No, that is not what the hospital does, and thus based on this heuristic, it is not its purpose. What a system does is not the same as the context-free outcome. It is the outcome compared to the outcome that could be expected without the system. You have to define your priors.
However, if the expected 5 year mortality for the cancer was 50%, and with this treatment 2/3 died, then the rule would apply. A choice to continue using that treatment could be criticized as equivalent to a choice to kill 1/6 more patients. Because despite the intention, the known outcome was more patients dying.
People often have trouble with this saying, and that trouble often boils down to the difference between intent and purpose.
The people who create a system have some intent for it. The system may or may not effectively achieve that intent, may or may not outlive the initial conditions that surrounded its creation, and may or may not have side effects.
Purpose is something humans assign. It is sometimes linked to intent. A carpenter's hammer is intended to drive and pull nails, and that is often also its purpose. The purpose of the hammer I keep in my basement is breaking open walnuts.
The phrase is stating that the purpose we should assign to systems when judging them is their outcome, and not the intent behind them.
Scammers can definitely get through it faster than you can. Whenever you attempt to address abuse in a system by increasing the complexity of that system, you implicitly bias it towards those with the time and inclination to study it, which always includes those with intent to abuse it, and generally does not include your users.
One thing that I think is demonstrated in the Safari "file" menu, the most common things you might need under that menu* have icons, which means you can scan to them very quickly.
* Note here that most common things you need under the menu are not the same as the most commonly used commands. New tab and new window are surely the most commonly used commands, but those you would almost never go to the menu for.
The real lesson here is that almost all modern SaaS applications have massively under invested in customer support in order to appear more profitable or sustainable than they really are. One of the major factors behind LLM development is trying to solve this problem before the house of cards falls down. Companies were enticed by the recurring revenue of SaaS, but don't want to pay for the level of support required when you are responsible for all your customers data as well as their access to the service.
I am actually not sure that this is an example of bias, at least not in the direction that you seem to be implying. Though I appreciate your strong connection to the subject, the purpose of the Wikipedia page for a topic is not to advocate, but to describe. I don't think it is very controversial to say that the term "feminism" has a more widespread common understanding than the term "men's rights." I empathize with the desire to have a place to put information about issues that affect men, and also with the frustration at being told that the correct place to put that information is under the heading of feminism. But I do not think it is unreasonable for the Wikipedia page on "men's rights" to discuss the various ways people use and understand the term, the history of its use, and criticisms.
The phrase does not mean that you can pick any single effect of a system and claim that is its purpose, as your linked article does in its examples. (Ironically, a form of reducto as absurdum.) It is a heuristic, a pattern of thought to attempt to overcome the bias towards judging systems based on the intentions behind them instead of the outcomes they produce. The point is that when you choose a course of action, you are implicitly choosing its negative effects as well, and the choice should be judged on all its effects. You are making a cost / benefit analysis, and if that is not explicit, it can easily be wrong.
However, if the expected 5 year mortality for the cancer was 50%, and with this treatment 2/3 died, then the rule would apply. A choice to continue using that treatment could be criticized as equivalent to a choice to kill 1/6 more patients. Because despite the intention, the known outcome was more patients dying.