Benchmark that spits out a nested mess of non-reference checkable data. Tests libraries ability to deal with fud.
Most high scoring libraries wrap their update routine in a time delayed context like a microtask or animation frame. I've followed suite.
Run it here.