Take, for example, the following:
$\ln(x-1)^2 = 4$
Given the rules of logarithms, we have two potential first steps. In one, we could move the exponent within the argument of the logarithm to be a coefficient:
$\begin{aligned} \ln(x-1)^2 &= 4 \\ 2\ln(x-1) &= 4 \\ \ln(x-1) &= 2 \\ e^{\ln(x-1)} &= e^2 \\ x-1 &= e^2 \\ x &= 1+e^2 \\ \end{aligned}$
Or, in another, we could first raise the base to each side, resulting, in this case, in:
$\begin{aligned} \ln(x-1)^2 &= 4 \\ e^{\ln(x-1)^2} &= e^4 \\ (x-1)^2 &= e^4 \\ x-1 &= \pm e^2 \\ x &= 1 \pm e^2 \\ \end{aligned}$
For each of these initial steps (as written, at least), the end results are slightly different. In the first case, carrying through to the end yields $\ x = 1 + e^2$, while the second yields two solutions: $\ x = 1 \pm e^2$.
The question is: How do we avoid missing one of the two solutions simply by virtue of our chosen order of solution steps? I.e., what am I missing here?
You can avoid losing solutions if you take care of preserving the domain of the original equation, i.e. using $ln(x-1)^2=2\cdot ln|x-1|$