How Kelo Social Was Born

Hello everyone,

A little over a month ago, I started writing the first lines of what would eventually become Kelo Social, an alternative to Bluesky built around the AT Protocol.

But the story actually began several weeks earlier.

For some time, I had been watching what was happening around Bluesky. I saw new accounts joining, others disappearing, and a huge range of different reactions. Some people felt the platform was too left-wing, others too right-wing or too centrist. Some were very happy with it, while others announced that they were leaving.

That made me ask myself a question: how can a social network that aims to be a welcoming place see so many users join and then leave?

I didn’t think politics alone could explain it.

So, for several weeks, I went through online reviews and comments. I wrote everything down.

Some people complained about how much space politics occupied. Others felt there wasn’t enough news. Some complained about the number of bots. Others simply said that the design of the app didn’t make them want to come back.

One day, I was sitting on a bench near my school with all my notes. At the time, a strike had been going on for about a month and a half, so we didn’t have classes.

My best friend came over and asked me what I was doing.

I explained all my research to her.

She replied:

“But that happens to lots of social networks.”

I explained that what really caught my attention was the constant movement of users. It’s normal for a social network to remove accounts for moderation reasons, or for some people to leave social media for personal reasons.

But seeing large numbers of accounts join while even more people voluntarily leave, again and again, seemed to me like a sign of a deeper problem.

I gave her all my research.

She read through it and said something that would ultimately kick-start the project:

“You’re good at building websites, and you know how to create social networks.”

I told her that was true, but I didn’t have a large enough community to launch an entirely new social network from scratch.

She looked at my notes and replied:

“But you wrote here that it’s decentralized and that several applications can connect to the AT Protocol. So start from there.”

That was the moment everything began.

She suggested that we work together during the week to go through all my research again: how the platform should work, the problems users had reported, the design, and the possible solutions.

The next day, we got to work.

She focused especially on the design. She draws a lot and takes design-related classes, which helped her think about what could make an interface more attractive and enjoyable to use.

I focused more on how everything would work.

One problem kept coming up: bots.

So my first thought was: we need a way to verify that a user is actually human.

But I quickly realized that wasn’t enough.

The AT Protocol isn’t used only by individuals. There are also companies, media organizations, institutions, associations, universities, and many other types of organizations.

So we needed to imagine a system capable of verifying several different kinds of identities.

My best friend then suggested creating a separate verification system, taking inspiration from the idea behind WIdentité used by WSocial, but without limiting ourselves to individual people.

That’s how the idea for Kelo ID was born.

Then another question came up: what should we call the social network?

Eventually, we decided to keep everything within the same universe: Kelo ID for verification and Kelo Social for the social network.

That way, whenever someone heard the name “Kelo,” the two services could naturally be associated with each other.

From that point on, I worked an enormous amount.

For more than a month, I would sometimes go to sleep around 3 a.m., wake up around 9 a.m., and start coding again.

My best friend would regularly call me to see how the project was progressing, what had changed, and to think with me about possible improvements.

Then one day, I called her and told her:

“Kelo is born. Kelo Social is born.”

But something was still missing.

Kelo ID.

She came to my house about fifteen minutes later, and we started analyzing Kelo Social as a whole.

We found bugs and things that needed improvement, but above all, we felt that we had built something that addressed a large part of the problems I had identified several weeks earlier.

Then she asked me:

“And did you build the identity system?”

That’s when I realized I had completely forgotten about it.

The main challenge was figuring out how to verify someone while still protecting their data.

So we imagined a gradual approach.

At first, Kelo ID could offer manual verification methods. Later, more immediate methods could be introduced depending on what was available in each country: NFC, identity card readers connected to computers, and so on.

The idea was to start with Belgium, since we are Belgian, then gradually expand to France, Spain, Portugal, Switzerland, Italy, and the rest of Europe before going further.

Then there was the question of organizations.

For them, we imagined a system capable of comparing the information they provide with publicly available information online. Whenever an automated verification cannot reach a sufficient level of confidence, the case could then be sent for manual review.

We spent several more weeks thinking, building, changing things, starting over, and coming up with new ideas.

And about two weeks ago, we looked at each other and said:

“That’s it. We did it. We have the foundations.”

So Kelo Social wasn’t born simply from the desire to create “another Bluesky.”

It was born from weeks spent reading user feedback, trying to understand why some people were leaving a platform they had initially chosen to join, and then thinking about what could be done differently.

And above all, it was made possible by one essential characteristic:

The AT Protocol makes it possible for different applications, communities, and ideas to coexist within the same decentralized ecosystem.

Today, Kelo Social and Kelo ID are still evolving. There is still work to do, bugs to fix, and certainly many things we haven’t thought of yet.

But it all started with a few sheets of notes, a bench near my school, a conversation with my best friend, and one very simple sentence:

“So start from there.”

3 Likes

Hello and thanks for the write up, congrats on getting things launched.

One thing (and we don’t have anything written down about this yet, so this is not your fault) — it would be great if you participated in the forum with your human / personal account rather than your brand account.

And thanks for posting in French as well

2 Likes

Welcome!

Looks like this is your website and atmosphere account?

I recommend searching around on this forum for things like ‘verification’ and ‘trust’. You’ll find some great topics relevant to your interests, like these:

1 Like

So contrary to popular belief, ID Verification and real name policies don’t actually make platforms safer / nicer, in fact the opposite is often observed. The paper Digital Social Norm Enforcement: Online Firestorms in Social Media (Katja Rost, Lea Stahel, Bruno S. Frey), looks at this in the context of openpetition.de — non-anonymous commenters were more aggressive than anonymous ones. In 2007, South Korea adopted a real names policy for social media, and it was scrapped shortly afterwards, as it only decreased malicious comments 0.9 percent.

There’s been other studies conducted on this too, and all you really need to do is look at the comments on facebook posts to know that these real name / real identity policies just don’t work, counter to popular cultural & political beliefs.

Also, age and identity information is a treasure trove for hackers and other malicious actors, as well as for advertisers who can use that information for microtargeting.

Already the recently launched W Social (also using AT Protocol under the hood) has seen controversy due to real names, where they had a user verify identity but use the handle & display name of TikTok. Ultimately they banned the account. Also, their data processing arguments fall flat as the controllers collapse together as both share execs & board members — it’s provably not acting as a separate entity (and W Identity actually delegates identity verification to Trust Anchor Group).

So based on all this, I’d be asking: is your assumption about real people vs bots actually going to hold up, or do you just need better labelers to hide and filter out bot behaviour (plenty exist already), if that’s even the reason why people left in the first place.

(Also, there’s some fairly infamous product development quotes which state “the user doesn’t really know what they want, just what they think they want”)

6 Likes

Bonjour,

Nous n’obligeons pas que l utilisateur affiche son nom et prĂ©nom Kelo ID vĂ©rifie juste que l utilisateur est humain si aprĂšs il souhaite s appeller “roi du monde” ou “pĂątissier amateur” ou autre. Kelo ID ne fait que vĂ©rifier que l utilisateur est humain ou une organisation c est tout et les donnĂ©s sont supprimĂ© tout de suite aprĂšs la vĂ©rification car les donnĂ©es ne servent plus rien une fois la vĂ©rification rĂ©alisĂ©e et accordĂ©

If verification info is discarded, how do you prefer people just
creating multiple “verified” identities to e.g. run bot networks?

Bonjour,

la vĂ©rification se fait via NFC ou pour les organisations via notre IA ou manuellement de notre cĂŽtĂ© le robot n’ont pas droit un badges humain ou organisation il y aura un badge robot qui arrive bientĂŽt.