↓
sketched by friend shweta, supposed to be me apparently
Select a theme for the best reading experience
Last updated: February 10, 2026 • 5 min read
A friend brought this up - "because nowadays everyone has an internet connection, the competition is low." We're both into competitive shooters so this kind of thing comes up.
I didn't immediately disagree. There's a supply/demand logic to it - more people flooding in, average quality drops, things get diluted. I can see why it feels true. Had to actually sit and think about it for a bit.
The more I thought about it, the more it seemed like the logic runs the other way. More participants means:
Doesn't matter if you're Bronze or pro - same thing applies.
That said, I get where the intuition comes from. Back when fewer people played, being above average in your local scene felt impressive. You were the guy. Now matchmaking throws you into the global pool and suddenly you're mid at best. That's not lower competition though - that's just finding out what the real competition was all along.
Old-school LAN tournaments and regional scenes probably did feel more intense - everyone who showed up was dedicated, no casuals. But that's concentration, not level. The ceiling is way higher now.
I wanted to see if my intuition holds up statistically. Rough idea - skill-relevant traits (reaction time, spatial reasoning, game sense, whatever) tend to follow normal distributions in the population. So if you add more samples from that distribution, you should get more competition at every level. Even if the existing playerbase is skewed (early adopters being more dedicated), mass adoption should push things toward normal anyway.
Asked Claude to help me formalize this properly:
Player skill \(S\) as a function of \(k\) parameters:
\[S = f(X_1, X_2, ..., X_k)\]
Each \(X_i\) is something like reaction time, spatial reasoning, strategic thinking - the things that actually matter for being good.
Each of these follows a normal distribution:
\[X_i \sim \mathcal{N}(\mu_i, \sigma_i^2)\]
If \(S\) is a linear combination of these:
\[S = \sum_{i=1}^{k} w_i X_i\]
Then skill itself is normally distributed:
\[S \sim \mathcal{N}\left(\sum_{i=1}^{k} w_i \mu_i, \sum_{i=1}^{k} w_i^2 \sigma_i^2\right)\]
Caveat on the linear assumption: The model above assumes skill is a weighted sum of independent traits - \(S = \sum w_i X_i\). In practice, competitive-game skill compounds non-linearly. A player with slightly better reaction time and slightly better game sense doesn't just get the sum of those edges - the reaction time lets them exploit openings that only the game sense could identify. Neither edge is worth much alone; together they cascade. The real relationship is closer to multiplicative (\(S \approx \prod X_i^{w_i}\)) or at least has significant interaction terms, which produces heavier tails than a Gaussian.
This actually strengthens the argument rather than weakening it. Heavier tails mean more density at extreme skill levels than the normal distribution predicts. When you increase population \(n\), you're sampling more from those fat tails - the skill ceiling rises faster than \(\sigma\sqrt{2 \ln n}\), and competition at the top intensifies more than the linear model suggests. The conclusion "more people = more competition everywhere" isn't just directionally correct under non-linearity - it's stronger, especially at the top end.
For \(n\) players drawn from \(S \sim \mathcal{N}(\mu_S, \sigma_S^2)\), the expected skill ceiling is:
\[\mathbb{E}[\max(S_1, ..., S_n)] \approx \mu_S + \sigma_S \cdot \Phi^{-1}\left(\frac{n}{n+1}\right)\]
As \(n \to \infty\), this grows like \(\sigma_S \sqrt{2 \ln n}\).
So skill ceiling goes up logarithmically with population. More people = higher peaks.
Number of players at or above percentile \(p\):
\[N_p = n \cdot (1 - p)\]
Double the playerbase, double the number of people you're competing against for any rank. Top 1% of 10 million is way harder than top 1% of 100,000.
Even if the original playerbase was right-skewed (dedicated early adopters), adding random samples from the general population:
\[\frac{\partial(\text{competition})}{\partial n} > 0\]
For the original claim to work, you'd need \(\frac{\partial(\text{competition})}{\partial n} < 0\). But it doesn't.
"More people = diluted quality" makes sense for some things. But competition isn't about average quality - it's about how many people you have to beat at your level. That number only goes up.
Look at pro scenes now - Valorant, CS2, Apex - they're more stacked than ever. The gap between casual and Radiant/Global Elite/Predator is insane. Internet access didn't lower competition. It just showed everyone where they actually stand.