I mean...it's not that surprising that an article from 16 years ago is out of date, right?
FizzBuzz was a low bar, but that's the point. You're right; you can't assert that anyone who clears that low bar is a good developer, but I don't think anyone ever claimed that. What you can do is assert that someone who can't clear the bar might not be worth the extra effort required to pull them through the hiring pipeline. If you were trying to hire a software developer/programmer and they couldn't even begin to solve FizzBuzz, would you hire them?
FizzBuzz is the "fairly simple algebra problem" from the article you prefer, because - in Spolsky's words:
"At one end of the scale, there are the unwashed masses, lacking even the most basic skills for this job."
"String compression", "valid anagram", and "valid parentheses" fill the same role as FizzBuzz today, but maybe aren't so played out? I don't think they're necessarily any harder, just that they haven't been talked about, publicly scrutinised, and blogged to death.
What does make them slightly better as coding-interview questions though - in my opinion - is that they provide much better jumping-off points to ask a candidate questions about their implementation: how it could be improved, how well it could scale, etc., giving the candidate opportunities to elevate themselves and demonstrate their knowledge/experience; giving them room to differentiate themselves from the candidates who completed the leetcode/cracking the coding interview grind on that particular problem and just moved on.