Exactly, it's like sifting a mountain of ore looking for diamonds. The original Korean team has essentially been doing a slightly guided brute force search in a domain that they thought was promising doing 100's or even 1000's of experiments. It's a clever approach: you have a real problem because the parameter space is so large but if you get 'close enough' to pick up an anomaly that might help you home in on something that is really promising. I'd love to see their notebooks and everything they tried. This is pretty much how we first got electric light: endless failures and then one success.