> …when you’re in a position to help

This clause is doing a lot of heavy lifting. One needs to have good judgement about when and how to help. A lot of people can imagine how things could go better if a bunch of other people changed their behavior in surprisingly simple ways. It's a much smaller subset of people that can correctly push the right buttons to get the other people to actually make those changes succeed at a systematic level.

In a small org it's actually not too hard for good ideas and feedback to get traction. In a larger org for broad concerns it can be fiendeshly difficult. Often the reason why a large project will fail is only truly knowable by a few senior technical people with enough experience and broad context to see the forrest for the trees. Past a certain volume of people involved you can not explain to people why it will fail fast enough to offset the army of clueless stakeholders incentivized to socialize a good-sounding narrative convincing everyone that we need to try. In these cases reductive explanations with the right counter-narrative can work, but they require significant reputational and/or hard authority to pull off.

This is why the article advocates picking your battles in a large org. Often the chance of actually helping is much lower than destroying your own reputation, even if you're right.