web applications - How do Facebook and have these amazing features?

i'm interested in learning how these 2 things accomplished:

facebook: when user logged in , sitting @ news feed, new items automatically show in news feed, red badges friend requests, messages, , notifications automatically appear, , instant messaging live. that's lot of database activity! on top of that, users have ability remotely end other sessions, of sessions have checking if they're still valid. many users, how work without killing servers?

stack overflow: there 60+ badges on stack overflow based on wide variety of actions user does. seems extremely complicated check virtually of user's actions see if earned them badge or not. same question: many users, how work without killing servers?

i appreciate guidance.

 planning:  

first step in process planning. facebook or stackoverflow cannot create badge today , 2 months down line, find problem. detailed planning done create category. can write long here, need tech; lets move ahead.

 object oriented:  

everything object oriented in these applications. take example careers.stackoverflow. ever related careers moving around careers. huge set of careers, resumes, present users of stackoverflow data present there. take example badges of users, they cannot make obeject, cannot used careers sub domain. create such badges object used @ different applications.

i promoted careers site of stackoverflow. :) 

.

database: 

when talk database, procedures, triggers, etc come account. maintaining database not enough. if 1 event occurs, 1 may required trigger database query pull useful data event. lets answering same question. while still tying, stackoverflow throws orange message bar on top stating "1 new answer has been posted". happens when user has posted answer while type. triggers used here.

ajax:  

when want small portion of page loaded randomly.

jquery, ext js:  

used nice javascript , presentation logic.

servers:  

within server create different node instances, farms @ different locations. route traffic accordingly different nodes on server. monitoring taking lot of memory. accordingly allocate new nodes, instances such traffic. use different farms maintain database. if 1 crashes @ la, pull data ny , bring both same platform.

r&d: 

last thing r&d, without cannot keep pace whats new in market. have upto new techs , developments.

hope answer question.