Introducing a network for thoughtful conversations

Proposing a new kind of tool for deliberate discourse on the Internet.

The quality of conversations on the Internet is terrible. Most platforms like Twitter, Facebook, Instagram, Reddit, etc. optimize for engagement as their key metric. This spawns and rewards behavior that leads to bragging of gameable metrics such as likes and followers, self-feeding thought bubbles, knee-jerk reactions, if you’re not on our side you’re on their side mentality, and thought policing even by people who think they’re fighting for a good cause. Today’s Internet doesn’t incentivize healthy conversations.

This isn’t a novel observation. Nevertheless, the problem seems to be getting worse by the day. For the last couple of months, I’ve been honing an idea of a new kind of network that I believe is a solution to the problem. A tool, specifically, that facilitates advancing conversations. I’m calling it LinkTalk.

LinkTalk is a network where you can only post links. You can’t caption your post or write any introductory text. None. To respond to a link, you post another link. This is how we can have debates, rebuttals, a stream of links that provide curated information on a topic, and more. That’s it, that’s the core idea. Converse in links.

While I’m primarily thinking of links as in blog posts, articles and research papers, other mediums of disseminating information such as podcasts and videos are also welcome.

Let me address the obvious question upfront: If people can’t write any text in posts, how will others know what to expect from the shared link?

You must have noticed when you share a link anywhere, most networks and messaging apps autogenerate a link preview. This includes a title, a subtitle or lead text, and a thumbnail. That is sufficient information when a user views a link on LinkTalk, the title and subtitle being the key. The goal is to make people click links and consume them, not look for (often inaccurate) TLDRs in the comments section. Besides, Twitter and other networks already exist so there’s little point in not breaking some notions and testing new waters.

Now come the juicy bits. What follows is my first ideation of the specific mechanics and incentive structure of LinkTalk.

Legit links

How do we get people to post legit links on the network and not just links to tweets, GIFs, or spam in general?

I have no technical knowledge of how such things work at an implementation level but three things come to my mind that can act as a basic posting filter.

  1. The system will check if the link being submitted has an associated RSS feed, and only allow posting if so. Blogs, many research journals, YouTube channels, podcasts, all have RSS feeds to aid reading and distribution, whereas things like tweets, Instagram photos, and LinkedIn posts don’t. So that’s a natively available filter on the web for us to utilize.

  2. Of course, an RSS filter would catch only some unwanted links. After all, RSS is a versatile beast also used for things like tracking changes in Wikipedia pages. So the next step would be to maintain a blacklist which would disallow posting links from unwanted websites in the context of the network.

  3. While we’re on this topic, let’s also disallow URL shortening services. Users should be able to clearly see which website they’re heading to.

Needless to say, it’d require quite a bit of work to even make and maintain such a blacklist but my hope is that these filters block a reasonable amount of spam. But since I have no knowledge of such things at a technical level, including how to handle bot problems, an expert needs to chime in here.


People consuming links on LinkTalk need to be able to react in some way, typically to throw their weight on or against a link’s quality. I imagine implementing this differently than the usual likes, retweets or shares mechanism. Each link could show the following buttons to vote on:

  • Good link

  • Bad link

  • Disproved

  • It’s complicated

People can vote links as good and bad the usual way—just select your reaction. But when a user selects any of the other two reactions, they are required to post a link as well. So when you click “Disproved” on a link someone posted, you need to reply with a link that you think proves it wrong. If you think a link has many gray areas or caveats, use “It’s complicated” to reply with a link that justifies that reaction to you.

The benefit of such a mechanism is clear. For every user that strongly disagrees with a post or thinks they know more about a matter, the system helps others ride along by having the relevant links available by design.

Displaying reactions

An interesting divergence in displaying reactions on LinkTalk from other networks would be to not show how many and which reactions a link got until you vote yourself. This is similar to how polls work. Let’s accept it, most people on social networks only upvote content because 100, 1,000 or 50,000 people already liked it. This is the self-feeding bubble we want to break by having people realize their honest opinion first.

Next, a post view on LinkTalk will always show vote count for each reaction separately. This is to avoid giving the impression that a cumulative score of say, +50, somehow makes the thing positive. This will help maintain some nuance, which is the easiest thing to get lost in conversations on modern networks.

Following, not followers

The network effect is real. An account with over 10,000 followers has their content better heard overall than one with 500, no matter the content. People also tend to follow people who have a large following over those who don’t—another vicious loop. So on LinkTalk, each profile would only show who they’re following, not who or how many are following them. This should incentivize people to find more people they actually want to follow, and not get hung up on a popularity contest. This concept is borrowed from, a platform with some neat ideas.


You can reply to a posted link with another on LinkTalk, even if you don’t react to said link. Each reply post has all the same properties as a regular one so it can be reacted and replied to the same way. Conversations can thus happen in multiple threads, not unlike Twitter, to as many levels as the replies go. The idea with replies on the network is that you either find a good link to reply with or write one of your own, which is where the next feature comes in.

Reply by writing a blog post

When you hit reply on LinkTalk, in addition to allowing you to post a link, you can initiate a new blog post via a button. This will open a new post editor on your blog, which you may have connected to via your account settings or during onboarding. If multiple blogs are connected, you can choose which one to reply with.

I think this is good for two reasons. First, writing a blog post requires a more nuanced mindset than impulsively replying on traditional social platforms. Second, even if the blog post’s response length is a mere 100-200 words, it’s fine. Since the response will go as a post on your blog, you’ll be more mindful of what you write for your readers too.

I think this feature is a good way to incentivize a culture of blogging, which has been lost, while keeping conversation quality in check. With this feature, LinkTalk also incentivizes you to own your thoughts via your blog, unlike the pseudo-ownership that social networks provide.

Besides, it has never been easier to start and write a blog.

Extending to the Web

LinkTalk seems relatively simple in terms of the number of things it needs to host a conversation. So why even limit it to a website? There are two ways I imagine sensibly extending LinkTalk.

  1. If site owners want, the comments section on webpages could host a tab for LinkTalk integration. People can then react and reply directly from a webpage. This tab could also show any existing replies to the link on the network. To integrate LinkTalk into a website, a mechanism similar to commenting systems like Disqus could be used.

  2. A browser extension, similar to Reddit Checker, that when clicked could show a list of LinkTalk submissions of the webpage you’re visiting, either as a shared post or a reply. It could also allow you to share the link on the network yourself or reply to it without leaving the page.

With these two extensions of LinkTalk, having conversations and finding them can happen organically on the open web, and not inside application silos. Unlike other networks, there’s not much incentive here to keep you on the network’s website. If anything, the more of the open web you explore, the more likely you are to add value to the network. So tools like these actually align the incentive structure more tightly to the core idea––better quality conversations on the Internet.

I have considered making LinkTalk federated so that people can spin their own, independent instances rather than have a centralized one like Twitter. But seeing the subpar user experience of Mastodon, and its moderation issues, I’m not convinced about federation being a net positive. Said that, I’m no expert and could be wrong.

Slow, deliberately so

At this point, it should be clear that LinkTalk discourages quick responses. While its tools are made to incentivize and accelerate a deliberate response itself, the network can, by definition, never be as spontaneous as Twitter, Facebook, Reddit, and the rest. Again, there would be little point if LinkTalk didn’t try to do something substantially different than already existing networks.

Said that, I do think LinkTalk has some resemblance to email conversations in terms of the mindset it places you in. Email conversations tend to be thoughtful because they’re asynchronous for allowing deliberation, they serve an established purpose of mutual benefit, and there’s incentive to not lose the other person’s interest. LinkTalk’s incentive structure positions it to benefit from such virtues, and potentially amplify them with the power of the open web.

Closing thoughts

So that was my first concept of a new kind of tool that encourages and rewards thoughtful conversations on the Internet, or so I hope. I will make additions or changes to this blog post as my thoughts refine and people provide feedback.

There are certainly many elements of the network I haven’t touched on, such as timeline, search and discovery, mentions, moderation (which is no joke), and much more. While I do have some more ideas to expand LinkTalk, I stop here for now. The more quantifiable elements you introduce to a network, the more it can be, and is, gamed. As such, any change or addition to the core idea of LinkTalk must be considered with caution—better to be safe than sorry.

At the same time, I’m sure that LinkTalk has its own demons, which would likely only be apparent and addressable if and when the network takes off. I don’t consider, or even intend, LinkTalk to be the ultimate solution for online conversations. It’s supposed to be a new tool that solves some problems better than other, existing tools. If LinkTalk increases discourse quality for even relatively small communities compared to large social networks, I’d consider it a success.

Now comes the bad news. I’m not going to build LinkTalk. Not now, not anytime in the near future. It’s not that I don’t care enough for what it intends to solve, I do. But I care even more about spending my limited time spreading the purpose of exploring space and the Moon, a journey I have barely begun.

Part of the reason I’m sharing the core idea of LinkTalk as this public blog post is that if it turns out to be good one, it shouldn’t remain locked with me. That’s the least I can do. My hope is that someone at some point builds LinkTalk or something similar. At its worst, it would be a successful experiment. At its best, I hope that I get due credit, and maybe some share of profits.

What I’m open to though is working on a small proof-of-concept (perhaps using a tool like Discourse?) with an interested person or a group of people. If you’re passionate about the same problem as I am, reach out to me with your background, your interest in the matter, and how and why do you think we should work together.