Chris Messina goes wide to explain where this next 20-year cycle of the internet is taking us. From the community-organzing power of the hashtag to decentralization and the massive shifts ignited by AI, he threads the needle on it all.
n 2007, the hashtag was a simple, yet revolutionary, idea that changed the way we organize and amplify content. Today, it is either endangered or more useful than ever, depending on whom you talk to. On the open social web, hashtags are an important unifying mechanism — not just for content but for people too.
Why is that? How did we get here? What’s next for this small but mighty feature and for the web at large? Here to tell us is Chris Messina, the inventor of the hashtag, the creator of the DiSo Project, and the No. 1 hunter on Product Hunt. In this episode, Messina goes wide to explain where this next 20-year cycle of the internet is taking us. From the community-pulling power of the hashtag to decentralization and the massive shifts ignited by AI, he threads the needle on it all.
Mentioned in this episode and/or acronyms for clarity:
🔎 Learn more about Chris at his website, ChrisMessina.me, or find him on Bluesky @chrismessina.me, Mastodon @chrismessina@mastodon.xyz, and Threads @chris.
✚ You can connect with Mike McCue all across the social web, including on Bluesky @mmccue.bsky.social, Mastodon @mike@flipboard.social and Threads @mmccue.
🌊 Catch the wave! Surf the social web and create your own custom feeds at surf.social, a new beta from the people at Flipboard. https://about.surf.social/
This transcript was generated by AI, which may affect its accuracy. As such, we apologize for any errors in the transcript or confusion in the dialogue.
It’s easy to take the concept of a social media hashtag for granted but it wasn’t always so. Back in 2007, the hashtag was a simple, yet revolutionary, idea that changed the way we organize and amplify content.
Today, the hashtag is either endangered or more useful than ever, depending on whom you talk to. On the open social web, the hashtag happens to be thriving as a fundamental unifying tool.
In this episode, you’ll meet someone whose two decades of work and advocacy not only birthed the hashtag, but also helped to define the social web itself.
Welcome to Dot Social, the first podcast to explore the world of decentralized social media. Each episode, host Mike McCue talks to a leader in this movement; someone who sees the fediverse’s tremendous potential and understands that this could be the Internet’s next wave.
Today, Mike’s talking to Chris Messina. Chris is the inventor of the hashtag, and the creator of the DiSo Project. DiSo led to the Activity Streams project, which was a predecessor to ActivityPub. He’s the No. 1 Product Hunter on Product Hunt and serves as a founder coach. He’s also an investor in vertical AI applications at the Ride Home AI Fund, which he helms with Brian McCullough, the host of the Techmeme Ride Home podcast.
We hope you enjoy this conversation.
Mike McCue:
Chris Messina, welcome to Dot Social. It is great to have you here.
Chris Messina:
I'm so excited to be here.
Mike McCue:
You created the hashtag. You were the first person to use that on Twitter. That's right. And December 17, I think it was Elon posted, stop using hashtags. They I think he said they look ugly or, well, you know what was going through your mind when you saw that?
Chris Messina:
You know it's, it's funny, right? Because the hashtag concept started back in 2007 right? So they've been around, they're like, it's part of, like, the media environment. And in in December, when, you know, Elon posted about it, a couple things happened, you know, first, lots of people told me about it, and they're like, oh, it's gonna, you know, that's it. The hashtags over and, you know, people have been declaring the death of the hashtag for years. And I just kind of accept that, you know, fine. Like, if you don't like it, don't use it. You know, that's not really my problem. But it wasn't the first time either, that Elon has come out against hashtags. The first time was actually in March of 2023 and I believe what he said then was, I hate hashtags. Hashtags missed the mark, whatever that that might mean. And so, you know, like, I guess if I were to take the, I don't know the like Elon's case, I can make it very clearly. You know, hashtags are ugly. They were never meant to be pretty. They weren't meant to be something that was like this elegant interface for tagging and labeling content. It was meant, you know, when many of us were working on decentralizing the social web back in the 2005 to 2008 era. You know, I was trying to create a solution to tag content across all social media, across any wherever you could type text the hashtag Could, could work, essentially. And so we weren't solving for, you know, adding a special interface that only exists on one social platform, or creating groups that only live on one server. It was like, how do you embed metadata that can't be stripped off of a post as that post moves around the social web and, you know, loses its original context, and so by embedding the metadata with the meat of your content, that meant that they were inseparable, and that was, that was a goal, that was a design decision. So one for Elon to say that they missed the marks, one kind of implies that he has no idea, like, what we were actually solving for. Because as an engineer, or one who proclaims to be an engineer, you would think that he would think about, you know, first principles, and that's how we arrived, you know, at that solution, and second to say that they missed the mark. I mean, fine, you know, you can do lots of other or solve the problem that hashtags attempts, attempts to solve in a narrow domain or in a single social platform, as people have done, you know, like Tumblr has communities, other platforms have groups. You can have a little text box to add all the tags that you want, and they're not visible and they won't be seen by anybody, but that inhibits their ability to teach other people how to use tags for, let's say, trending topics that have never been described before. #Black Lives Matter or #MeToo are examples of the type of generative outcomes that hashtags enable. So the question is, how would those groups have found and organized themselves across all social media had the hashtag not been a tool that they could have used? And, you know, up until now, I've not heard sort of a sufficient answer to that question.
Mike McCue:
Yeah. You know the thing that I think people. People don't necessarily appreciate with a hashtag is how, when you have a truly unique hashtag, how that does organize communities across different platforms. One of my favorites is hashtag. Believe in film. That hashtag is used on Twitter, Mastodon, Bluesky, YouTube. I see it everywhere. It is amazing that that, that that community of people who love film photography, right? It is a very it is it? It symbolizes a community and allows people to engage across these platforms.
Chris Messina:
And that was the whole point. That was the purpose. Like you should be able to use whatever tool or medium or platform or community space where your community is active. And if you choose to set up shop on YouTube, for example, but you've got a bunch of people who would love to see your content, all you have to do is tag that content and then it, it travels out there, you know, it's, it's interesting to sort of think about, I oftentimes think about, like, how, you know, we use email as the metaphor for the decentralized social web. There's lots of different servers. You can send an email from one server to another, but it's a private context, and the purpose of the hashtag is to allow people to tag public content and then to share that public content to the world, and then define the people that are interested in it because they've subscribed to the tags. And so that's kind of what you're describing, like this idea of, you know, believing in film is something that can exist on all the different media platforms, and as a result, all you have to do is add that extra metadata and people can find and discover it, even though you don't know who you're sending the content to.
Mike McCue:
And that's the beauty of it, exactly. And, you know, Pixelfed has really had a fantastic kind of, you know, growth spurt recently, and it's but I love it. I personally really love Pixelfed. I've gotten back into sharing photos again. It's a lot like what Instagram was in your early days of Instagram. It's a much more pure photo sharing experience.
Chris Messina:
Actually. This is, this is you'll appreciate this too. I think we just hit the 10. Let's see 20, 1050, the 15th anniversary of the hashtag on Instagram. Oh, so Wow. Well, there you go this past weekend.
Mike McCue:
Happy, happy Instagram hashtag day!
Chris Messina:
Well, and the reason why I bring that up is because, you know, Cory Doctorow talks about something that he calls adversarial interoperability, and it's essentially where, by creating interoperability with an existing platform, you can expand, you know, the usage or the way in which a system can be adopted to new purposes. And it was Instagram, before Facebook acquired it, that used the hashtag as a tool for tagging and labeling the content which were photos that people were posting, because this is before computer vision, and so it was a necessary means to add some metadata that people understood. And so they adopted something that already existed on Twitter. And it wasn't until that happened that Twitter Inc was like, oh my god, like we could have trademarked this, or patented or whatever they thought they were going to do, which, of course, they could not have. And they tried to prevent, like, the social web from having it. And it's just one of those examples where you know when you understand the history of these things and why people do certain things and why, for example, Elon, would be against it. It's because he doesn't want to lose control. And I think it's very important for those of us who are building the social web, who care about the open social web, that these tools and technologies and behavioral technologies need to remain as part of the commons.
Mike McCue:
Yeah, this, you know, this concept of this kind of adversarial kind of collab, you know, collaboration, or coordination, the ability for a community to exist across platforms is crucial. One of the other great examples I really love is the NBA Threads hashtag. Yes, even though Threads doesn't that was one of
Chris Messina:
the first on Instagram. It was it, yeah, in fact, I think it was like the first, or at least the first that they promoted, because it was another one of those great examples where, you know, it was a bunch of like, geeks and nerds like us who were using hashtags, you know, to go to SXSW and talk about the events that we were happening like that were happening in real life, but it was NBA Threads that and, well, it wasn't NBA Threads back then, but it was NBA, probably on Instagram, that was bringing together the basketball ecosystem on social media, because you still had to teach people what social media was for, and if you could find out what the players were doing and get this intimate peek into their lives, suddenly people understood, oh, that's what social media is. It's like this insider view. It's no longer like going through the maybe it's going through gatekeepers, but not in the same way.
Mike McCue:
The NBA Threads phenomenon, which grew on Threads as a pretty significant hashtag community, which then a bunch of people went to Bluesky fragmenting that community, right? But the hashtag is still used on Bluesky, even though the name of the hashtag actually is the platform that it originated on, which is kind of wild, right? But there's still lots of posting. It's so it's such a great example, right? People are posting on Bluesky, they're posting on Mastodon, and they're posting on Threads with NBA Threads, and you can see it all in one place, right? Of course, you need apps and clients to, like, bring it all together, right? Thankfully, we have Surf now for such a thing. That is one of the things we're doing with Surf. But like that concept of, like, a community, basically being able to sort of migrate and be and exist across different platforms, or coexist or coexist Exactly. It's not just about moving from one to the other, right? I just think that's, you know, an under, radically underappreciated element of the hashtag. And you know, for, you know, a long time, and I will confess to being when I was on the Twitter board, I was relatively uninformed about the hashtag. I kind of thought it was sort of an ugly, kind of techie thing, and I kind of thought it was one side. Well, I mean, I was more on this sort of side of, like normal users. From my vantage point, was like, you know, my mom doesn't understand hashtags and things like that, right? But it wasn't about my mom, really, right? It's about a whole new generation of users. And I remember Twitter actually wanted to do a and they did do an ad campaign around the hashtag, right? And they have all these hashtags on billboards and things. And I was like, Guys, I remember, why are we doubling down on this, like, techie, geeky thing, right? I was, I was completely, frankly, unenlightened about what the power of the hashtag, right? And when I listened to you talk about it, you know, I now, I've now grown to appreciate just how big of an idea this is, right, that it's totally cross platform. It's embedded in the post, and it is not something that the app, the app or the platform maker has to actually support, because it's just an organically generated thing, right? I mean, that just that alone is brilliant.
Chris Messina:
It's also, and this is, this is an important part, which is that learning to use a hashtag like you said, like, you know, maybe your mom will never get it, but she'll see these weird words, you know, floating around that tend to show up, maybe around content that she likes or engages with, and then maybe someday she does type a hashtag. Maybe she doesn't, but it's the use of the tag is embedded in the design of the tag itself. By, you know, it's like, you can, you can type $1 sign in front of a number, and suddenly, now you're talking about currency. And that's, that's how this idea is, is transparent, and so once you learn how to use it, then you can use it everywhere, yeah?
Mike McCue:
Well, that, yeah, that is, that is awesome. And, you know, I can see also where, you know, a lot of people put hashtags, you know, 1000 hashtags on there,
Chris Messina:
right? Abuse is real. 101st to admit that exactly it is real. And there,
Mike McCue:
there are ways for a client, a client app, to deal with that, right?
Chris Messina:
And sort of, you know, Mastodon has actually done some things where they'll show, I think, the first three tags, and then they'll kind of hide behind some toggle the rest of them. And it's a fairly I came around to accepting, okay, like, that's not a bad solution when people are doing keyword stuffing or hashtag stuffing, because obviously the real dilemma is whether the hashtag is being used to communicate with other humans and the signal to them that this is a topic of conversation, or that I'm trying to say something about a trending topic that I want to drive people to, or whether I'm simply adding a bunch of filler words to signal to the algorithm that my stuff should be shown when someone does a search for again, like dogs, right? And that, to me, is sort of a misuse of the tags to some degree, because now you're adding metadata for the machines, and the machines are going to be much better actually, at figuring out that your photo is of a dog you know, and of a specific breed of dog, even then adding some, you know, generic hashtag that's going to be polluted anyways.
Mike McCue:
And I think you know that is, that is one of the things, again, is an underappreciated aspect of the hashtag, is that it is not about topically identifying a piece of topic, as much as it is attaching it to a community. Sure, that's when it's like at its most powerful use, right, like believe in film or NBA Threads. And I'm curious, what was the first hashtag that you typed into Twitter?
Chris Messina:
The first hashtag was hashtag Barcamp, and that was for an event that is now as of 2025, 20 years old. Barcamp was an event that we organized in Palo Alto. We organized it in six days. It was meant to be kind of the open source fork of Tim O'Reilly's food camp, which he would hold up in Sebastopol at O'Reilly headquarters. I was a an early, let's say, or a recent transplant, to San Francisco. I had arrived in 2004 Or and, you know, we were just kind of getting the community going, and we wanted a space for us to come together and to talk to other people who are just part of this early web to o ecosystem that we're using these technologies to meet and connect with our friends and and it was kind of like the first big moment where the social technology movement kind of like put itself on the map. We had Michael collori from Wired was there, and he covered the event. And so that kind of inspired this movement where other people around the world saw what we were doing, you know, in Silicon Valley and in San Francisco, and they wanted to get involved. And it was, you know, we used Twitter to, you know, get the word out, but we also were using IRC, and so we were able to use hashtags as a way of seeing the growth of this ecosystem emerge all over the world. So from the very beginning, the hashtag, especially for me, was a lot more about bringing people together in the real world who are otherwise sort of mycelially connected through the internet, but maybe just didn't know it that.
Mike McCue:
That is really cool. So after you did that, yeah, um, what? What was kind of the next level of adoption of the concept of the hashtag? What happened after that?
Chris Messina:
The hashtag was, was inspired in many ways by the tagging system that Flickr had adopted. And of course, Flickr was kind of that era's Instagram. It's where people would upload their photos after and during events. Actually, I still have an account I love. I still do too, yeah, and there's a lot of like, historical stuff there that's like worth perusing, but, you know, so I wrote a 2000 word blog post, kind of describing how I thought hashtags could work. Some short time after that, we had our Barcamp event and then, and that was because we had South by Southwest earlier that year. And so a bunch of people started using Twitter, realizing that it was this great platform for sharing and posting updates, sort of like mini blog posts. Micro blogging is, I guess, what we called it. And so in the early days, Twitter wanted to be a protocol, you know, they built support for XMPP. They built support for a web API, and I started to kind of nudge the developers of Twitter apps to add support for the hashtag. Because I brought my proposal to Biz Stone, you know, one of the co founders of Twitter, and proposed the idea to him. And, you know, like you said, this is for nerds. Like, this is dumb. Like, Why would, why would we do this? We don't need this, you know, we're gonna solve it with algorithms.
Mike McCue:
Yeah, that, that is sort of the initial assumption that a lot of technology people make, right, which iron, which, you know, you can't do. You can't do community things. Like #BelieveinFilm.
Chris Messina:
Yeah, like, the illegibility of algorithms that inhibits actual community formation. It's, it's the act of creating content where you add the tag that is a signifier that says, I want people to come around, you know, the space. It's, you know, going to a park and like, stepping up on a park bench and sort of, like, saying, Hey, I have something to say. Or, I don't know. Like, what are the signifiers that allow other people to know? Oh, this is a party, as opposed to this is like, some random crazy person, right? So algorithms are not going to solve that problem for you. And so that's why I was like, Okay, look like you're building Twitter. You sold your company to Google. You're obviously smarter than I am, but I have this need for these Barcamp events that I'm running and CO organizing. And so we need a solution now, and so I'm just going to start using it and gradually talk to and convince all these developers who are building on the Twitter API to add support for the hashtag. And that's really how it got started. So I could be clear, adding support for the hashtag that means so basic, making it a hyperlink, right? That's basically it. It's just like, turn it into, it's just like, you know, we had at mentions. At mentions were already there. The funny, you know, one sort of funny historical thing, is that Foursquare rip used to be able to allow you to check into a location using a text message. And you would check into location using an at symbol, and you'd say, like, you know, at Blue Bottle, and then it would check you into that space. Well, eventually we took those at like locations and then turn them into at mentions. And so the same code, minus one change in the regex that allowed you to mention someone was the same thing that would allow you to then create a search for hashtags. Now, in the beginning, Twitter didn't even have search. I mean, it took, I don't know, like, a decade for Twitter to get search. And so there was a third party company called summize, which built Twitter search, and then they added support for hashtags. And so it was amazing, because any tweets that had hashtags in them, you could click on it and it would like produce a search. You're just like, well, of course, duh. And yet, Twitter didn't build that. And so Twitter acquired some eyes, and then when they built it in, it was sort of like this trojan horse movement moment where the semi search results became Twitter search results, which then validated the hashtag as a useful way to get content into the system and to have people discover and find related tags.
Mike McCue:
Yeah, yeah. And Mastodon actually enabled you to follow a hashtag.
Chris Messina:
Which was part of my original spec.
Mike McCue:
Interesting we should do you have is that spec online somewhere. We show that in the show notes.
Chris Messina:
It's if you go to bitly.com/tagchannels, that will take you there.
Mike McCue:
What is next for the hashtag? Like when you think about, okay, well, what else? What would be a better way to support the hashtag. What are some other things we could be doing around that?
Chris Messina:
So here's a sort of a grab bag set of things, you know. One is back in the original spec. The idea, and part of this is about both encouraging people to use more hashtags, and then maybe even, like, suggesting, you know, like two to three hashtags that might be relevant or that are anticipated. Let's say by an LLM right. So let's say, you know, I post a lot of stuff about, let's say Spotify app updates personally. You can imagine that for every time I'm posting about Spotify, there's an autocomplete that suggests, you know, hashtag new Spotify, because that's the hashtag I use, so I don't have to type it myself. And that way, instead of, like, forgetting it, or if on on the go and on my mobile device, it's kind of like taken care of. So lowering the friction to adding hashtags is great. Yeah, the second part of that is then making it easier for people to either follow or mute certain hashtags from certain people. So if you don't care about my Spotify updates, then why don't you be able to, like, you know, whether you tell Mastodon or you tell Threads or whatever server you're on, look, just mute Chris whenever he mentions this hashtag I care about, like, his other stuff, but not that that would be great. And then from a labeling perspective, I think that's where what Bluesky is doing with the intermediaries and composable moderation could be another way to create, perhaps, like, long lists of essentially the inverse of block lists, where you want to follow a community. And let's say there's like a half a dozen hashtags that just with that hashtag, exactly right? So maybe, maybe there's some variants of, you know, believe in film, or some adjacent tags, and you have a little collection of those hashtags that are relevant for that community, that type of thing could also be very useful.
Mike McCue:
Yeah, for example, you could have, like, a curated list of photographers who you just absolutely love, they use plus tags, right? Plus tags, right? It's the handing of that that's right. Yeah, no, totally. I mean, this is the kind of stuff that I think is incredibly powerful, especially when you make a cross platform, right? When you say, okay, when they post on Pixelfed or Threads, or Instagram or YouTube or Mastodon or Bluesky, I still see it, right? Yeah. But of course, that also gets at another thing, which is, like, identity, right? Like, which is something you've been working on for a long time as well, right, correct. You've done a lot in the world of social web. I think you've, you've kind of, in some ways, already mapped out the future, 10 years ago or more, you know, 20 years ago, even, in a lot of ways, right, with what you did on flock, what happened, you know, where's the hashtag stuff, OAuth, you know, identity, you know, activity streams. There's a lot of things that you worked on here that have kind of been a lot of the forerunner for what we're currently now seeing, right? The whole rise of new protocols and new apps and a whole new, whole new kind of breath of fresh air for the concept of the open social web. Hashtags are already inherently cross platform, right? But identity isn't, you know, right now, for example, if you do a search on me, you'll see I will come up with, like, seven different accounts on the social web, right, Pixelfed acid on Bluesky Threads. I mean, Flipboard goes on and on, right? So what's your take on that front?
Chris Messina:
You know, it's, there's this interesting dynamic and mix between, I think, what people want or would prefer to have, especially when it comes to usability, familiarity, accessibility, and then the way that the Internet and Internet technologies actually work. And then there's a third part, which are the downsides or risks that come with certain implementations or approaches. And, you know, I think earlier on when we were trying to build internet scale identity, especially with OpenID, and we believed that a website address, a URL, would be the ideal way to sort of identify a person, one because it's public, so anybody could go to your domain, you know, chrismessina.me from that domain which I own, you know, I paid for. You could then find all my other accounts, and that would be a very useful way, one again, to have multiple tools or mediums or platforms that I might personally prefer to use in a way that I then, you know, maintain control over one the publication of those things. Like, we had something called realm me, where you could, like, point to another account and then point back to the original domain to say, Yes, this is the same, yeah. You know, very simple. What we have found. And I think this again, is it's both like, what's. A contextual but it's a consideration is that, one, a cost of a domain can be prohibitive for many people in many countries. Two, technically setting it up, you know, just like, just like crypto, introduces a level of complexity that, at least in this era and generation, when you want to have widespread adoption, creates a kind of risk or Surface area that can lead to exploitation. In other words, I mean, crypto is the worst, because the level of risk gets into like, dollars and cents, whereas, like, if I figure out your domain or I hack your domain, maybe I can. I mean, there's lots of bad things that could happen but in so much as with a domain, there's a lot of, you know, services and providers that make it possible for you to get and register and own a domain. I mean, this is like, kind of where Bluesky is going with their business, you know, that allows you to create a profile that is then portable. And so if Bluesky, you know, goes foul and turns into, I guess, dark sky is a different thing, but like, you know, right, right, right. Then you can still own your domain, and from your domain, you can point to another AT Proto provider, and all the people that were following you are none the wiser. So you get choice the people who are following you continue to be able to get your updates, and there's a persistence in that. Now that's that's great again, the ease of use and the adoption become challenging. And I guess like to your point about having multiple, you know, mics showing up when I do a search for you, you know, Mike and Flipboard. There are other challenges where people will use, let's say, a business or company, you know, as their domain or their identifier. You know, like I used to be messina@google.com Well, I presume no one has my email address there anymore, but if my identity were tied to my employer, and then I leave, now, I've actually lost access to that. And so just like, you know, you are, you know, my Flipboard like, and presuming you are probably going to say that for quite a while, you have also had employees, you know, that started out, let's say Steve at Flipboard. I don't know what that is, but then he leaves, and he no longer has access to that that that profile or that identity, and so you will have this other problem of recycling identifiers, or the ability for someone else to squat on your identity, like there are many people, for example, that have deleted their ex accounts or their ex profiles because they want to protest against our friend the hashtag hater. And what that means, though, is that someone else can come and squat on your previous profile or name in a way that then exposes you to all sorts of other types of risk. So identity becomes much harder when you get into the digital space because of these issues that when you're just thinking like, oh, I want a profile on the internet and I want my friends to be want my friends to be able to find me, you're not thinking necessarily like two to five to seven years down the road, when things can go sideways. The TLDR of your question and answer is that when it's complex. Two, it probably needs to be somewhat complex. And three, it may be that there's another generation that comes up who, like, grows up with the internet and is willing to accept some level of that complexity in order to defend against the type of corruption or concentration of power that we're now seeing on a number of platforms you know that exist out there, but I just don't see how you enable freedom and a great deal of simplicity, unless you have a totalitarian system, you know.
Mike McCue:
So this is where decentralization necessarily means that it's going to be more complex than if it was all perfectly centralized.
Chris Messina:
Yeah. And I guess, like, it's important to identify and call out, like, why decentralization, one is necessary, and two, why it needs to be somewhat more complex. I mean, we, like you could imagine, and I think the United States has done a pretty good job of attempting to federate or decentralize some of the powers that in other countries are centralized into, you know, one high level government system. We instead push a lot of control and power and influence down into the states, and that allows you to choose from 50 to 51 maybe soon 52 I don't know different choices, where you could go live, where the rules are different, right? Just like having different Mastodon servers and where the choice of where you live and where you digitally homestead actually, like, has a great deal of import on your lived experience, and so those decisions are becoming more salient as having a digital presence on the Internet becomes more commonplace. But 20 years ago, when we were working on these things, we anticipated these issues, but we were unable to tell a story that captured people's imagination such that they would care and then actually get involved in using those solutions that we built. And so it feels like we're kind of at this crossroads where a lot of things that we were worried about are now coming to pass, and the question is, will we be able to take back control over the Open Social Web before, you know, the powers that be, you know, really compressed. Own, the choices that we have available to us.
Mike McCue:
I do think you're right that there's a there's a period in time here where, you know, there's fantastic momentum at the moment, but yeah, we need to accelerate that momentum in a way that, you know, assures that this is going to exist in the long run. And the challenge, of course, is that some of these decentralized, you know, components does make it harder for normal people to kind of adopt this new world.
Chris Messina:
So one of the things that's worth contemplating in this part of the conversation, I think, is what an appreciation for what moderation is and what it does. And, you know, we have some friends that spend a lot of time and a lot of resources, you know, a lot of their their goals, so to speak, on solving that problem and on attempting to keep people safe. And it's a very, very hard, thankless job. It's one of those costs of social media that you kind of just expect. It's sort of like a janitorial service where people who use Mastodon servers, perhaps will donate but the act of dealing with all the terrible things that people will publish and share and creates a burden on those who are offering those services. We are in this interesting moment where LLMs might be able to assist in two factors. One is, if you imagine that people become more comfortable with adopting kind of a decentralized home on the open social web, whatever their digital homestead might be. Well, in choosing that, you are choosing a set of rules that govern participation, you know, like whatever the community vibes are of the space. Maybe it's MAGA and, you know, very much conservative values and against you know, progress in a certain way, or progressivism or liberalism, maybe you are on a server that's incredibly like liberal and wants to be able to talk about things in a safe way, where more conservative viewpoints are either suppressed or aren't promoted. Well, for those people, presuming they're humans that are acting as moderators in either of those contexts, the LLMs and those types of services could actually be a facilitator of two things. One, is obviously identifying content that is borderline questionable or problematic. Two, it could be a way to perhaps intervene or interact with a person who's acting in a way that is outside of the norms of the community. So whether it's a chat bot or something like that, that sort of says, and Twitter had these little buffers and these little roadblocks before these speed bumps and sort of say, Oh, it sounds like you're about to, like, say something really mean to Mike. You know, most people on this community don't say things like that. So you could automate that. And then there's a third part where, and I don't know exactly how this would work, but there was a product recently called Social AI that meta acquired. And, you know, Zuck has talked a lot about creating artificial entities or characters that will exist on social platforms that people interact with. You know, there's an interesting aspect of norm setting that may happen as a result of generative AI being able to produce lots of content that causes people to maybe behave towards one another in a more positive or pro social way. I don't know what that's going to look like, but I think character AI and things of that nature demonstrate something along these lines.
Mike McCue:
You know, this is a really, really key point. The use of LLMs to categorize content, to find signal through all the noise, to flag problematic content. From a moderation point of view, these are, these are very, very powerful techniques, techniques I definitely believe in. I you know, in fact, a lot of that stuff we do on Flipboard, right? There's always someone watching over these things. And you know, when it comes down to certain moderation steps, there's always it's, it's effectively creating a bionic moderator, right? Or a bionic curator, right? And that's that is something that I think is is going to continue to be a very positive development for AI as it relates to social media. Another thing I've been thinking about is reputation. Because one of the things that I've noticed that x does now with their ex AI, is you can go to someone's profile and you can kind of get a summary of like this person, what do they tend to post about? Who you know, etc, right? Are they right? Brock, the yes, the grok thing that's built into X, have you? Have you tried this where you get a summary of someone's profile? Yeah, it's quite good, actually. Yeah, it is. It is, I have to say, it's definitely good. And, and, and. So I wonder if there was a way also to kind of use as a way to have some sort of reputational signal around. You know, for example, if you're a bot, it's much harder to establish it a quality reputational signal, right? If you're a spammer, right? And some of that can be done, you know, without LLMs, you don't necessarily need AI to make those, some of those determinations. You know, you can just look at they're following 1000 people, and they're followed by two. So maybe they're a bot, right?
Chris Messina:
So they're obviously bot farm, but farms that are all following each other, so sometimes those numbers are misleading, exactly.
Mike McCue:
So you it's not foolproof, right? But this is where you start to get into kind of the use of of more advanced kind of AI type of techniques to try to figure out reputation, is that something that you know, you thought about much?
Chris Messina:
As someone who is quite prolific on Product Hunt, I see a lot of these tools that are being created to essentially, you know, create posts from AI in order to build followings and for users. You know, if they see this content and they engage with it, they don't really care if there is a person or reputation behind it. All they care about is the content. I mean, this is what TikTok has shown us. So I would balance what you're saying and the value of what you're saying with what the goal or purpose is. So is there a question like that? I'm trying to get real or original reporting about something, in which case, reputation is important, and that's why, historically, we've used institutions or the fact that someone has worked for a company, you know, and they have, in theory, you know, requirements or interview tactics to evaluate someone's veracity or their value, right? If I said I worked for Google, that might mean something to certain people who know what Google is, versus, you know, a random account that signed up two weeks ago and isn't connected to anything. So in that case, I think we have to maybe embrace more of a plurality where reputation one is contextual. So you and I might have very high reputation with each other because we both worked on the open social web for a long time. Web for a long time. But if I were to show up, you know, in a vintners retreat, you know, and started the opining about how I thought the wine tasted, actually, I think you would probably do fine, but, like, I would just have opinions, right? Yeah, so that I think reputation is performed contextually. So to your point, maybe the way to evaluate that in these in a decentralized sense, is what are the number of people that are, I guess I would call this contributory reputation. Who are the people that interact with Chris? What servers are they on? What do they tend to talk about? And you know who, I don't know if I want to say like, who tends to be right, but what is the level of authority that he demonstrates over time in a certain domain or area, and then that could provide some context within an individual post to know, okay, like, the reason why this is in my feed, maybe I don't even follow him, is because these other people that are reputable in their domains interact with him.
Mike McCue:
Yeah, that is pretty cool. I think that there's a sort of reputation by, you know, association, right?
Chris Messina:
Yeah, that's not my proxy universe of guilt by association, right?
Mike McCue:
Reputation by association, exactly, yeah. And by the way, there is the other two, which is also helpful from a moderation point of view, right? Yeah, yeah.
Chris Messina:
These all, well, it's, you know, the harder patterns are a way of, kind of, like capturing and productizing and distilling, you know, reputation and so when you join, you know, whether it's Bluesky or massive on North Threads, you know, someone has curated and collected a set of people that they believe has a set of reputation or value, where the content that they produce is worth it for you to see.
Mike McCue:
Yeah, I very much believe in that approach. The, you know, the algorithms that generate feeds, you know, are going to magnify whatever the choices are that people, whatever, whoever put the inputs in right? And so I think it's critically important what inputs you put into these algorithms for like, a for you feed, for example, right? And, and so that is the Starter Pack concept. Is people curating sources, not individual posts, right, right? And that, that is a lesson we've learned in Flipboard, right? We have, you know, millions of users around the world and lots of different languages and lots of different regions. We have a very small team, and, you know, like six or seven people who basically moderate the entire world and all of our topics and all of these things on Flipboard. And the main way we do that is by curating what goes into the algorithms. And it is it, is it, you know, it's a select group of people and sources that we put in that we then have verified that are worthy of being magnified by these algorithms, right? That is something that, that I think is a is a great opportunity with Bluesky Starter Packs lists. You know, if you make a list, that is a really big high signal for me that these are all trustworthy people, right?
Chris Messina:
It sort of feels like there's also a value in teaching people to become better moderators or better curators, you know, like, I believe I did make a feed for Surf. And what would be really great going forward would be a set of tools that allow me to express my taste or my preferences. Piece, which then allows me to build, you know, a feed, or set of feeds, so then I understand that my role in CO producing these feeds has downstream benefits. Now let me give you an example of kind of like what I'm imagining the same behavior is the behavior that I engage in on Spotify when I create playlists in that act. One, I'm doing curation. Two, I'm doing affiliation or association between in this case, it's on the level of tracks. It's not really at the level of the artist. But from that, what Spotify then does, let's say that I have like, 20 tracks on a playlist, they'll start to automatically play other tracks that other users have curated into similar playlists. And that allows you to keep going down this rabbit hole of you know, content that's like, super valuable. And so that is a behavior that makes sense in the media sort of music space. I think it makes sense for us to consider doing similar types of, you know, tooling in the social space.
Mike McCue:
When you go to see a trending hashtag, you know, hashtag, TikTok, bam, right, right? You're going to see all sorts of posts, right? And so the question is, right now, there's like, typically, two different ways, right top and recent to sort that, right? But wouldn't it be awesome if it was like, here's all the people that you're following, and those people are following, and what with they've posted with that hashtag as an example, right?
Chris Messina:
And also like, who are emerging voices, or who are the voices that have yet to be discovered? You know, where the LLM is is able to almost do kind of like a negative space thing, where it's like, what are the opinions that are interesting and off the beaten path, because there's so much regurgitation and so much like, you know, watching DeepSeek, you know, in the past week, you know, you kind of get, like, the zeitgeist very quickly, because people have to publish something, whereas people that are deeper or have unique insights are the ones that you actually want to probably like hear from, so that you get a plurality of views and your understanding happens much faster.
Mike McCue:
No, I do think that this concept of reputation, you know, Reddit has karma points, right? Have you? What do you think about that type of a system?
Chris Messina:
Yeah, you know, again, so, so this is something that we did talk about with Open ID, like, a long time ago, right? So with activity streams, the idea was that you get a receipt for something that you did on some social platform, and that would be syndicated back to whatever your identity provider was, which would be your Open ID. Your Open ID provider would aggregate all of your activities and then syndicate them downstream to whoever might be following you. Now that's one sort of cheap way of getting at a type of reputation, because it's like, what does Chris do? Like, what does he post about? Where does he hang out? And so you get this kind of ambient sense of, you know, what someone's up to. But the reputation piece, like I said, is kind of a set of actions or activities that are performed in a given context, where you behave in a way that is either pro social or beneficial to that context. So acting in a certain way on Reddit means that I'm curating, or I'm uploading, or I'm like, adding comments that are, like, useful or helpful, and then interactions that happen on that platform, identify those activities, which I mentioned before, as either being, you know, positive or negative. Just because I added a bunch of like, you know, asshole comments on Reddit doesn't mean that I should have good reputation, right? So syndicating that reputation means that I have to trust Reddit as a reputation authority or provider, and when I give it an identifier for a person, Reddit come come back and say, Well, this is this person's karma within all these different, you know, communities, and then maybe here's his total, you know, reputation for that space. I absolutely think that that type of portable reputation is valuable and useful, but I do think that there's not, I don't I just, I'm very skeptical of global reputation being, like, useful or able to be measured in a way that, you know, it's sort of like your credit score, you know, like, there might be some mistake that you made and you might have a really bad credit score, but like, there's a reason and there's no context to it, yeah, and it's also a chicken and egg thing, right? How do you get a great reputation, if that's rep, if you're factored out to begin with, unfortunately, now you've built an industry for people that will sell you the like, you know, the reputation thing for a piece in the measurement, right? So that's, again, when you have things you have to be like, very mindful of how those reputations will be used, and whether or not they will create worse behavior.
Mike McCue:
Yeah. You know, coming back to your point about Starter Packs that that that is a sort of, that is a reputational system in a sense, right? Yeah, you know, you have someone that you respect who has selected someone else that they respect and put them in the Starter Pack. That's a big deal.
Chris Messina:
I've seen cases, actually on Bluesky where people, like, created a Starter Pack, let's say for, you know, starter Threads, or, like, starter tech or something, and they'll add a bunch of like, people who are of high reputation, and then they will edit the Starter Pack for a fee to add other people. People who are trying to get followers, thank God, really, and this is what I mean.
Mike McCue:
Where are you spending the bulk of your social media time right now? Which platform are you mostly on?
Chris Messina:
You know, I was, I was the 19th user of Instagram, and so I spent a long time kind of growing with that platform and real user, wow. Yeah, that's cool. It was called bourbon in the beginning. Bu, I remember, oh yeah, it was, it was, it was just like, Foursquare, yeah. So I was there from the very beginning. And so, you know, that's how I got Chris as my handle. I, you know, it's been hacked, but I've kept it the whole time. And, you know, I kind of moved away from Instagram because it just, it got so much more focused on video and on ads and culture, and it was kind of less about the social sharing aspect. So when Threads came out, you know, and I sat on Chris for so long, was like, Oh, okay. Like, this is cool. And for what it's worth, like, Threads is well designed. It's well built. They've been moving really fast. Like, I'm very impressed with what they've done. So I would say, more recently, though, I think Bluesky is very promising. And, man, I feel so conflicted about like Mastodon, but it is. It's a harder platform to use. It really is. And I used to defend and fight for that complexity, as we did this time. Like sometimes the complexity is necessary to decentralize. But I suppose what will be very interesting to see is whether or not Bluesky and like Freer feeds or any of these other initiatives that are meant to create multiple AT Proto you know servers or instances works, and once you have multiple Bluesky or AT Proto servers come online, whether that adds a level of complexity that is the same or less than Mastodon.
Mike McCue:
What is your opinion on the most technically hard part about Mastodon, or what is the hard user experience part of Mastodon?
Chris Messina:
For you, I support both like design goals, and perhaps in the next two to five years, there will become a an alignment between these platforms. But what Bluesky is solving for is more about identity, portability and the ability for you to create kind of a private key where you sign your content, and then if you want to leave, you know, because of enshittification or something else, the whole, well you can the whole dead strategy. I mean, it's, it's um, in so much as one, I use my own domain for Bluesky, right? So I have that portability, and that was something we were trying to solve for before that all is very promising. I don't have that on Threads. I don't have that anywhere else. It's incredible. That's very promising, wholeheartedly. At the same time, Mastodon is a lot more like the states model that I described, where choosing a server could determine the type of social experience you have now, because I'm kind of a gadfly and, you know, man about the internet, you know, like I kind of prefer to publish once, syndicate everywhere, a little more slutty about my content in that way, and so what Mastodon offers me or offers its user base is less valuable because I am trying to publish to the commons, whereas if I wanted to have a smaller server, almost like a group chat server or like Discord server, I would say that Mastodon actually provides more value because my identity is kind of constructed within a certain context. I don't see that those things have to be mutually exclusive, but they are currently. So I think long term, we need to support both one where you can have an identity and reputation that exists within one context or server, and you have the ability to have portable identity. And so I know people are working on some of those things right now, but they're currently not compatible. And it's, it's a choice.
Mike McCue:
It's so interesting because, well, for the record, I personally love Mastodon, and I've loved it since the day. It is, it is amazing. And yes, and I know you also have been a big supporter as well. And it's gotten, by the way, a lot easier to use a lot of the kind of challenging user experience, things that people talked about when they first joined those things have been dealt with. Eugene's done an amazing job at continuing to improve it. That said, I do think that Bluesky, because they're still centralized, they've been able to not have some of those same kinds of clunky issues, like, Oh, you went to this server. How do I follow you? I can't. I'm having trouble following you. Those kinds of issues because they haven't federated. They're not decentralized yet, right? And so ironically, you know, you talk about the portability on Bluesky, and you're right. Architecturally, I do think it's a better strategy for how they've gone, but I've never moved my identity on Bluesky. And currently, whereas I have moved my identity on Mastodon, I was originally on Mastodon dot social, and I moved it to Flipboard dot social. And you know what? It actually worked. It actually worked all my followers, all my following. It all it was, it was actually. To be relatively seamless. Now, that said one thing I learned through all of this whole thing was it does a lot. It's not just about the protocol. It comes a lot down to the app, right? And how you deal with portability, right? And what I love about Bluesky is that you can take, you could write an app right now where you could copy every post I've had, every every you know post I've made, and every single reply and copy and put all package all that up, and then put it on a different server that is technically feasible right now, right? And that is really great. And then the whole did thing, I think, is also great, because you can change your name space, and it doesn't mess with your your namespace is not tied to a server, right? And that also is, I think, the future. I think that's a great model for identity. So it's weird, because I think both Mastodon and Bluesky have advanced the world of portability in different ways, but nobody's totally solved the whole thing, the whole package.
Chris Messina:
Yet it really causes me to wonder, like, if these things are ultimately not reconcilable, either just based on like mathematics or based on product expectations, right? So, for example, like crypto, you know solved, in some ways, the idea of durable identifiers that you yourself own right, not your keys, not your crypto. And while people will bristle at how crypto, you know, was used, you know, nefariously for various schemes, the reality was that, you know, PGP had tried to solve these problems a long time ago, but for whatever reason, the crypto world was able to solve this in a way that gave you control over something that you owned and could sign and could provably demonstrate that it was yours. So that technique, which is adjacent to what Bluesky is doing, seems to have some level of persistence given the sheer amount of dollars that goes through the crypto network. So they solved for a very important part of the problem. However, the act of creating like a Discord server, or creating a Reddit community, or any other type of like space that is an architectural concept, you know, is more like a Mastodon server and instance where maybe you want to have an identity that is only within that space. So the point that we were making before, you know, a global name space for finding and discovering all of Mike's profiles may actually not be what you prefer, because you want to have your little German dungeon identity that no one else knows about that's not connected to all your other identities. So we need to support both.
Mike McCue:
Do you believe that activity pub and AT Proto are going to, you know, meet in the middle at some point. Like, what's your take on, how or you think one's going to win over the other. You've seen these kinds of protocol things in the past. What's your take on, what's going to happen here?
Chris Messina:
You know, in some ways, it feels like, and I've made this analogy, and it I'm always cautious about believing, again, that, like, history repeats versus even rhymes. You know that if activity Pub is kind of like the Betamax to AT Protos VHS, that AT Proto potentially could go a lot further, because they're just dealing with messages and message signing, they don't really and yes, there's a lexicon, but as you know, the more players you have in a space that are trying to decentralize, the harder it is to coordinate and move a format or a protocol format forward to answer your question. What I'm actually most interested in is thinking about the alignment, and this may be a curveball, but the alignment between the semantics of activity pub and what you can do with LMS that when you combine an activity pub feed as a way to signal either intent or a certain type of payload or a certain type of structure of content that an LM can use and make sense of that that now I don't know this is actually kind of like this interesting race condition where LLMs might be able to just look at a message and figure out what it is that's that's going on. Maybe the semantics are not that important, but my intuition is that that pre processing of the type of activity that someone is engaged in, whether I liked your post, or whether I reposted your post, or whether I followed you, or whether I wrote a movie review, those types of things, can actually help improve the training data that goes into these LLMs that might be social in nature. And we've we haven't even seen what that's going to look like yet. So there is a little bit of a race condition where you have generic content marked up with, you know, AT Proto that kind of like, you know, spans and syndicates and gives people updates. Then there is activity pub, which could be very useful for IoT or for other types of applications where the semantics really matter, that will enable all sorts of new things to occur. So that's what I'm mostly interested in, if I think about, like, the future and where it's going, that I haven't seen a lot of experimentation with, but I would love to see more.
Mike McCue:
Yeah, well now, you know, before we go, I'd love to get your sense of you know, you've been around the block multiple times here and now there's a whole new thing happening? What's the advice you give to people that are building products in this new space?
Chris Messina:
Sure, it's a great question. We are coming to the end of the web, two, oh, generation, in many respects, generations typically are 20 year cycles where a certain set of ideas become, in some ways, institutionalized, but can also lead to a type of sclerosis in what's possible. Because those of us who have been through it are like, Oh, we solved this problem before, or like, here's the trade offs. Why would you do that again? The thing that I have to keep reminding myself is that every 20 years, so many other things have also changed, that the constellation of relationships between the things that you thought were static or set fall apart. And I look at what's happening with AI and generative AI and LLMs, and I think about the conversational nature of those experiences in those products. And I think about my time at Uber, where Travis's goal was to make transportation as reliable as running water. And of course, in the developed world, running water enables so many possible things to occur that once you have intelligence that's as accessible as running water in everything that we build, the way in which we construct applications and interact with software and interact with each other through software also can change. So, for example, you and I having this conversation right now, but there's no reason why, shortly in the future, we wouldn't have some other agent that's part of this conversation that's either reminding you of a certain thing that we should talk about, or, you know, doing real time fact checking, which might be annoying but then causes us to be more factual about our conversation, or is literally diagramming our conversation as it's unfolding and emerging. And we can go back to this topic that I brought up before that we didn't get into, and then continue down that path. So the way of augmenting conversations feels like something that is like very full of possibility and potential. And given my experience, you know, on the outside, advocating for a solution, you know, which became and is the hashtag. It's about understanding the fundamental way in which people want to understand and connect, and being very observant and slowing yourself down to being like, what is it that I'm trying to get across right now, and why am I doing it in this way? What is it that the medium that exists is forcing me to do, you know, to you know, like the change from square photos on the Instagram profile to vertical, you know, rectangles is one of those moments where it's going to cause all these people to change how they express themselves, because they're being forced into a structure that is based on someone else's priorities. Going forward, you can imagine, I'm starting to imagine AI more as a medium, and if you think about it as a medium and not a tool, that allows you to start thinking about the ways in which you can express yourself more clearly, more with greater veracity or greater empathy, in a way that our current media prioritizes speed as opposed to comprehension and understanding. So those, to me, feel like great opportunities to build social software that are opportunities that we haven't solved for in the past 20 years because we've just been getting people online. So what does that look like? And how do we create not more slow, necessarily, but more understanding. When people are communicating in these mixed contexts where context has collapsed, maybe it's about standing context back up, back up again.
Mike McCue:
That's pretty cool. Yeah, there. There are a lot of things that have changed, a lot of assumptions that have changed, but then there's also a whole set of things that have not changed. You know, how people want to be able to connect to each other and communicate with each other, and those experiences together, right? So it is, it is interesting, the the technology tools, sometimes you can forget about those common fundamental truths, because the tool just doesn't accommodate that so well, right and right. And so going back to those and then asking yourselves, kind of, from a first principles point of view, what could these tools be doing better to help us connect.
Chris Messina:
There's a younger generation growing up on TikTok where it appears that their attention span has gotten very, very brief and very, very short. And so you try to have a longer form conversation with them, or you try to get them to read books. And I don't want to, like overgeneralize, but it seems like there's just less capacity for all the filler stuff that you used to add into and especially when books were paid for based on the number of pages in them, not the quality of content. And so there will be a pendulum swing that goes back and forth between these things and understanding the ways in which people want to connect and commune and share and learn and discover, create all sorts of opportunities, I think, for people to improve the technologies that we already have. I mean, you know, as McLuhan or his assistant, said, You know, we shape our tools, thereafter our tools shape us, which previously was, we shape our buildings, and then our building shape. But whatever, what like, whatever the neologism might have been. I think there is a fact aspect to that, that when you have mass mediums, it causes the shape of our thoughts to change, and as a result, once you become aware of that, you can push back to say, actually the shape of this thought needs to be a balloon shape, or needs to be a giraffe shape, or whatever sort of like crazy thing is. And you can go about constructing that now that this medium is becoming a lot more interactive, like the fact that software writes itself now, should open up a whole new opportunity for a generation to basically construct disposable little apps that are about demonstrating an idea or a concept, or expressing or sharing something, or creating art. And that's very exciting, you know.
Mike McCue:
And I think the simplicity with which you've attached, you've attacked some of these ideas like the hashtag, to me, is just it's like an inspiration as a product builder, right, to really kind of think through all these different components and come up with a simple, elegant thing that has stand stood the test of time. And in fact, I think it's actually become more valuable now, and and I think will become even more valuable in the coming, you know, months and years. So I, you know, thank you for everything that you've done, and for the work you've done in your career, and the products you've helped build, and the idea. So it's, it's been, it's been fantastic talking to you. Chris.
Chris Messina:
Yeah, well, thanks for having me. I appreciate it.
Well, thanks so much for listening.
You can learn more about Chris at his website, chrismessina.me You can also find him on Bluesky @chrismessina.me and on Threads at Chris.
You can follow Mike at @mike@flipboard.social and mike@flipboard.com.
Big thanks to our editors, Rosanna Caban and Anh Le.
To learn more about what Flipboard is doing in the fediverse, check out the link to Surf in this show's notes or at surf.social.
Until next time, see you in the fediverse!