Extremely mild punishments for academic misconduct

After looking at My professor is rigging data and plagiarizing. What can I do?, I read through some of the case reports of the Committee on Publication Ethics (COPE) and I noticed that punishments seem to be very light. When COPE receives a case, sometimes an attempt is made to contact the institution, and then the institution does nothing: perhaps the wrongdoer is too politically powerful, or the institution doesn't have any will to act. Sometimes the investigation simply fizzles out after nobody bothers to look at the lab journals, even though the lab journals presumably exist. Sometimes the journal is told how to avoid such problems in the future, but no punishments are forthcoming for the existing case.

This is very different from what I believed before - that if someone did decide to falsify/steal data, their employment would be terminated, and nobody would ever want to associate with them again.

So my question is: where is the discrepancy between how I perceive the consequences of academic misconduct, and what COPE reports as actual cases? Is it perhaps because most of the misconduct cases are from no-name departments and journals that do not care about their reputation? Or are my impressions of the consequences of misconduct incorrect?

In fact, I have experienced this on my own: one non-mathematics professor at my school took my mathematics manuscript, added his graduate students as coauthors despite their having no relation to the work, and tried to submit it to a journal. I don't care about this specific publication, as it is so worthless that I prefer it to be rejected as to not be associated with me. The professor thinks it is significant only because he knows nothing of mathematics. But it does make me question the kind of integrity I assumed of academics. He is a chair professor at a top10 university, and his (non-mathematics, science) department is very well-respected, so brand name is not the cause here.