[majormajor]

> In software, you make a typo and you get an error immediately. You fix it and go on.

This level of thing is usually silly, ultimately-meaningless errors. (The rare exception being the metric-vs-imperial type things that literally blow up eventually.)

There are many other errors we make every day building software that lead to years or decades of tech debt. The thing we need to figure out is how to get better at making better up-front decisions on that. Being better at monitoring the long-term feedback produced by our designs, so that instead of saying "we should rewrite this in a new framework" we can say "we should restructure this doing X and Y to prevent these types of problems from coming back."

[toast0]

The costs of making better decisions need to be measured against the costs of making poor decisions. Most (but certainly not all) software is ephemeral. If you knew what would be in use in ten years and what would be in the bit bucket, you might make different choices. If you were liable for your choices and mistakes for years and years after your product shipped, you would examine them closer as well.