He's just standing there either for ceremonial reasons or for training (possibly both). In any case, his only task is to stand there and call the police if there's trouble.
An attack on the guard himself is likely to lead to a resolute counterattack.
Soldiers guarding typically have limited police jurisdiction of the area they guard in most countries.
This should be common knowledge.
Your point of view is not always correct, so it should definitely not be common knowledge.
Local mileage and laws may vary, but erring on the side of caution when you see a soldier means that it's prudent to act as if he can legally fire upon you -- even if that assumption is not actually true at that specific time or place.
It's this sort of blame-driven, individual-focused, ask-the-unachieveable answer that makes it completely impossible for organizations to move beyond a relatively low level of quality/competence. It's satisfying to say, because it can always be applied and always makes the speaker feel smart/superior. But its universal applicability is a hint that it's not going to actually solve many problems.
If you'd like to learn why and what the alternative is, I strongly recommend Sidney Dekker's "Field Guide to Understanding Human Error":
His field of study is commercial airline accident review, so all the examples are about airplane crashes. But the important lessons are mostly about how to think about error and what sort of culture creates actual safety. The lessons are very much applicable in software. And given our perennially terrible bug rates, I'd love to see our thinking change on this.
Keeping one's finger off of the trigger is not unachievable; indeed, every infantry organisations instils that from the beginning in its riflemen. It's really Not That Hard. And indeed, one sees that negligent discharges are actually pretty rare, even in war zones.
That approach quickly plateaus. Then more discipline harms improvement because people just start covering up mistakes to avoid punishment. In addition to being covered in the book on air safety I already mentioned, your "beatings will continue" theory is basically why Detroit spent decades trying and failing to make cars as reliable as Toyota's. This was true even when Toyota went to great lengths to teach GM how to do it.
That experience was covered engagingly in This American Life's episode NUMMI:
Systematic measures to reduce problems are wise.
Carrying a weapon implies the possibility that it will need to be used at a moment's notice, and it's a hell of a lot quicker to flip a safety off than reload. These measures could wind up with someone dead.
I agree with the GP - if accidental discharges where that much a problem, then the problem is poor firearms discipline, something that can be trained and drilled into people. These are better called NDs (negligent discharges), and they're called "negligent" for a reason: the shooter was doing something horribly wrong.
NDs will get you kicked out of recreational ranges in this country at a minimum. It can also be a ticketable or jailable offense.
Firearms discipline 101 basically breaks down to keeping your finger out of the trigger guard until you intend on shooting something, and possibly the weapon on safe (depending on situation).
Hardly unreasonable or "unachievable" as another put it.
If the guards are incapable of this one simple thing, something that recreational shooters are held to a high standard on, they should be replaced with people who can.
These positions are largely ceremonial, an incident requiring them to use their weapon will happen less than once a decade.
Also, Recreational shooters aren't holding a rifle 8 hours a day every day for years at a time. The increased amount of time combined with the monotony of the job dramatically increases the risk of mistakes.
Ultimately the odds of there being an incident which requires the guard to be able to start firing immediately is less than the odds of accidental discharge.