[austin-cheney]

The article does not provide numbers. The reason why local development is important is because of speed, like the article says, but the article supremely under sells this. We could easily be talking about several orders of magnitude performance improvements.

When I first joined travel company Orbitz they had a build that took just over 90 minutes. Of course it was a Java first environment, so I needed to test a small one line change to the UI (that had absolutely to do with Java) I still had to wait 90 minutes. So, I just planned on doing nothing all day.

In my personal software I start crying if my builds take longer than 12 seconds. The difference is that I really enjoyed getting paid to watch YouTube for 90 minutes stretches 5 or 6 times a day. It wasn't my time. It was their time. With my own software, though, it absolutely is my time.

Small improvement increases to trivial items is a cost savings that adds up. Its like a flash flood. A rain drop isn't going to drown you. A bunch of rain drops are still insignificant, but when there are too many rain drops you are under water.

Just think about how floods work, because its still just tiny rain drops. One increase does almost nothing on its own, but it enables other things to occur more frequently. When that happens everywhere you have a performance flood. Suddenly you can test automate several hundred features of your application end-to-end in less than 10 seconds. When that does occur it changes peoples behavior and their perceptions of risk, because now all options are discoverable in a nearly cost free manner for everyone. You will never ever get to experience that freedom if you are drowning in dependency and framework stupidity.