On his book "The Secrets of Consulting", Gerald Weinberg advise against improving more than 10% of performance, and if so, of hoping to have any credit.
Just like the article, his reasoning is that if you improve performance too much, it makes management/the team look bad for not doing it before, while a smaller improvement in performance make management looks good.
I'm only just realizing now why I was treated so strangely after discovering a pretty severe security issue that had been in our software for about 7 years. I had only been in the team for about 12 months when I discovered it.
The only time I have seen a good response from the discovery of a large security hole was when it was by total accident, and the person mentioned it in a public channel by accident. They almost got fired, but the bug got fixed quickly.
Just like the article, his reasoning is that if you improve performance too much, it makes management/the team look bad for not doing it before, while a smaller improvement in performance make management looks good.