I think that if you want google to be able to index the article text, then the article has to be accessible from the search result - you can't show different pages to google's robot than to human visitors or they'll block you. So the paywall is not allowed to kick in if you are coming directly from google. (Works for the NYT as well, probably anywhere else that is indexed on google too).

http includes a referer field that tells a webserver what page linked you to it. It is therefore possible for the server to disable the paywall for visitors who arrive through google.