Roast my LinkedIn

The number at the top is one thing only, the share of that account's recent posts that were not worth a stranger's time. Pass counts fully, Skim counts half, Read counts nothing. A 20% account writes mostly for readers. An 80% account writes mostly for the algorithm.

Worth reading does not only mean useful. A post that is genuinely funny counts. So does a real story, a real opinion with a reason attached, or one specific result. What does not count is anything ten thousand other accounts could have posted unchanged.

Every post gets the same short list of yes or no questions. Does it say something specific. Does it say something a regular reader has not seen a hundred times. Is it asking for a reply instead of earning one. Is it selling. Is it empty words. Then a six-level quality read, from pure filler to the kind of post that breaks out.

Skim is not a soft Pass. It is the honest answer when a post is short, fine and forgettable. Pass is reserved for bait, pitches, platitudes and posts with no idea in them.

Engagement bait is the big one. Any post whose real job is to collect a reply rather than to say something. Comment X to get Y, agree?, repost if you, a question with nothing behind it.

A promo is a post that is there to sell, launch or hire. One a week is a person with a job. Ten in forty is a billboard. Empty words are advice or motivation anyone could have written. Chatbot vocabulary is the delve, leverage and tapestry layer, underlined in every post so you can see exactly where it crept in.

Reposts and replies are skipped, not scored. A repost is someone else doing the talking and a reply out of its thread cannot be judged fairly.

One more judgment is about texture, not authorship. Does the post read like a specific person wrote it, or like a chatbot draft nobody read back: a templated hook, three parallel beats, smooth balanced sentences, generic claims, and no name, number, place or lived moment anywhere. That grade is blended with how dense the chatbot vocabulary is.

Above the line, the post is labelled "reads like unedited AI" and lands in Pass. The line was set by scoring human posts from before chat assistants existed against our own generator with the guardrails off, and putting it where the two sets split. A person can write like this and a model can write well, so the label is about the text, never about who typed it.

It reads the words in recent posts. That is the whole input. It does not see follower counts, likes or who you are, and it never claims to know whether a person or a model wrote a post. A human can write slop. A model can write a good post. The score is about the reader's time, not the author's tools.

It only sees the most recent forty posts, so a good week and a bad week can give the same account two different numbers. Roast again in a month.