You're missing the point here. The causation is different, so a direct comparison is invalid. An injury caused by a mechanical failure on a job site is different (accidental) than a criminal who decides to hurt a cop (intentional). The end result is the same, but the causation is different. Someone makes the decision to hurt a cop (The recent PA trooper ambush, for example), as opposed to your example of a steel beam hitting someone on the head. The beam has no intent of causing injury. Intent is everything, and (in NY, anyways) the basis between similar criminal acts, ie: criminal trespass to burglary, manslaughter to murder, etc.Person A gets killed, while doing their job, by a steel beam tearing loose and falling on their head.
Person B gets killed, while doing their job, due to criminal attack.
In regards to danger they're the exact same spectrum.
Things are just as dangerous as the statistics say they are.
You want to argue that folks doing a civil service deserve more praise for doing an equivalently dangerous job? Sure, I can see that. Want to argue that folks with an identical death/injury rate are doing a "more dangerous job" because it's for the public benefit? You're conflating the "danger" of the job with the choice to engage in civil service.
No matter what the cause - if the rate of death/injury has gone down, the job is less dangerous by definition.
When new training among loggers in log handling and safety equipment usage reduces death/injury rates - logging gets less dangerous.
OSHA was created to make the workplace safer, and it's worked. Maybe they should start a campaign telling criminals to stop trying to hurt cops, to therefore make their work environment safer. Sounds ridiculous, because it is.
A lot of people's perceptions of law enforcement (good and bad) have no doubt been influenced by personal experience. A lot of other people's perception has no doubt been based solely from biased internet babble and news media accounts. Their whole perception is based not on experience, but word of mouth.