Tech: Unpacking the Digital DNA of Social
I’m writing this the week before my summer holiday, so I’m already deep in the summer mood. The living is… well, if not easy, at least a bit slower. That makes it the perfect time for thought experiments that don’t require too much seriousness. So, I’m kicking off a summer series1 that might be a series of one, but I have a stretch goal of getting at least a second post out before the leaves start to turn. So grab a cold drink, and let’s get into it.
We often discuss “social media,” throwing platforms like Facebook, TikTok, Reddit, and Discord into the same bucket. But what if we’d classify them by their core technology2? By focusing on the underlying architecture, we might gain a clearer understanding of how these services work and why they feel so distinct.
This architectural DNA dictates everything, including a topic everyone should understand: the filter bubble. Let’s break it down into five levels.
Level 1: The Forum (Asynchronous & Channel-Centric)
The simplest architecture. Think of a classic internet forum or a Subreddit.
Tech: Communication is asynchronous (posted and read at different times) and organized into channels or topics, not around individual users.
Result: You have to actively seek out content. This “active-pull” model means there’s no real filter bubble; you are in complete control of what you see.
Level 2: The Chat Room (Synchronous & Channel-Centric)
Technically similar to a forum, but with one key change. Think Discord.
Tech: Communication is synchronous (real-time). It’s still organized into channels or “servers.”
Result: The experience becomes a live conversation. Because it’s still channel-based, it’s more about focused community interaction than broad content discovery. The risk of a filter bubble remains low.
Level 3: The Social Graph (User-Centric & Chronological)
This is the architecture of early Facebook and Twitter, marking a fundamental shift.
Tech: The organizing principle is the user-centric social graph (i.e., who you follow), and the feed is presented in chronological order.
Result: The platform is a passive aggregator of your friends’ posts. This architecture gives birth to the user-created filter bubble. The bubble is a direct and transparent result of your own choices—who you decide to follow or friend.
Level 4: The Algorithm (User-Centric & Curated)
The most significant leap in the taxonomy encompasses most modern social media platforms.
Tech: A powerful, opaque (or open source) set of algorithms and advanced analytics3 is inserted between you and your social graph. It decides the order of your feed, not the user.
The Result: This is the shift to the platform-created filter bubble. The algorithm optimizes engagement, often learning that emotionally charged content keeps you scrolling.
Your reality is now curated by a black box, transferring control from you to the platform. This curation is a critical distinction. From an objective technical perspective, the moment a platform decides what to show you, the responsibility for that content—including illegal or hostile material—moves clearly to the platform. There are no longer any valid arguments for claiming the platform to be a neutral host4; it’s an active publisher, which should have significant legal implications.
| Dimension | Chronological Feed (Level 3) | Algorithmic Feed (Level 4) |
|---|---|---|
| Control | User-controlled via follows | Platform-controlled via algorithm |
| Transparency | High (Newest first) | Low (A “black box”) |
| Bubble Type | User-Created | Platform-Created |
| Responsibility | User / Community | Platform |
Level 5: The Persuasion Engine (Aggressive Algorithm)
The final stage is exemplified by “For You” -platforms like TikTok, modern Instagram, and Threads5.
Tech: The Algorithm moves beyond analyzing explicit actions (likes, follows) to tracking granular, subconscious behaviors in real-time—like scroll speed, re-watches, and even mouse movements.
Result: This creates the most potent form of the platform-created filter bubble. The system builds an incredibly detailed behavioral profile to deliver a hyper-personalized stream of content designed to maximize your time on the site, often with proven consequences for the user’s mental well-being.
A Framework for Builders and Regulators?
This framework isn’t just an analytical tool; it’s a practical one. For digital professionals, it provides a precise, shared language. Instead of talking vaguely about “connecting to social media,” we can discuss the specific architectural realities: Are we integrating with a Level 3 chronological feed or a Level 5 persuasion engine? The technical and ethical requirements are vastly different.
More importantly, this could act as a scaffold for regulation. The term “social media” is too broad to facilitate the creation of fair laws. This taxonomy allows for a more nuanced approach. Perhaps the high-responsibility, high-risk platforms at Levels 4 and 5 warrant significant regulatory oversight, while the user-driven models of “traditional” social media at Level 3 and the community spaces at Levels 1 and 2 should be treated very differently6. It enables us to focus our attention on the areas where architectural risk is most significant.
Footnotes
-
I was supposed to write on a more serious topic, but… we should return to that when it’s raining, dark and cold. The way I like it. ↩
-
Most of the taxonomies are based on the business or social characteristics of the platforms. These taxonomies often end in a set of fuzzy, unsatisfying, and disputable categorizations. ↩
-
It’s this set of advanced analytics - including, but not limited to, AI technologies like machine learning - and the algorithmically sorted feeds that people have started to call “The Algorithm.” While I’d like to scream and shout at people who use the term, that would not change anything. Thus, it’s likely more appropriate to do as we did with the ‘internets’… and use the capital’ Algorithm’ for the set above and the non-capitalized ‘algorithm’ for an algorithm. ↩
-
Technically, there might be a small loophole here if we look at services like Bluesky and Mastodon - where the user can choose or develop their own feed algorithms. However, even in these cases, the platform provides the default Algorithm, so the taxonomy stands. ↩
-
As with levels 3 and 4, your mileage may vary. The level of data collected by late-era Twitter was likely limited to tracking how long a post was visible before the user pressed ‘like.’ In contrast, some platforms measure real-time data, ranging from the user’s typing speed to the amount of background noise at the user’s location. ↩
-
The taxonomy also clearly explains why a chat or a forum is not a social media, not from the user or technology perspective, at least. However, on the internet, the elegance or correctness of arguments likely has minimal impact on anything. Except to prove I’m silly enough to care. ↩