Ciarán Murray on building a better Media

With trust in news platforms collapsing and AI flooding the internet with noise, the answer may lie in the wisdom of crowds

“You are the media now,” said Elon Musk in a speech earlier this year encouraging people to turn to citizen journalism on X and away from legacy media outlets. It’s a statement that captures both the growing distrust in traditional media and the shift toward independent journalists and social feeds as primary news sources.

Its no new thing that trust in legacy media has been declining. For the past decade, studies and focus groups have come back with the same conclusion — people are growing increasingly sceptical of mainstream outlets. When you watch the polarised theatrics on mainstream news channels, can you really blame them?

In response, more people are turning to independent platforms like Substack, or following individual voices on X and YouTube, looking for something that feels more authentic, more direct and seemingly more accountable to them, not to institutions.

The idea of a decentralised media landscape powered by citizen journalism sounds great but what happens when there’s no real quality control or accountability? Are we heading for a future of misinformation, algorithmic junk, and headlines that are just plain wrong?

Traditional media tackled this with layers of editors, lawyers, and oversight. But when the call is for decentralised news with no central authority, things get messier. X tried to address this with Community Notes (although I’m seeing fewer of those lately), but with the sheer volume of content and no real consequences for getting things wrong, it feels more like a band-aid than a solution.

Ciarán Murray at Olas Protocol (not to be confused with the AI agent platform) thinks he might have the answer an open platform for citizen journalism combined with decentralised fact-checking, powered by web3 incentives that reward accuracy through consensus.

Having recently stepped into independent journalism myself and covering web3 for years, it’s an idea that caught my attention. Could we one day have a media ecosystem with no centralised control? And, dependent on that, could we also build a future where AI models are trained on better, more accurate information?

Key takeaways:

• Traditional media is broken, with centralised ownership shaping narratives read by the majority of the population. Trust in traditional media is at historic lows.

• Decentralised infrastructure and frameworks similar to prediction markets, where people have to put skin in the game to register their opinion on certain information could help, creating a kind of quality control system for news articles.

• There have been attempts at doing this before but only in the past two years have a combination of technologies come together in a way that could make meaningful impact, enabling micro-payments, proof-of-humanity and mechanisms to encourage collective intelligence.

• These protocols could in turn create an environment that equates to more accurate AIs. If the information going into the AI is verified as true by independent individuals, provable free of collusion, its outputs can be more accurate.

Food for thought:

• Ciarán talks about Bayesian Truth Serum, a scoring system that can be implemented to encourage truthful participation in surveys where objective truth is unknowable. Basically, participants are rewarded according to how common their answers are, theoretically incentivising truthfulness. It’s a framework that was first proposed in 2004 but, for Cieran, it could be particularly impactful now, combined with other available technologies – this is a link to the original paper in case you want to learn more, along with two more recent reflections on its effectiveness, here and here.

• I found this article featured on the Cosmos Institute’s Substack recently which looks into AIs impact on humans' continued ability to think. It builds on some of the studies (another is here) I’ve seen published which found that with continued usage of AI, humans' ability to think critically and for themselves may be reduced. This could be particularly relevant in an environment which calls on humans for information verification. On one hand, a decentralised approach could result in more accurate verification which I’m inclined to support, but on the other, if humans are thinking less anyway, and rewarded based on the most common answers, the results may simply follow an unquestioned consensus.

• Last year, while working for Digital Frontier, I spoke to Winn Schwartau, an encryption and cybersecurity expert about his book, Metawar, where we discussed (among other things) the impact of centralised entities and the use of algorithms on the shaping of global narratives – “Metawar is the manipulation of your reality” kind of sets the tone of the conversation which concluded in a call to action for more mechanisms for critical thinking and collective intelligence. You can read the full write up here and listen to it in the context of building effective AGIs in a subsequent podcast I led here.

Image by Cassiopeus

Isabelle Castro
Maybe we can start with your inspiration for the idea behind Olas Protocol? And explain to me a little bit about the process of setting up Olas protocol because I looked it up twice. The first time I looked it up, I thought we were going to be talking about monetising AI agents.

Ciarán Murray
That's the other Olas.

Isabelle Castro
Yes. I realised that when I looked on your LinkedIn.

Ciarán Murray
We're in a name dispute with them. They are more well known but they rebranded from Autonolas and they never took out trademarks. I did a trademark check when they were Autonolas, so I even knew of them when they weren't Olas and I took out trademarks and now we've got this confusion, but that's not us.

I'm sure there will be AI agents on our protocol. But it's very much a human based system for human crowd wisdom that I'm sure will eventually feed into AI. Because where else does AI get its information other than from humans? It is getting incredibly smart and being able to generate knowledge itself from what it knows. But in our context, the news comes from the physical world. And AI doesn't live in that world. It can't see stuff happen on the streets. It can't see corruption happening in governments. Well, it may do actually, but not as easily as humans. And all these things we rely on the media to inform us for.

So I'm just trying to make a better media, basically one that is more economically sustainable, one that's more trustworthy and ultimately more accurate. And we're devising loads of mechanisms to those ends.

Isabelle Castro
Okay, take me through those mechanisms. How does it work? How does your framework equate to a better media?

Ciarán Murray
Okay, so first of all, media has since the dawn of the printing press been controlled by somebody, some entity has controlled it. And it tends to be desired by powerful people because they can have huge influence. And that's why you have oligarchs or even just rich businessmen like Bezos just buying a newspaper because he can use that for his own ends. So that has been a problem throughout history because you've got a conflict of interest immediately and that often manifests as bad information, propaganda or even if it's not overtly, you know, a direct influence you often have because they are centrally owned and centrally controlled. You can often have group think and people miss seeing the wider truth. By having the first ever ownerless media platform, it removes that conflict of interest and reduces the possibility of groupthink.

And then we have a new economic system that we believe information is really a natural public good. It's not really something that you can easily charge money for without being a bit of a hack. It's supposed to be free. Once somebody says something, you can't put it back in a box. So media business models, particularly advertising, are kind of something that's tacked down on top and they actually make the product worse.

Every time a newspaper that's particularly reliant on advertising is trying to make more money, they're trying to optimize for the advertising model and not their actual product often. So there's a tension there. And even if that tension didn't exist, you know, the way the Internet's gone, it just is not sustaining the media industry and you're having a lot of gaps arise, particularly in local news. This information that is really useful and not just useful, but demanded, isn't actually being supplied because the current business models can't supply it.

So we believe that you can just embrace the fact that it is supposed to be free and actually get people to choose voluntarily pay for it. Under the current system, that would never work. I mean if you say that to somebody who works media now, theyscoff at it. And rightly so because the friction involved taking out your credit card to make a payment, particularly if it's going to be a small payment, you get likes of wikimedia, they get 2% of their readership to periodically pay. That's very periodically. And they take in $200 million a year just from that alone. Even with the friction of having to take out your credit card and they ask for bigger payments, $15-$20, we think with the way the Internet's going and you being a journalist are well aware of this. Payments are going to become more cash like, more like walking by a busker on a street or more in a digital context, more like just liking a post on social media. Basically completely frictionless to send somebody 5 cent, 10 cent, 20 cent.

So we believe if you actually just remove paywalls and largely remove advertising, we probably won’t be able to completely remove it because people build on top of our protocol and they might instigate it. That's up to them. We're just the base layer. But by improving the reading experience and making the tipping or payment experience so trivially easy, we think we take away paywalls and somebody gets to the end of an article that they really like and they'll just go, “You know what? I'll just like that with 10 cent, 20 cent, 50 cents.” And because there's no paywall, the amount of eyeballs on that are greatly increased. So we think a larger readership coupled with getting the tip rate up from let's say Wikimedia's 2% to 20-30% and then suddenly we have vastly improved revenues in the system.

So that's the money coming in, problem solved along with I guess the trust problem by not having a biased owner. And then there is the quality control systems where if we don't, we're not a centralized institution so we don't have this hierarchy where there's an editorial review board. We have to go to the crowd. And to do that you generally rely on the two best tools we have for decentralized quality control. That's markets and reputation. But there are huge risks involved with that because sometimes the crowd can be stupid, particularly if you're dealing with small barriers to entry and your tip is just 20 cent to have a say in this article you might just, it doesn't really hone your calibration, shall we say to use forecasting speak.

It's in markets where you have to drop thousands, where you really stop to pause because you're the skin in the game there is so much higher. So we've had to devise mechanisms to kind of tease out the smarter members of the crowd. Also prevent collusion because you know, for the same reason people buy media platforms today you will find people try to game outcomes in this for their benefit outside the system. So we have to have all sorts of anti collusion controls, proof of humans, so there's only ever one person, no bots, all sorts of bells and whistles like that basically to make sure the crowd wisdom really is wisdom rather than nonsense, I suppose.

Isabelle Castro
So who's this kind of geared towards? Is this for like the individual creator, kind of like a substack? Or is this more for kind of newsrooms? Because with this crowd intelligence, I guess it doesn't need a newsroom.

Ciarán Murray
No, but there will be collaboration tools. A lot of journalism requires, you know, multi jurisdictional or you know, across the globe collaboration and people. Because you're getting judged by the crowd subsequent to publication, you may want other friendly eyes on it before you publish. So you improve your article score and hence your payout. So there will be kind of more looser newsrooms that will come together on a periodical basis are, you know, it's more fluid basically rather than direct hierarchy and ownership. But in terms of answer your question, it could be for anyone, even Substack riders who may want to just get a stamp of approval from Olas that says this went through our quality control reviews. So people have a higher level of trust in it. But we do envision that it will probably look more like a newspaper.

Like we're going to have systems for photographers to attach NFTs to their work and let people bid on them to be put on the front page or whatever, or any page. It'll look very much more like a newspaper, but more modular. So you can build your New York Times or you can get an AI to do it for you every day you'll just feed the AI some prompts. Anytime this journalist writes an article, I want it in my daily publication, any type. There's an article about Arsenal Football Club, I want to see it. Well, it might be that broad, otherwise you'd have thousands a day. But you get me and you can basically build a publication in your image.

But then there'll be also human creators that will basically be the newspapers of the future, much leaner, that will actually sift through our gigantic database of trusted articles and build a publication either on a daily or weekly or monthly basis themselves. And they will take a cut because they'll handle a lot of distribution. So they'll come to arrangements with journalists. But I do suspect the power balance between curators and journalists will be much more slanted towards the journalists, whereas these days they're much more slanted towards the media owners. It's the Michael Bloombergs that make all the money these days. Where I expect in our system, journalists will make a much bigger cut of revenues.

Isabelle Castro
I mean a lot of people, they now get their news from social media and I guess In a way that I've done a couple of pieces on it, but it's kind of like a collective intelligence in itself. You've got a lot of posts about how trust in the media is kind of broken, but you don't think that we're past the point of no return? It sounds like.

Ciarán Murray
No, I don't. Look, social media has played a huge role in this kind of race to the bottom for the click under the advertising model. But if you take that model away and have a new model, it does free you up from that kind of sensationalist are just sometimes just like completely falsifying information for getting clicks. But don't get me wrong, you're still going to have attention warfare on the Internet. But I think social media itself is reforming because people know Community notes is a first step and I know a lot of the centralized protocols lens and all these are putting a lot of thought into decentralized crowd control as well or implementing AI. So I think as a whole the information industry is beginning to think about truth and accuracy a lot more.

At the end of the day, although we demand sensationalized information, we're left unfulfilled by it. And that's why our trust levels of media are so low. And people don't want to be stupid either. And so I think if you have these, if you change the culture where you have this little tick that says this went through these review protocols, it'll just become a thing that you won't actually trust information unless it has gone through something, if not us, like somebody else's. I think ours is by far the most advanced of things out there. So you won't need this type of review for all types of information.

If you're writing a bit of celeb gossip, I don't think you need to go through Olas or if you're making a post on Twitter X, you know, you probably don't need to have this massive crowd market on it. But for stuff, if you're writing an article, let's say the equivalent that goes through Bloomberg today, that might move a market trillions of dollars. People are really going to want to make sure the journalist is staking their rep and money on that and all sorts of other really important information. But yeah, no, I don't think we're past the point of no return. I think the fact that it's so discussed, everyone's aware it's a problem, they just haven't really figured out a way of solving it yet. And I guess the solution hasn't been there.

The technologies to make something like this possible are really only coming out in the last year or two. Like World ID is what, two years old? And any type of system you build on this needs to be bot resistant, needs to be cibil attack resistant. Blockchains haven't been cheap enough for the payments aspect. Like a lot of things are literally just starting to come together on like literally in the last years. So that gives me hope that people building on this, these technologies can really start moving the needle.

“Media has, since the dawn of the printing press, been controlled by somebody […] By having the first ever ownerless media platform, it removes that conflict of interest and reduces the possibility of groupthink.”

Isabelle Castro
I'm glad that you mentioned tech because that's what I wanted to ask you about. What combination of tech has allowed you to build this now? Because blockchain has been around for a little while. I've heard of a couple of decentralized media organizations trying to set up. What is the combination of tech that has come to fruition that has allowed you to build this now?

Ciarán Murray
Okay, so yes, blockchains. But more important blockchains are in a while now like they're no longer a new tech, you know, but they haven't scale. So I guess L2 developments. We're building on stark net and we can do transactions for like nothing. They're basically free. It's 0.00003, something like that in transaction fees to send money. So micropayments are now possible. They weren't before on any tech stack.

Proof of human is absolutely essential. And that is brand new. And not only is it brand new, it hasn't been plugged into a lot of systems. So we've got a bridge to starknet from World ID, which is huge crowd wisdom tech if I'd call it that. Or protocols. These really cool designs like Bayesian truth serum, which allows us to settle markets on subjective information or unverifiable events. That was just an academic interest, very niche one that we have kind of plucked out of obscurity and we're going to put into production and a live app. You just wouldn't be able to settle things like opinion pieces without something like that. So that's another piece of the jigsaw.

What else have we got? We need decentralized compute. We've got quite. Because that Bayesian truth serums are in the markets that's really computationally heavy. It's not just person a bet, $60 and the market settled here and they get paid out. It's a big long. It's like a little bit like an AI LLM code in the background that we have to compute that and we have to do it trustlessly. So we've got decentralized compute platforms like the Internet computer ICP protocol. We're going to implement that.

And then of course you need the articles themselves to be uncensorable. So we've got IPFs for that. I guess that's probably 90% of our tech stack there.

Isabelle Castro
Okay, cool. Yeah, I'm glad that you kind of brought up the opinion pieces because I mean with AI coming into the newsroom increasingly like the amount of times I have been asked are you scared about your job by people because of AI and because of its ability to kind of like collate news and all that kind of stuff. But saying that I do think there will be a shift for the newsroom for journalists towards kind of more investigative stuff, towards more kind of, not opinion, because obviously journalism is always based on fact but you know, you can see it, a lot of people's political leanings or personal opinion have been coming into even traditional media.

So the fact that you are targeting that is quite a big deal actually.

Ciarán Murray
Yeah. I think if I was in media right now, it would be the support people who support journalists I'd be worried about from an AI perspective. AI becomes this assistance to everyone but it can't go and do the work of journalism. It just can't. It can't get in a room when there's a press conference. It can't, you know, interview people on the streets. It can't go around door to door whatever they need to do where every journalist, that's all in the physical world or I suppose a can do opinion, but not if that opinion is based on again information a journalist has gotten from the real world. So that's just the realities of it. We will need human information always.

And even if we didn't, I don't think it's a good idea that we should just give up the truth to AI if we're worried about it being an existential threat, we really worry about it if it was our newspapers. So we just need to keep it, use it as a tool that helps us, but not as one that completely dominates the media landscape.

Isabelle Castro
Yeah, I Think one of the conclusions that I came because I. Alexandra reached out to me for this interview based on a newsletter that I did a couple of weeks ago on Truth. And I kind of concluded it with how critical thinking is going to be more and more important, especially with AI coming in. It sounds like mechanisms for critical thinking may be being baked into what you're developing?

Ciarán Murray
Yeah, if you think about the way that kind of the Internet gave everyone a printing press, but it didn't give everyone powers of editorial control of review. So we got this kind of democratization of content creation, but we didn't have the democratization of quality control. So you have this sea of bullshit, basically, that we have to wade through from the individual who has their lives to live. It's just too taxing to figure out what's true and what's not. And I think AI is supercharging that again, because it's just made content creation so trivially easy.

So what we're trying to do is democratize quality control and make it trivially easy as well. And then we get to open up access to information creation, which is a good thing if we can also open up access to quality control so we have more and better information from broader sources.

We don't want the scenario we have today where you have like, media moguls controlling, what goes on the platforms where 90% of the people get their news from. That's a serious risk to society. Those people get in a room and decide what's what, essentially. They do, I'm sure. So we can blow that all open without having the system devolve into social media, which is, you know, a cesspit of bullshit, basically.

Isabelle Castro
Yeah, no, I. Yeah, I agree. And I think it's getting worse as well. Like, sometimes I scroll. Like, I've really had to curate my feed because otherwise you just end up descending into a cesspit like you say. Okay, my final question is kind of looking forward, what are you most excited about in the coming year both for Olas, but in general for kind of media reporting within this new kind of possible landscape?

Ciarán Murray
Okay, well, for us, we are in about, I think a month, touch wood, going to launch competitions based on our beige and truth serum opinion markets. So we won't have our own publication platform as yet, but we will be able to kind of enable people to pass judgment on tweets or posts on any platform. We'll try get in as first responder and say, have your say on this and be able to hone our mechanisms with all the data we get from that. So that's one thing we're really excited about. And then that will feed into our full version one launch, which will have a full publication platform for journalists with all these review protocols tied onto it, I suspect in mid-2026, maybe a little bit later.

So that's from an oldest perspective in terms of excited about journalism in this new regime, so to speak. I honestly don't really. Anyone else is doing this. They're doing decentralized substacks, decentralized mediums, decentralized Twitters, which are more straightforward to do because you can reinvent the wheel there. That will look like Twitter. They look like they'll just be under centralized infrastructure. Not very hard to implement. That's why people are doing them. To reimplement the news in a decentralized fashion and protect it from being captured by organized interests is a gargantuan task. And I haven't seen anyone else take it on outside of what we're doing. I don't know if I am excited by much else. Oh, sorry, there's one Project Control X. They're kind of. She comes from a more a journalist background where I'm someone with an interest in media comes from a web3 background. I've just gone full web3 on it. Where she is going to rely still on licensing and copyright and stuff while using Web3 for payments, which I follow that with interest. She's trying to move the needle. We might collaborate, she might use our quality control mechanisms as well. So that's something to keep an eye on. But outside of us, I don't really think there's much going on.

Isabelle Castro
Yeah, no, as I said, I'm interested in decentralised media. I think it could be impactful. I really want it to happen. But you're right I do listen to people talk about it and I feel like this is the first time I've thought that this, this will actually happen at some point soon. So just to be clear, you haven't launched the full platform yet. That's going to be next year. Can people use the protocol yet?

Ciarán Murray
The first kind of time people will be able to play with this will be this summer. We’ll announce it and let people play with it. Then I'm going to run ongoing Competitions. And then we'll probably do another competition for people developing AI agents as fact checkers after that. Gonna do little piecemeal launches and then stitch it all together for full version one next year. It's online, so you just have to sign up through a WorldcoinID.

Isabelle Castro
Okay, so it's not like a hackathon kind of thing.

Ciarán Murray
No, it'll be ongoing. There'll be a market set up, like two or three markets every day on. On a certain opinion. And if you feel like you have insight into whether this opinion is good or not, get involved. If you don't, it's probably best to sit it out because you just damage your score. And then at the end of every month, there will be payouts for the top performers of reward tokens, and then those reward tokens will be transferable for real Olas tokens when we launch. So it's a way of getting a skin in the game or getting your hands on some Olas, basically.

Isabelle Castro
Okay, I will keep an eye on that. But thank you, this has been really interesting and yeah, have a good rest of your week.

Ciarán Murray
Cheers. See you. Bye. Bye.