(the NY subway ticket system is similarly archaic!)
A significant part of this is differences in credit card usage patterns. It's fairly common in the US for people to have a large number of credit cards (think ~10), not least because various stores have their own cards which get you discounts. From what I can tell, other places people have a much smaller number of cards.
If a person has 10 cards, and each of those card issuers wants their card to actually be used, the game theory is against them making the person try to remember a PIN to use their card. Even if you posit that they can solve the coordination problem and all agree that they will simul-roll-out a requirement to use a PIN (which is not trivial at all, given lack of widespread support in terminals), they all worry that with 10 PINs to remember the cardholder will just say "screw it" and stick to using only one or two cards, and it won't be theirs.
Yes, the cardholder can go through the trouble of changing PINs so they all match. But then you have to gamble on them doing that...
So that's one reason issuers haven't been pushing very hard for chip+PIN.
Next, credit card fraud in the US is currently limited through various Orwellian big data measures, whereas Europe was rife with credit card fraud when EMV was originally deployed, so the cost benefit analysis made it much more appealing in Europe.
Basically, path dependency is a thing. Americans routinely use checks to pay rent for similar reasons.
Physical credit card theft is a negligible source of fraud, it’s all about card cloning, which EMV does a moderately good job of preventing.
Prior to that event, I can’t think of any transaction we have had in decades here in the US where we were required to pay by check. Cards or electronic payments of one sort or another have almost always been available to us.