How to Raise Millions for $100 — The Lucrative Shadow Economy of Fake GitHub Stars

A new investigation just dropped, and it’s deliciously cynical: researchers scanned more than 18,000 repositories and uncovered roughly 6 million fake stars. The undisputed champions? AI startups (the non-malicious kind, at least).

At first glance, who cares? Stars are just digital likes, right?

Wrong.

Venture capitalists are obsessed with them. A partner at Redpoint Ventures openly admitted that many funds run scrapers across GitHub every day. The #1 filter? Star count. On average, you need ~3,000 stars to get a serious look for a seed round and ~5,000 for Series A. It’s not the only signal, but it’s a brutal gatekeeper.

And once something becomes a gatekeeper, a market appears.

One star costs between $0.03 and $0.85 depending on quality (fresh bot account vs aged profile with commit history and avatar).  
Want 3,000 stars for seed? That’s $90–$2,500.  
If you actually close a $2–5 million round, your ROI is absurd.

This is Goodhart’s Law in action: “When a measure becomes a target, it ceases to be a good measure.”

The authors of the report have a simple fix: stop worshipping stars. Look at forks instead.  
Forking requires real effort — someone actually cloned your repo and did something with it. Faking forks at scale is way more expensive. Even better: track the forks-to-stars ratio. Real traction shows up there. Pure hype evaporates.

This pattern is universal:

• Telegram → views (cheap) vs meaningful engagement (likes + comments + reposts);
• Mobile apps → installs vs Day-7 retention (DAU);
• SaaS → sign-ups vs paid conversions or 30-day usage.

Bots will always flood the cheapest metric. “Just detect them better” doesn’t work — detection always lags. The winning strategy is to move the goalpost to the metric that’s expensive to fake.

P.S.
We’ve seen this movie before. Higgsfield AI spent months farming fake blogger mentions and “organic” coverage.

The product was actually impressive, but the growth-hacking burned their reputation so badly they’re still scrubbing “Shitsfield AI” memes off the internet years later.

Meanwhile, entire Telegram channels exist that are 100% fake — bought followers, stolen content, botted views, comments, and shares — yet blue-chip advertisers happily pay top rates because the vanity metrics look perfect.

The economy of fake stars isn’t a GitHub problem.  
It’s a human problem: we love easy numbers, and the market will always sell us the easiest ones.

Want real traction? Make the signal expensive to fake.

Also read:

• AI Is Kryptonite for Critical Thinking — Here’s Exactly How It Melts Your Brain
• Startups Are Reviving Dead Brains — And Big Pharma Can’t Look Away
• The Battle for Humanity’s Last Dataset
• US Intelligence Agencies Secure $9 Billion for AI Chips and Access to Anthropic’s Mythos Model

Thank you!