They do when there is good reason, but there isn't always good reason.
In fact, a theoretically better standard could actually be worse in some situations if people are accustomed to a different standard. Imagine if America had settled on red for go and green for stop, and then tried to change the standard to match the rest of the world. It would be a calamity as some drivers continued to adhere to the old standard that they knew.
Stop signs were originally yellow, so such a change isn't unprecedented, but in that case they also contained the word STOP so the colour wasn't critical to understanding the sign. Standards can more easily change when there is backwards compatibility available. More recently, some jurisdictions have started adopting traffic light shapes (square = stop, diamond = caution, circle = go) but retain coloured lights for backwards compatibility.
Like, the WALK / DONT WALK crossing signals that were around in my youth were slowly replaced with the HAND / WALKING PERSON ones, I'm sure because they're better for people who don't read English.
But also there is a cost to change, and it's often not worth doing if the benefit from the new system or standard isn't a lot better than the old.
When ATMs were first implemented they would dispense money before or at the same time as giving your card back.
This resulted in premature conclusion errors. They went to the ATM to get money. They got their money so they left... forgetting their card.
When ATMs were updated they fixed the design error. Now the card popped out and had to be removed before money would dispense. This resulted in a spike of people leaving with money hanging out of the ATM because they had been trained that removing their card was the end of the task. (Which is why money now gets sucked back into the machine if it hasn't been removed fast enough).
The point of the anecdote (other than that you hire HCI experts before implementing an interface) is that implementing a new and objectively better system doesn't necessarily result in an objectively better outcome when replacing an incumbent worse system.
Indeed, sometimes they do and sometimes they don't.
Why doesn't the rest of the world (outside the US) have a 3rd middle brake light (CHMSL), as was mandated in the US in 1986?
My guess is because, despite measurements of ~22% crash avoidance at the time, that number turned out to be between 0% and 4% in practice. Maybe the rest of the world didn't care about 4%, or maybe they thought it was as stupid then as it turned out to be.
But the CHMSL standard in the US has refused to die.
In fact, a theoretically better standard could actually be worse in some situations if people are accustomed to a different standard. Imagine if America had settled on red for go and green for stop, and then tried to change the standard to match the rest of the world. It would be a calamity as some drivers continued to adhere to the old standard that they knew.
Stop signs were originally yellow, so such a change isn't unprecedented, but in that case they also contained the word STOP so the colour wasn't critical to understanding the sign. Standards can more easily change when there is backwards compatibility available. More recently, some jurisdictions have started adopting traffic light shapes (square = stop, diamond = caution, circle = go) but retain coloured lights for backwards compatibility.