I am looking for smart proofs of the inequality $$\sqrt[\alpha]{2} < \frac{\alpha}{\alpha-1} \qquad \text{for} \ \alpha> 1$$ this inequality comes from the comparison of the median and the mean of a $\mathrm{Pareto(\alpha)}$ distribution.
Here is the proof I found but I would like a list of different ways to approach the problem.
The inequality rewrites as $$ 2 < \left( 1+\frac{1}{\alpha-1} \right)^\alpha $$ the RHS expands (Newton generalized) as $$ 1 + \alpha \frac{1}{\alpha- 1} + \frac{\alpha (\alpha -1)}{2}\left (\frac{1}{\alpha- 1} \right)^2 + \frac{\alpha (\alpha -1)(\alpha-2)}{3!}\left (\frac{1}{\alpha- 1} \right)^3 + \dots$$ and thanks to Leibnitz criterion (alternating signs might begin at the fourth term) this is greater than the fist two terms which are greater than 2.
What I am not satisfied with is that it looks quite a long proof (actually, despite is trivial, the case $\alpha \in (1,2)$ should be treated separately) and indeed the inequality can be stronger substituting 2 with $e$. What I would like is a (let's say) creative proof rather than simply checking that the limit at $+\infty$ is $e$ and showing that the function is decreasing for $\alpha>1$.
$f(x)=2^x$ is a convex function on $[0,1]$, hence its graph lies above any tangent and below the secant line through $(0,1)$ and $(1,2)$, namely $y=x+1$. It follows that $$ \forall \alpha>1,\qquad 2^{1/\alpha} < 1+\frac{1}{\alpha} $$ that is stronger than the given inequality, since $$ \frac{\alpha}{\alpha-1} = \frac{1}{1-\frac{1}{\alpha}} = 1+\frac{1}{\alpha}+\frac{1}{\alpha^2}+\frac{1}{\alpha^3}+\ldots $$