A Journey Through the YouNiverse
Understanding the Scale of the YouNiverse
Over the past two decades, YouTube has evolved from a simple video-sharing website into one of the largest social ecosystems ever created. With billions of viewers and creators interacting around the globe, it has become a global stage where cultures spread, trends emerge, and ideas collide.
Today, YouTube is not just entertainment: it is a reflection of society, a mirror of human curiosity, and a massive accelerator of cultural exchange. From music to science, from gaming to politics, its content shapes our habits, our conversations, and often, our worldview.
Revealing a Hidden Cosmic Landscape
Given the sheer scale of interactions happening daily on the platform, we are convinced that YouTube cannot be understood as just a collection of videos; it behaves more like an immense digital world with its own internal order.
And if we zoom out — far out — this world begins to resemble a vast cosmic landscape. Individual channels appear as stars; they cluster into large thematic galaxies, and within them, smaller constellations emerge around more specific interests.
A Universe We Cannot See
Yet, the inner order of this digital cosmos remains invisible. We travel through YouTube every day without ever seeing the structure that shapes what we discover, what we watch, and ultimately, what we think.
With billions of interactions steering opinions, habits, and cultural currents, YouTube now functions as a parallel society — one whose architecture remains hidden from its own inhabitants. This raises a fundamental question: how can an ecosystem of this magnitude remain a black box? And a more personal one: how much of what you watch is a choice — and how much is gravity? Understanding this structure is not optional; it is essential. YouTube offers a rare, large-scale laboratory for studying how communities emerge, evolve, and influence one another.
Our Expedition Map
To navigate this digital cosmos, our exploration is guided by four fundamental questions:
Before We Explore: First Glimpses of the YouNiverse
To answer these questions, we turned to the YouNiverse dataset. And the name truly does it justice, its size is that of a universe. Data from 73 million videos, 137 thousand channels and 8.6 billion comments from 449 million users on a timeframe from May 2005 to October 2019. The dataset claims to only hold data from english speaking videos (more on that later).
Looking at this universe with the naked eye, we see a system that has grown exponentially since its 'big bang'. What started as a few flickering lights has evolved into a sprawling ecosystem. However, this growth is not uniform.
When we peer through our telescope at these celestial bodies (channels), we notice a stark reality: size is not distributed equally. The distribution of views and subscribers follows a power law, a mathematical signature of winner-takes-all dynamics commonly observed in large-scale social systems. A few 'Supermassive' channels command the vast majority of the YouNiverse's light, while millions of smaller stars twinkle in the background. The top 1% of channels capture over 46% of all views, a cosmic inequality that shapes the entire ecosystem.
To understand the nature of these stars, we first look at their "Spectra" — the content categories assigned to them. This provides a first-order view of what content dominates the YouNiverse, before asking how these channels interact. While this gives us a hint of their composition (Gaming, News, Education), categories alone don’t tell us how these stars interact.
To see the true architecture of the YouNiverse, we must board our spaceship and look at the gravitational bonds between them.
Constructing the Cosmic Map: From Comments to Galaxies
To understand our YouNiverse, our objective is to see what channels actually share common audiences. By doing this we will be able to run community detection algorithms and visualize types of interactions.
To do this we will use the biggest file in our dataset, the comment data. We want to know which channels users most commented on. "We establish gravitational bonds between channels: when users orbit both, commenting frequently on each, we draw a connection weighted by the strength of their shared audience. The edge of these weights will then be first determined by the number of commenters they have in common. Two channels with many shared commenters are linked by strong gravitational bonds — they orbit in the same region of the YouNiverse.
Demonstration of graph creation using the top 3 channels of the user.
Normalizing the gravitational bonds
In space, a massive sun has more gravity simply because it is big. On YouTube, two "Black Hole" channels (like PewDiePie and T-Series) might share many commenters just because they are famous, not because they are related. To find true communities, we had to "level the playing field" by normalizing our edge weights. We developed a Similarity Score to penalize the sheer size of a channel, ensuring that a "Galaxy" is formed by genuine audience overlap rather than just popularity. Imagine trying to see stars near the sun — their light is drowned out. Similarly, mega-channels like PewDiePie overshadow genuine connections. Our Similarity Score acts like a solar filter: it dims the giants proportionally to their size, revealing the authentic audience overlaps that define true communities.
To see the true shape of the galaxies, we had to apply Gravitational Shielding. By using our similarity score, we effectively "muted" the blinding light of the Black Holes. This allowed us to see the faint, genuine connections between smaller stars that were previously invisible in the glare of the platform's giants.
🧮 The Mathematics of Normalization (Similarity Score)
The Problem: PewDiePie and T-Series might share 100,000 commenters simply because they each have 100 million subscribers. Meanwhile, two niche gaming channels sharing 500 commenters might have a much stronger cultural connection.
The Solution: We penalize large channels proportionally to their size:
$$size\_factor = \left( \frac{total\_commenters}{median\_commenters} \right)^\beta$$
$$penalty(c_1, c_2) = (size\_factor(c_1) \cdot size\_factor(c_2))^\alpha$$
$$Similarity\_Score = \frac{shared\_commenters}{penalty}$$
- size_factor: How much larger is this channel than typical channels?
- penalty: Expected overlap if connections were random.
- α, β: Tuning parameters (we used α=0.5, β=1).
Result: Two mega-channels need exponentially more shared commenters to appear "connected" than two small channels. This reveals genuine communities rather than just popularity contests.
Filtering the Cosmic Noise
Even with gravitational shielding applied, our universe still contained faint, unreliable connections, like distant radio signals lost in static. To build a clean map of meaningful relationships, we applied two critical filters to remove the cosmic noise.
- Minimum Subscribers Filter: We excluded channels below 200,000 subscribers. These "proto-stars" haven't yet established stable audiences, making their connections unreliable indicators of community structure.
- Minimum Edge Weight Filter: We removed edges with fewer than 25 shared commenters. Like eliminating background radiation, this ensures we only map gravitational bonds strong enough to define true communities.
Why these thresholds? Two reasons, aside from computational efficiency. Because of our top-5 channels logic, small channels frequently had very few shared commenters and by consequence too few edges for the analysis to actually be significant. Similarly, small weight edges were not consequential given the size of the dataset and the remaining channels. Through iterative testing, we have found out that the two chosen hyperparameters were the best balance to get meaningful communities.
📊 Impact of Filtering on Network Structure
Before Filtering:
- • Nodes (channels): 129,996
- • Edges (connections): 32,408,399
After Filtering:
- • Nodes (channels): 19,129
- • Edges (connections): 572,732
With our normalized map in hand, we applied the Louvain Community Detection algorithm. Think of this as a way to find where the "gas clouds" of users naturally condense into distinct structures.
Louvain seeks to maximize Modularity — a metric that measures how much "denser" the connections are within a galaxy compared to a random universe. Our map achieved a Modularity score of 0.655. In the world of network science, a score above 0.5 is a significant discovery. It confirms that the YouNiverse is not a chaotic cloud of random stars, but a structured system of distinct, high-density galaxies bound by shared cultures, languages, and interests.
The Architecture of the YouNiverse
By applying the Louvain algorithm, we successfully charted 52 distinct communities within the cosmic map. It became immediately clear that this universe is highly polarized: while some regions are vast, mainstream Super-Galaxies, others are small, isolated Niche Constellations.
The scale varies dramatically, the smallest of these communities contain as few as two channels, existing like distant binary stars in the deep void of the YouNiverse.
Before diving deeper, let's label our communities with the biggest channel in each. This gives us the first valuable insights into what sort of galaxies are there in the YouNiverse.
Inter-Community Flows: The Cosmic Connections
To visualize how these 52 galaxies interact, we map the flows of attention between communities. Each ribbon in the chord diagram represents the strength of connections between galaxies, revealing which communities share audiences and how tightly they are bound together.
The thickness and opacity of each connection reflect the normalized flow strength — stronger gravitational bonds appear more prominent, while weaker connections fade into the background. Hover over any connection to see the exact flow weight between two communities.
The Core vs. The Periphery
In the center of our map lies the Galactic Core. Here, the mainstream galaxies are so massive and their audiences so interconnected that their borders often blur together. Yet, despite this density, the Louvain algorithm reveals they remain distinct cultural entities with their own centers of gravity.
Stretching out from this core are thin, elongated edges: the Interstellar Bridges. These represent bridge channels that link the mainstream center to peripheral niche clusters. These channels serve as the connective tissue of the YouNiverse, allowing users to travel between vastly different content worlds.
A Voyage into the Core
To understand the internal divisions of the most dominant communities, we must look past the "space dust" of the periphery. By zooming into the core, we can observe the fine-tuned interactions and galaxies that shape the most popular regions of YouTube.
Showing the 5 biggest galaxies, those that form the core of our universe.
First Contact: A Glimpse Into Three Galaxies
Before we systematically analyze all galaxies, let's descend into three interesting ones, yet very small compared to the size of the major galaxies. We will quickly analyze what sort of channels lie in them to determine how they came to be. They define three contrasting types of communities. Do not hesitate to click on the little nodes to get the names of the channels and their total strength in the graph.
Tennis Terror
Galaxy #22 • 6 channels • Biggest channel : Tennis TV
A small, tight-knit community around a passion for one sport: Tennis.
Maghrebi Power
Galaxy #25 • 32 channels • Biggest channel : 7liwa
A single galaxy with what seems to be two different galaxies connected by two heavily interacting channels. Quick overview of these channels leads us to see that on one part we have some Moroccan singers and on the other side Tunisian ones.
English Learners
Galaxy #24 • 45 channels • Biggest channel : Go Natural English
Probably a hub for recharging your spacecraft before going exploring. This is a community based around learning the biggest language on the website (and the world), English.
These three examples showcase the diversity we've discovered. In the next chapter, we will systematically analyze what truly distinguishes all major galaxies in the YouNiverse.
What We've Discovered
By building our cosmic map from 8.6 billion comments across 19,000 channels, we've revealed something remarkable: YouTube isn't a chaotic cloud of random content. It's a structured universe with 52 distinct cultural galaxies, each with its own centers of gravity, its own audiences, and its own identity.
The modularity score of 0.655 confirms what we suspected: users don't wander randomly across YouTube. They orbit within specific communities, occasionally traveling between galaxies through bridge channels, but largely remaining within their cultural home.
With our map complete, we can now begin the real voyage: descending into each galaxy to meet its inhabitants, understand its culture, and trace the journeys of travelers between worlds.
A Decade of Transformation
Slicing Time: Investigating the Invisible Shift
To truly understand the evolution of the YouNiverse, viewing it as a static map is not enough; we must observe it as a living organism. A dataset spanning 14 years flattens history, it hides the dramatic shifts in culture, influence, and structure that occurred between the first viral hits and the modern creator economy.
To capture this motion, we employed a temporal snapshot methodology. By slicing our data into distinct one-year windows, we isolated the interactions of each era. This allowed us to build independent universes for 2010, 2011, through to 2019, enabling us to watch the laws of digital gravity change in real-time. What we found was not a linear expansion, but a complete structural metamorphosis from a unified community to a fractured multiverse.
The Big Bang of Communities
The most fundamental shift in the ecosystem is a massive +518% fragmentation of the network structure. In 2010, our community detection algorithms identified only 16 distinct clusters. At this stage, YouTube acted as a "digital town square": audiences were forced to mingle because there were simply fewer distinct cultural spaces to inhabit.
By 2019, this number had exploded to 99 distinct communities. This fragmentation was driven by two major cosmic events visible on the timeline: the Gaming explosion (starting around 2012) and the Indian expansion (triggered by Jio in 2016).
The Implication: This is not an accidental fracture, but a necessary adaptation to abundance. As the platform flooded with content, the "shared experience" became impossible to maintain. To survive the cognitive load, audiences naturally segregated into cognitive silos. Viewers retreated into hyper-specific clusters — from K-Pop to Minecraft — turning shared interests into hermetic worlds that rarely interact.
The Connectivity Paradox
This creation of silos leads us to a startling paradox: as the universe expanded, it became less connected. While the number of channels grew, the structural density of the network collapsed.
To understand the scale of this collapse, we must look at how density $D$ is calculated:
In 2010, the network density was 0.05. In simple terms, out of all possible conversations between communities, 5% actually happened. It was a "small world" where distinct groups frequently rubbed shoulders. By 2019, this density had flatlined to 0.002 — a staggering 96% drop in relative connectivity.
This curve is the filter bubble made visible. The mathematical collapse of $D$ represents the death of serendipity. As algorithms got better at prediction, they stopped building bridges between galaxies. Viewers stayed in their comfortable corners, and the "global conversation" of 2010 dissolved into thousands of private, parallel conversations.
Galaxies Rising and Fading
The heart of the YouNiverse has undergone a seismic transition. Our PageRank centrality analysis — a measure of structural influence based on how audiences move between channels — captures the fall of an empire and the birth of a new one.
- The Great Passing of the Torch (2012): In 2010, Music (red curve 🎵) reigned supreme, commanding over 30% of the platform's total influence. By 2012, the graph shows a historic crossover: the beginning of the Gaming Era (green curve 🎮). Let's Play creators shifted from peripheral niches to becoming the platform's new centers of gravity.
- The Decline of Music: Once a central pillar, Music saw its structural influence steadily collapse as discovery migrated to dedicated streaming services. This trend shows YouTube evolving from a "video radio" into a home for community-driven creator content.
- The Quiet Ascent: While Gaming and Music fought for the top spot, other galaxies grew steadily. Categories like Entertainment (blue curve 🎭) and People & Blogs (purple curve 👥) saw consistent growth, proving that the YouNiverse has diversified far beyond simple passive entertainment.
The Emergence of New Powers: The Jio Effect
While the Western core was busy transitioning into the Gaming Era, a seismic shift was happening on the periphery. Our temporal tracking reveals the emergence of massive regional superpowers that fundamentally changed the scale of the YouNiverse.
The most prominent example is the "Jio Effect" of 2016. As seen in our fragmentation data, the entry of millions of Indian users led to a sudden spike in community detection. This era birthed the massive T-Series galaxy. As shown in the "Evolution of Top 5 Identities" chart, by 2016, T-Series shifted from a peripheral node to the third-largest community on the entire platform.
Parallel Worlds: What makes this growth remarkable is its isolation. Despite becoming a global giant, the T-Series galaxy grew with almost zero audience overlap with Western giants like PewDiePie. This proved that the YouNiverse was no longer a single shared experience, but a multiverse of parallel worlds. By 2019, YouTube had become a collection of massive, independent islands — vibrant and rich, yet structurally invisible to one another.
Visualizing the Drift: A Universe in Motion
Animating these annual snapshots through our physics engine reveals the structural explosion of the network. In 2010, the ecosystem appears as a compact, monolithic block: there are few communities, and they are all tightly interconnected within a restricted space, forming a cohesive mass.
Over the decade, the sheer pressure of growth fractures this block. The number of communities rises drastically, creating a centrifugal force that propels new "niche galaxies" outward. The final result in 2019 is no longer a unified network, but a vast archipelago: a fusing core persists at the center, surrounded by a multitude of specialized islands floating on the periphery, orbiting far from the central gravity without ever touching it.
Interactive Model: Drag nodes to explore connections, use the slider to travel through time.
Understanding the Identity of Galaxies
We've mapped 52 galaxies, but what makes each one unique? YouTube assigns categories to channels, but these official labels only scratch the surface. A "Music" galaxy in one region might be entirely different from another. To truly understand each galaxy's identity, we need to look deeper: into the content itself, the language, and the behavior of its inhabitants.
Reading the Stars: Topic Detection with LDA
We deployed Latent Dirichlet Allocation (LDA) on video titles and descriptions from the 10 largest galaxies. Using POS-filtering (nouns, proper nouns, adjectives) and bigram detection, we extract the hidden thematic structure within each community — the actual topics people talk about, not just the labels YouTube assigns.
The results largely validate YouTube's categorization while revealing rich subcultures within each galaxy. Regional ecosystems also emerge, with entire non-English communities thriving in an allegedly "English-only" dataset.
The Language Barrier: Regional Galaxies Emerge
One striking discovery: language creates galaxies. Despite the YouNiverse dataset claiming to contain only "English-speaking" content, our analysis reveals massive non-English communities bound together by shared language.
- 🇮🇳 Galaxy #0 - Indian Entertainment Hub (3,187 channels): Bollywood dominates (Filmfare, Salman Khan, Varun Dhawan). Hindi beauty tips, Navratri celebrations, and PUBG Mobile.
- 🇵🇭 Galaxy #7 - Filipino Entertainment Hub (596 channels): Mobile Legends: Bang Bang is the gravitational center. ABS-CBN content, Himig Handog music, and Filipino language (ang, lang, ako, ikaw).
These linguistic gravitational bonds are so strong that they override thematic connections. An Indian gamer has more in common with an Indian news channel than with an American gamer — at least in terms of shared audience.
Categories Confirmed: LDA Validates YouTube Labels
Interestingly, our LDA analysis largely confirms YouTube's category assignments. The detected topics align remarkably well with the official labels:
- Galaxy #1 (Gaming 48%): GTA mods, Roblox, Minecraft roleplay 🎮
- Galaxy #2 (Music 64%): Ariana Grande, Nicki Minaj, chill playlists 🎵
- Galaxy #3 (Howto & Style 31%): Vlogs, makeup tutorials, hauls 💄
- Galaxy #9 (Autos & Vehicles 42%): Car reviews, bikes, off-road 🚗
This validates that co-commenting behavior reflects genuine content affinity. Users who comment together genuinely share interests that match the channel's official category — the community structure is real.
Behavioral Signatures: How Galaxies Engage
Beyond topics, each galaxy has a distinct behavioral fingerprint. By analyzing engagement metrics across millions of videos, we uncovered dramatically different patterns of audience interaction.
The Engagement Paradox
Our analysis reveals a fascinating paradox: the most-viewed content isn't the most engaging.
Most Engaged: Galaxy #3 - Vlogs & Lifestyle
26.9 likes per 1000 views
Makeup tutorials, hauls, and vlogs create devoted communities. These viewers don't just watch — they comment, share tips, and build relationships with creators.
The Viral Void: Galaxy #8 - ASMR & Kids
133K median views, only 4.0 engagement
Slime, soap cutting, and satisfying compilations — optimized for the algorithm, not connection. Viewers consume passively without forming lasting communities.
Duration Tells a Story
Video length isn't random — it reflects the content type and audience expectations of each galaxy.
(Workout videos, vegan recipes)
(Music videos, playlists)
(Let's plays, GTA mods)
Fitness content (Galaxy #5) demands the longest attention — Buff Dudes workouts, vegan recipes, and Law of Attraction manifestation videos average 18+ minutes. Pop Music (Galaxy #2) lives in quick 8-minute bursts: music videos and playlist compilations.
The Galaxy Classification System
Combining topic analysis and behavioral metrics, we can classify our galaxies into distinct types:
Global Galaxies
#1 Gaming, #2 Music, #8 ASMR
Content transcends language
Regional Galaxies
#0 Indian, #7 Filipino
Bound by language
Engaged Galaxies
#3 Vlogs, #5 Fitness
High engagement, tight communities
What We've Learned About Galaxy Identity
Our deep dive into the 10 largest galaxies reveals that YouTube's official categories are only part of the story. The true identity of a galaxy emerges from the intersection of:
- Thematic content: What topics dominate the conversation
- Language: The gravitational force that binds regional communities
- Engagement patterns: Active fans vs. passive consumers
- Content format: Long-form immersion vs. short-form consumption
With the identity of each galaxy now mapped, our next question becomes: how do travelers move between these worlds? Do they stay loyal to their home galaxy, or do they explore the broader universe?
The Black Holes of the YouNiverse
On platforms like YouTube, attention — understood as the capacity to maintain both broad connectivity and intense interactions — is the most critical resource, even more than views or subscriber counts. Attention enables channels to cross community boundaries, tap into multiple audiences, and capture flows from across the ecosystem. It is precisely this capacity that makes attention so powerful — and potentially so dangerous.
Given this, it becomes essential to study how attention is distributed across the YouNiverse, both within and across galaxies.
Our goal is simple: to determine whether attention is relatively evenly distributed, or whether it collapses around a small number of dominant actors — the Black Holes.
Within galaxies
We begin our exploration at the scale of individual galaxies by providing a broad estimation of how interaction intensity (strength) and connectivity (degree) are distributed internally. To quantify these inequalities, we rely on the Gini coefficient.
Two main observations immediately emerge.
First, attention is highly unevenly distributed inside most galaxies. Gini coefficients are typically above 0.7 and can reach values close to 0.9, revealing extreme concentration among a small subset of channels.
Second, we observe a clear relationship between connectivity inequality and interaction inequality. Galaxies with highly unequal degree distributions also tend to exhibit highly unequal strength distributions. However, inequality in interaction intensity is consistently slightly higher than inequality in connectivity, suggesting that attention concentrates even more strongly in how intensely channels interact than in how many connections they have.
How much do the giants absorb?
Having established that attention is globally unequally distributed within most galaxies, we now quantify how much of this attention is captured by the most dominant channels: the Black Holes. We define black holes as channels belonging simultaneously to the top percentile of interaction strength and degree within their community.
At the 1% level, we already observe a striking concentration of attention within galaxies. Expanding to the top 5% largely reinforces this picture. In Community 20, the top 5% subset alone accounts for around 70% of total interaction strength, revealing an extreme collapse of attention around a narrow core.
Comparing these thresholds highlights two important patterns. First, within a given galaxy, concentration often increases sharply when moving from the top 1% to the top 5%, suggesting that dominance typically arises from a small core of influential channels rather than from a single superstar. Second, rankings change across scales: some galaxies are driven by an ultra-elite at the very top, while others rely on a broader but still highly dominant core. This variability shows that black holes do not all share the same internal structure.
Internal structure of the black holes
To validate this hypothesis and better understand the nature of black holes, we examine the internal distribution of attention inside the top 5% subset across different galaxies.
The patterns vary substantially across communities. For instance, Community 7 shows a very large gap between median (or mean) and maximum values, pointing to a sharply peaked internal distribution. In contrast, communities such as Community 0 display smaller gaps, suggesting a more distributed dominant core.
Who are the black holes?
After characterising the internal structures of black holes, we now turn to their concrete identities. Which channels actually concentrate the largest share of attention inside each galaxy?
Use the selector below to explore all communities and inspect the top black-hole channels. For each galaxy, we display the share of total community strength captured by each dominant channel.
-
“Single-superstar” black hole —
one channel dominates the core.
Community 7: the leading channel alone captures ≈ 16% of total community strength, followed by a sharp drop.
Interpretation: one extremely dominant hub surrounded by a weak halo. -
“Two-tier” black hole —
a small inner elite.
Community 2: the top two channels capture ≈ 12% and 8%, before a rapid decline.
Interpretation: dominance is shared by a very small inner core. -
“Broad-core” black hole —
many channels share dominance.
Communities 0, 5, 15, 16, 17 show a gradual decay (≈ 6%, 3%, 2%, 1%, …), while Communities 9, 10, 20, 29 exhibit even flatter profiles.
Interpretation: attention collapses into a dense influential nucleus rather than a single superstar.
Conclusion — Within galaxies
Inside the YouNiverse galaxies, attention rarely spreads evenly. Instead, most galaxies are shaped by a gravitational pull: a small number of channels draws in a disproportionate share of interactions, bending the internal structure of the community around them.
Yet these black holes are not all alike. Some galaxies orbit around a single dominant star, while others are held together by a compact inner core or a broader shared nucleus. Inequality is universal, but the way it takes shape tells a different story in each galaxy.
Across galaxies: How is it structured?
After examining how attention is distributed within individual galaxies, we now shift perspective to a broader scale and compare galaxies with one another.
We associate the amount of attention concentrated by a galaxy with its total interaction strength (`total_strength`), or in other words, the overall intensity of exchanges taking place within a galaxy.
In a first step, we focus on characterizing the global distribution of total interaction strength across galaxies, in order to assess how unevenly attention is allocated at the ecosystem level.
The histogram reveals a highly skewed distribution of total interaction strength across galaxies. A large majority of galaxies exhibit very low total strength, while a small number reach extremely high values. This strong asymmetry suggests that attention is far from evenly spread across the YouNiverse.
To quantify how this skewness translates into the global allocation of attention, we now examine the cumulative distribution of total attention across galaxies.
This cumulative view confirms what was already suggested by the histogram: attention is extremely unevenly distributed across galaxies. We observe that the top 10% of galaxies, ranked by total interaction strength, already concentrate close to 85% of the total attention in the YouNiverse. This indicates that a very small fraction of galaxies dominates the overall interaction activity.
Why such a gap?
Having established the existence of such a strong global imbalance, we now turn to the question of why this concentration emerges. To do so, we identify the galaxies belonging to this top-attention group and examine their main structural properties.
A first observation is that all galaxies belonging to the top-attention group also rank among the largest in terms of number of channels. This suggests a simple explanation: the amount of attention concentrated by a galaxy might be primarily driven by its size.
To test this hypothesis, we now examine how total interaction strength evolves with galaxy size by plotting total attention as a function of the number of channels.
This plot reveals a clear overall trend: larger galaxies tend to concentrate more total attention. This confirms that galaxy size plays a role in shaping the distribution of attention.
However, the relationship is highly irregular. Galaxies of comparable size can exhibit vastly different levels of total interaction strength, and several strong deviations from any smooth growth pattern are visible. As a result, size alone is insufficient to explain the observed disparities in attention concentration.
Looking more closely at galaxies with abnormally high total interaction strength relative to their size, a clear pattern emerges: they also exhibit a very high average degree. For example, Community 17, which is comparable in size to Communities 0 and 9, stands out with an average degree close to 93, versus 50–60 for the others, and correspondingly much higher total strength. Similarly, Community 20 achieves a higher total strength than larger galaxies despite having nearly 1,000 fewer channels, driven by its exceptionally high average degree (≈ 81). This indicates that connectivity density, rather than size alone, is a key driver of attention accumulation. A higher average degree creates more interaction pathways, allowing attention to circulate and amplify more efficiently, and explains why the size–attention relationship appears highly irregular.
Charting the Future of the YouNiverse
Our voyage through the YouNiverse has revealed a structured cosmos hidden beneath the surface of YouTube's endless content stream. We've mapped galaxies bound by shared audiences, identified echo chambers and bridge channels, and exposed the black holes that shape attention flows.
As the platform continues to evolve, so will its cosmic structure. New galaxies will form, old ones will merge or fade, and the forces of attraction will reshape the landscape in ways we can only begin to imagine.
The YouNiverse awaits its next explorers. 🚀