Yes, I often have done such refactoring. It's hard to get that last 20% of speed, much less last 5% without making a large number of small changes in code, usually after getting accumulated trace results (however presented.) .1% = one in one thousand. Keep stacking up those improvements and you do get somewhere, and often you'll easily find one hundred very small improvements in a day (10% altogether!) if it's green code, along with the bigger inefficiencies to fix. Not to mention that ignoring a presently-small inefficiency is predicting that future use won't swell that into a problem - which can be most unwise. Also, looking for tiny improvements gives you a good chance to stumble upon larger improvements you'd missed earlier. Optimize much?

How much of the economy uses Bitcoin, though? That's the denominator to use.

Do explain.