Title is misleading. The article basically covers specific cases where modals aren't a great solution, but the title suggests they are NEVER useful. Clickbaity if you ask me.
It doesn't suggest that they're never useful. It says you that you shouldn't use them — i.e. that using them is bad.
There are lots of things that you generally shouldn't do, but sometimes must do under certain circumstances. Doing something you shouldn't because you must, doesn't excuse you from the fact that you're doing something "bad"; it just means that "doing something bad" is the best/only option you have at that point.
Compare/contrast: jerking a dog's collar to pull it back from running in front of a car. You shouldn't abuse animals; but in that specific case, you "must" do something that will hurt the dog a small amount, to prevent it from experiencing much greater harm. Doesn't mean you're not doing something bad, though! It'd be better to have planned ahead so you wouldn't be in the situation where you'd have to jerk the dog's collar in the first place.
You shouldn't use modals. Sometimes you end up in a situation where you must. That doesn't mean they're "not bad" in that situation; they're still bad! And as such, it'd be better to plan your UX, so that you don't end up in a situation where modals are the best/only solution to your problem. :)
A general "should" statement in English — e.g. "you shouldn't use modals" — recommends (assigns utility weight to) a concept; it's not an imperative command to do (or not do) that thing, right now, in the specific situation you're in.
This is in contrast to "it would be best to do X" when used in equivalent contexts. While "you should" conveys an absolute judgement on the general / global / context-free utility of an action, "it would be best to" conveys a relative utility-ranking of actions presented as (perhaps implicit) options.
Something can be the best option in some scenarios (= relative contextual ranking), but still a bad option in global static ranking (= absolute judgement), and so would be something you "shouldn't" do.
You shouldn't climb over an electrified fence. You might have to — it might be best to do so — if you're being chased by a bear. But it's still anti-recommended as a course of action. You should avoid any series of choices that results in you having to climb an electrified fence, because even when it's the best course of action, any future where you have chosen to climb an electrified fence, is going to suck compared to one where you didn't have to do so.
You should brush your teeth before sleeping. In the situation where you're hiding up a tree from the aforementioned bear, it would be best to not do so. But that doesn't remove the general recommendation that you "should" be brushing your teeth; and doesn't remove the twinge of regret you get for not being able to do the thing you "should" be doing.
(I think you might be confusing "should" for the similar word "shall" — a thing you must do but shall not do would indeed be semantic nonsense.)
Alternately, if your argument is that you need the words "in general" in the statement to make it true — no, you don't. "In general" is the default context for "should" statements, and especially ones not addressed to a specific speaker. If a "should" statement is made without reference to/knowledge of you or your specific context, how could that statement possibly be in reference to what you specifically should do in your situation — vs. being a generally-agreed-upon context-free absolute value-judgement of the concept?
I’m not confused about what any of these words mean.
> You shouldn't climb over an electrified fence. You might have to — it might be best to do so — if you're being chased by a bear.
And in that case you should climb over the electric fence. It’s nonsense to say that you “must” or “have to” climb over the electric fence but that you “should not.” The first sentence is, as is abundantly clear to all English speakers, a shorthand for “you shouldn’t climb over an electric fence except for very rare cases where it will be obvious to you that you’re encountering an exception.”
Yep, and it got me. I clicked thinking "how can they say you shouldn't use modals when there are definitely great use cases for them?" and of course, they're not saying that at all.
The editorialized title by @flothebre you mean, as @KerrAvon points out.. that's not the current title (nor possibly the prior one based on slug, it looks like the article was last updated July/22)
