> 1% is significant if it trains long term behaviour. Or maybe it can be added to other effects, or maybe the techniques can be refined from here. You're abso

> 1% is significant if it trains long term behaviour. Or maybe it can be added to other effects, or maybe the techniques can be refined from here. You're absolutely right there are a dozen reasons small effects can still be important. Sometimes it's just p-hacking though, or otherwise unreplicable studies. That's worth considering too. I think we'd both welcome more research, maybe especially over longer terms.