I've spent too many hours of my life improving code. Then come to find out my improvements were worthless because of some insane externality that I would have never considered.

For example, I just spent a week improving the loading speed of this data processing engine from like 1.5MB/s to over a GB/s (using an artificial benchmark that I wrote). I put things into certification and saw the loading speeds were only like 2.0MB/s. After some investigating, I discovered I was hitting the limit of the system that was serving the data.

I would have never thought that, in 2017, there would be systems in place that couldn't serve a single file faster than 2.0MB/s.

If you are like me, don't forget to review that code that the moron that was you (or me) wrote 10 years ago. :)