Claude
If someone has 80,000 followers on Twitter, their account is nearly useless to us.
Not because they're a bad person or even a bad candidate. But because the account has been optimized for something other than honest self-expression. At some point — usually around 5,000 followers — people realize they have an audience, and they start performing for it. Posts get more polished. Opinions get safer or more provocative, depending on which direction drives engagement. The raw signal degrades as the curation improves.
What we actually look for is the opposite: accounts where posting behavior is clearly not strategic. Infrequent and technical. Lots of replies to niche conversations. Engagement so low that no reasonable person would keep going unless they genuinely cared about what they were saying. A thread from 2021 that got four likes but turned out to be completely correct.
These accounts are harder to find and harder to read. That's the point. Nobody has been filtering them for palatability. You're seeing what the person actually thinks, how they actually communicate, what they actually spend time on.
The influencer account tells you who someone wants to be seen as. The quiet account tells you who they are.
This has implications for how we weight signals. Engagement metrics — which most tools treat as a quality signal — can actively mislead. A post with 10,000 likes might be optimized content with no original thought. A post with 12 likes might be genuinely important, just written for an audience of twelve people who would get it.
CloudScout tries to read past the performance layer. Not because big accounts are bad, but because the signal density is just much higher in the accounts nobody is watching.