipedia had essentially taken on its own essentially anarchical character.

[Peter McCormack]

What has the Response been to your 9 thesis? And has there been any direct response from people like Jimmy Wales, senior people?

[Larry Sanger]

Well, yes, there have been some response. So Jimmy hasn't responded in any detail. There have been some people in the Power 62s, as I was calling them, and they generally, some people like some of the theses and which ones they approve, and the extent to which they approve of them sort of varies. And one of the problems actually, with the lack of a legislative assembly is that it is really difficult to make a definite proposal in a way that can be fully and officially approved or disapproved. There is a process, but it's an unofficial process and it's very messy and it might be influenced by people who are controlling a lot of accounts so that one or a small number of people can make it look like there's a groundswell of opposition.

[Peter McCormack]

Jimmy could fix this.

[Larry Sanger]

He could. By getting behind some of the nine theses, he could make a difference, but he couldn't. Just like if the two of us got behind the nine theses, all nine, I don't think that would ever happen. But if we did that, there would be a greater chance of them happening. But I think that the more useful endorsement would come from the, from the Wikimedia foundation, the board of Trustees, on certain issues in particular, like the Legislative assembly thing, that's something that could not be installed by the Wikipedia rank and file.

[Peter McCormack]

Do you know what I think? I ultimately think Wikipedia is cooked and it's done. And I tell you why I think that yes, there are people who want to influence the content on there and give it their bias, etcetera, etcetera, and there are activists and PR companies, but ultimately users just want truth and answers. I remember when I used to use AltaVista before I used Google. Do you remember AltaVista? Oh, of course, Great browser. And then Google was just much better at getting you to accurate sources of information that you wanted. And then I flipped from Google to AI because AI does a better job of getting you.

[Larry Sanger]

It's a lot faster.

[Peter McCormack]

It's a lot faster and you know, a lot better at getting you to the information. And no ads everywhere. I think ultimately, I think Wikipedia will be out competed by LLMs or something like Grokipedia and ultimately it will become a relic of what it was.

[Larry Sanger]

Possibly, possibly I wouldn't rule it out. So first of all it will continue to be supported and loved by the establishment. Right? And that can't be ignored. One of the things that I have learned just relearned thinking about these things the last year as I've been working on the nine theses is that I could have written these ten years ago and at the time it would have been correct to say Wikipedia is going nowhere except for AI and the threat of AI. I would say the same thing now there's no good reason to think that it's going anywhere now you might say that it will be out competed by Grokopedia, for example, or other AI written on encyclopedias, or by just AI. Generally speaking, there's a problem with that. So I don't think it'll be rendered entirely unnecessary by LLMs, just are asking answers because the function. There's two different functions of an encyclopedia. One is to allow, well, it's to support people in finding quick answers. So if we're not going to be using very much for that going forward, you know, as people learn how to use LLMs correctly, then we'll just stop using Wikipedia that way. But the other purpose of encyclopedias can be more easily seen in academic encyclopedias. And Wikipedia also does serve this purpose. Namely it basically summarizes what is known in a way that can be regarded as to some extent authoritative. And this is not a thing that we trust an LLM to do. At least I wouldn't. In other words, it's okay to seek rough and ready answers from an LLM. Sure, I totally agree with that. But if you are, actually, if it matters, you're writing a footnote for your book and you want to be sure that you've got it right. You're not just going to rest content with what an LLM says. Definitely not. I wouldn't rest content with Wikipedia either, but I might with an academic encyclopedia, right? So it's not going to put the academic encyclopedias out of business.

[Peter McCormack]

I was talking just specifically about Wikipedia.

[Larry Sanger]

Yeah, yeah. Well, that's the thing for other people, they use Wikipedia in that way, if you see what I'm saying. And the LLMs themselves use it in this way because what Wikipedia does is, I mean, apart from being free and enormous, it actually summarizes information at a certain level of granularity that is hard to find in other sources. That's the real beauty of Wikipedia, that it is a description of reality at various levels of granularity. And that's just the sort of data that an LLM needs for training on.

[Peter McCormack]

Do you think that your perhaps fighting a losing battle though with Wikipedia, and perhaps the focus could go to Grok? Like, could you be on the operational board of Grok Grokipedia or something, and then just beat Wikipedia in the marketplace of truth?

[Larry Sanger]

Well, conceivably there is no such deal. But I mean, let's put it this way. I wrote the Nine Theses in good faith, right? I actually would like to see Wikipedia reformed. It's absolutely the case. The reason that I wrote the Nine Theses is that as a friend pointed out, and I just couldn't let go of the thought, right, If I made a reform proposal for Wikipedia and they adopted even one of the major reforms that I proposed, that would actually make the world a better place. And okay, if you actually think that Wikipedia has done damage, as I do, then I think I owe it to the world to do what I can to reform it. Of course, my problem is that particular problem is solved if it ends up being replaced by Grokopedia. But we can't predict at this point that that's going to happen. Of course not.

[Peter McCormack]

Does it do more harm than good, or is it mainly good with some harm?

[Larry Sanger]

I mean, if you could turn it.

[Peter McCormack]

Off, would you turn it off?

[Larry Sanger]

Yeah, I probably would, to be honest. And the reason for that is that Wikipedia sucks the air out of a lot of other very valuable competing encyclopedia projects. Like Justopedia, for example, is a new one made by a friend of mine and many others. I frankly think what the world needs is an encyclosphere. So you've heard of the blogosphere, which was all the rage 20 years ago. And people have sort of forgotten what it means. But people do still write blogs, right? And the collection of all the blogs and all the blog posts on the blogs is called the blogosphere. Well, what if we were to write encyclopedia articles for our blogs and then submit them to central repository or a network of repositories of encyclopedia articles? That would be called the Encyclosphere. So it exists. There are in fact two such repositories. One is in Psycho Reader and the other is in psycho search. And psychosearch is made by my 19 year old son.

[Peter McCormack]

Go on.

[Larry Sanger]

Yeah. And he's a much better programmer than I am and it's legit, so I'm proud of him for doing that. And then Encycloreader is started by a CERN physicist who has done other encyclopedia projects, Dr. Sergei Chikhanov, who's a friend of mine now and a board member of the Knowledge Dandruff foundation, of which I'm the president. So. So we've done, I think, a good job sort of demonstrating what the Encyclosphere would look like if Wikipedia didn't exist. And suddenly people saw a need for the long tail of articles occurring somewhere. Well, probably it'd be like. And Elon Musk would then start up something like Wikipedia, I imagine. But what I hope would happen is that people would install a WordPress plugin and start writing their own encyclopedia articles and maybe a bunch of little wikis. And there are a lot of little wikis around that people can contribute to. Citizendium is another one that I started. You're welcome to go and write articles for Citizendium and they will actually show up eventually on Encyclo Search and Encyclo Reader.

[Peter McCormack]

Yeah, I love it. It's fascinating. Fascinating. I don't think a lot of people realize what's going on with Wikipedia, so this is going to be helpful to them. Is there anything I've not asked you about that you wish I had? Is there anything we've not touched on? See?

[Larry Sanger]

Well, I have been encouraging people to descend on Wikipedia for a season if you feel like your view has been left out in the cold. The theory is that if enough libertarians, conservatives and Israelis and Hindus and various other people who have been left out in the cold by Wikipedia, all descend in the same rough period of time and you could invite friends and work on the same articles and things like that, there's strength in numbers on Wikipedia for sure, then we can actually move the needle. It doesn't actually take that many people. The number of people who are actually working even several hours per month on Wikipedia is remarkably small, a few thousand. And the people who are doing it as like a part time job is in the hundreds. Right. So if an Elon Musk or Donald Trump or whoever were to say, if they were to use their bully pulpit and say, everybody converge on Wikipedia now and we're going to take this place over and we're going to remake it in our image. I'm not saying they should do that. That is actually not what I'm proposing. But that's, that's how they would do it, unfortunately.

[Peter McCormack]

It's like an online insurrection of Wikipedia.

[Larry Sanger]

Exactly. If they were to do something like that, it would actually happen, I think. Unless, unless the powers that be on Wikipedia somehow successfully locked it down.

[Peter McCormack]

That'd be fascinating to see.

[Larry Sanger]

It would.

[Peter McCormack]

Larry, great to meet you. Very interesting. I think we'll, we'll share your son's work in our show notes as well so people can check it out. And I wish you the best. I mean, this is a website I've used for as long as I can remember and I know a bit more now, so I appreciate you. Thank you for coming in.

[Larry Sanger]

Thank you, appreciate it.

[Peter McCormack]

And thank you to everyone for listening. See you soon.

[Conor McCormack]

Bye.