There is a popular twitter API endpoint which is undocumented and unsupported, yet insanely popular due to its functionality. It's the tweet-count-for-URL one; ex: https://cdn.api.twitter.com/1/urls/count.json?url=http://www...
Twitter is now cancelling that endpoint (on 11/20) and recommending for thousands of developers to switch to their streams API for the functionality, which requires a program to maintain a constant socket connection and listen for filtered URLs to show up. Which, sure, that's a reasonable architecture for some cases.
However, what used to be a simple API-query-for-a-known-URL is now going to be a daemon program running 24/7 for months and months that listens to all of twitter for all conversations about the base domain, and then tries to emulate a count.