Comment by eli

I agree that seems presumptive. But of course twitter conveniently didn’t share the source code for how that factor is calculated. Turns out it’s hard to evaluate a big complex distributed system by looking at just one slice and no data.