Delta can't just be epsilon because then things wouldn't work.
You want to be able to prove that whenever [MATH]\left|x - \frac{1}{4}\right| < \delta[/MATH], then [MATH]\left|8x - 2\right| < \epsilon[/MATH]. But if you took delta to be epsilon, then if [MATH]\left|x - \frac{1}{4}\right| < \epsilon[/MATH], it would only follow that [MATH]\left|8x - 2\right| = \left|8\left(x - \frac{1}{4}\right)\right| = 8\left|x - \frac{1}{4}\right| < 8\epsilon[/MATH]. That lets it be too big.
I think introductory examples of these proofs may too often omit this step of stating clearly just what has been shown, because it is obvious to us who write them! The process we show is how to find a delta that will work, and hidden in that process is the fact that it does in fact work. But the solution you show does exactly this, in the last three lines. They are showing there that this choice of delta does what we need it to do. (On the other hand, you could choose any value of delta less than epsilon/8, and it would still work. We only need to find a delta that works, not the only, or the best, delta.)
By the way, you should try to type in what you can rather than just use images, to help others quote you and interact more easily. When there are special characters, you can either paste them in from some source such as Windows Character Map, or use the calculator icon in the tool bar to surround "\delta" or "\epsilon", which is the simplest use of the LaTeX formatting system. You don't need to do all the fancy stuff I did above, but a little bit helps a lot.