Useful feedback describes your experience of the work rather than issuing a verdict. Say where your attention dropped, what you expected that did not arrive, and what you were unsure of. Point at the work, never the person. Offer diagnosis before prescription, and be specific enough that the note could not apply to anything else.
Yes — when it arrives too late to act on, when it addresses a constraint the author cannot change, or when it is a verdict with no observation behind it. Accuracy is necessary but not sufficient for feedback to be useful.
Three substantive points is usually the practical ceiling. Beyond that people triage by what feels easiest rather than what matters most, so a long list quietly buries your important note among the minor ones.
Say that you skimmed, and where you started skimming — that is genuinely valuable information. What is not useful is skimming and then commenting as though you read it closely.