I like to think of analysis as the laziest field of math, because in the end (the limit), it's all the same.
Since you just need terms to go to 0 eventually, you end up seeing expressions like $$\epsilon = \frac{\epsilon}{3}$$ all the time.
You can really fudge the error terms in a proof, since in the end it doesn't matter anyway.