Stocks -- or, more accurately, investors -- always react to quarterly results. Analysts keep detailed models of their expectations, and the average of these models is known as the “consensus estimate.” If you miss consensus, as Google did, you get punished.This is exactly wrong. There's no punishing going on. The problem is that most people think that a stock trades based on some absolute reality of the company's "value" right up until earnings are announced, and then it goes all silly. Instead, there is no real value for the company. It's set by consensus, based on what lots of different people think the value of the company is (along with supply and demand and other factors, but I'll arm wave those a bit, here). When earnings are about to come out, the stock's value has already been trading based on those estimates for some time. If the earnings report shows that the company's profits went up, that doesn't really matter. What matters is, how well is the company doing, relative to the estimates that have already set the price.
For example, let's say that Google is trading at $500/share. Analysts look at the company and determine that it's worth $600/share based on what they're expected to bring in this quarter. The stock moves up to $600/share. Now earnings come out and it turns out that the company is doing 10% better than they were the previous quarter. 10% of $500 is $50, so the fair price for the stock is $550, but it's been trading at $600 because we all thought it was going to be worth more than it was. We're not punishing the company by lowering the price to $550. If anything, we were briefly giving them the benefit of the doubt and trading their stock at a level higher than their previous earnings report justified. This new change is just a correction back to non-speculative levels.
So, the next time you hear, "you can make more money, but if you miss estimates you'll be punished," you'll know it's complete nonsense.